glen
/
teapot-spreadsheet
1
1
Fork 0
Code Issues 48 Pull Requests Releases Wiki Activity
teapot-spreadsheet/src/common/sheet.c

1913 lines
50 KiB
C
Raw Normal View History

Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* #includes */ /*{{{C}}}*//*{{{*/
#ifdef DMALLOC
#include "dmalloc.h"
#endif
#include <assert.h>
#include <errno.h>
#include <limits.h>
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
#include <stdbool.h>
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "csv.h"
#include "default.h"
#include "display.h"
#include "eval.h"
#include "main.h"
#include "misc.h"
#include "parser.h"
#include "scanner.h"
#include "sheet.h"
#include "utf8.h"
#include "xdr.h"
/*}}}*/
/* #defines */ /*{{{*/
#define HASH(x,s) \
{ \
const unsigned char *S=(const unsigned char*)s; \
\
x=0; \
while (*S) { x=(x<<5)+((x>>27)^*S); ++S; } \
x%=LABEL_CACHE; \
}
/*}}}*/
/* variables */ /*{{{*/
static int upd_clock; /* evaluate clocked expressions */
/* Used during evaluation of a cell to specify the currently updated cell */
Sheet *upd_sheet;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location upd_l;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
int max_eval;
/*}}}*/
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
/* copycelltosheet -- copy a cell into a sheet*/ /*{{{*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
void copycelltosheet(const Cell *fromcell, Sheet *sheet2,
const Location to, LabelHandling lh)
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet2 != (Sheet*)0);
Implement a fill-with menu item This new option fills the entire marked region with copies of the current cell. Implements #35.
2019-07-27 08:08:42 +00:00
Cell *tocell = safe_cell_at(sheet2, to);
if (fromcell == tocell) return;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
freecellofsheet(sheet2, to);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (fromcell != NULLCELL)
/* copy first cell to second */ /*{{{*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet2->changed = true;
Cell *alsotocell = initcellofsheet(sheet2, to, NULL);
Implement a fill-with menu item This new option fills the entire marked region with copies of the current cell. Implements #35.
2019-07-27 08:08:42 +00:00
assert(tocell == NULLCELL || tocell == alsotocell);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
copycell(alsotocell, fromcell, lh);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/*}}}*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
/* dump_cell -- write internal contents of cell to standard out */ /*{{{*/
static void dump_cell(Sheet *sheet, int x, int y, int z)
{
Cell* c;
char buf[2048];
assert(sheet != (Sheet*)0);
if (x < 0 || y < 0 || z < 0)
{
printf("TEADUMP: Requested cell &(%d,%d,%d) has negative coordinates.\n",
x, y, z);
return;
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (!LOC_WITHINC(sheet, x, y, z))
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
{
printf("TEADUMP: Requested cell &(%d,%d,%d) outside sheet dims %d,%d,%d.\n",
x, y, z, sheet->dimx, sheet->dimy, sheet->dimz);
return;
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
c = CELL_ATC(sheet, x, y, z);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
if (c == NULLCELL)
{
printf("TEADUMP: Cell at &(%d,%d,%d) is empty.\n", x, y, z);
return;
}
printf("TEADUMP of &(%d,%d,%d):\n", x, y, z);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (TokVariety tv = BASE_CONT; tv < CONTINGENT; ++tv)
{
printf(" %s: ", TokVariety_Name[tv]);
printtok(buf, sizeof(buf), 0, 1, 0, -1, 1, c->tok + tv);
printf("%s.\n", buf);
}
if (c->label == NULL) printf("\n No label.\n");
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
else printf("\n Label: %s.\n", c->label);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
printf(" Adjustment: %s.\n", Adjust_Name[c->adjust]);
printf(" Float format: %s.\n", FloatFormat_Name[c->fform]);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
printf(" Precision: %d\n Attributes: ", c->precision);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (ColorAspect ca = FOREGROUND; ca < NUM_COLOR_ASPECTS; ++ca)
printf(" %s color: %d\n", ColorAspect_Name[ca], c->aspect[ca]);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
if (c->updated) printf("updated ");
if (c->shadowed) printf("shadowed ");
if (c->locked) printf("locked ");
if (c->transparent) printf("transparent ");
if (c->ignored) printf("ignored ");
if (c->clock_t0) printf("clock_t0 ");
if (c->clock_t1) printf("clock_t1 ");
if (c->clock_t2) printf("clock_t2 ");
if (c->bold) printf("bold ");
if (c->underline) printf("underline ");
printf("\n\n");
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
/* dump_current_cell -- dump_cell of the current_cell */ /*{{{*/
void dump_current_cell(Sheet *sheet)
{
assert(sheet != (Sheet *)0);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
dump_cell(sheet, sheet->cur[X], sheet->cur[Y], sheet->cur[Z]);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* swapblock -- swap two non-overlapping blocks of cells */ /*{{{*/
static void swapblock(Sheet *sheet1, int x1, int y1, int z1, Sheet *sheet2, int x2, int y2, int z2, int xdist, int ydist, int zdist)
{
int xoff, yoff, zoff;
assert(sheet1!=(Sheet*)0);
assert(sheet2!=(Sheet*)0);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet1, x1+xdist-1, y1+ydist-1, z1+zdist-1, NULL);
resize(sheet2, x2+xdist-1, y2+ydist-1, z2+zdist-1, NULL);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
for (xoff=0; xoff<xdist; ++xoff)
for (yoff=0; yoff<ydist; ++yoff)
for (zoff=0; zoff<zdist; ++zoff)
{
Cell *t;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
t = CELL_ATC(sheet1,x1+xoff,y1+yoff,z1+zoff);
CELL_ATC(sheet1,x1+xoff,y1+yoff,z1+zoff) = CELL_ATC(sheet2,x2+xoff,y2+yoff,z2+zoff);
CELL_ATC(sheet2,x2+xoff,y2+yoff,z2+zoff) = t;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet1->changed = true;
sheet2->changed = true;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
/* signum -- -1 if < 0, 0 if 0, 1 if > 0 */
static int signum(int x) {
return (x > 0) - (x < 0);
}
/* cmpflt -- -1 if <, 0 if =, 1 if > */
static int cmpflt(FltT x, FltT y) {
return (x > y) - (x < y);
}
/* cmpflint -- -1 if <, 0 if =, 1 if > */
static int cmpflint(FltT x, IntT y) {
return (x > y) - (x < y);
}
/* cmpint -- -1 if <, 0 if =, 1 if > */
static int cmpint(IntT x, IntT y) {
return (x > y) - (x < y);
}
/* cmpintfl -- -1 if <, 0 if =, 1 if > */
static int cmpintfl(IntT x, FltT y) {
return (x > y) - (x < y);
}
#define NOT_COMPARABLE 2
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* cmpcell -- compare to cells with given order flags */ /*{{{*/
/* Notes */ /*{{{*/
/*
Compare the _values_ of two cells. The result is -1 if first is smaller
than second, 0 if they are equal and 1 if the first is bigger than the
second. A result of 2 means they are not comparable.
*/
/*}}}*/
static int cmpcell(Sheet *sheet1, int x1, int y1, int z1, Sheet *sheet2, int x2, int y2, int z2, int sortkey)
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
int ascending = (sortkey & ASCENDING) ? 1 : -1;
Token leftval = gettok(safe_cell_atc(sheet1, x1, y1, z1), CURR_VAL);
Token rightval = gettok(safe_cell_atc(sheet2, x2, y2, z2), CURR_VAL);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* empty cells are smaller than any non-empty cell */ /*{{{*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (leftval.type == EMPTY)
if (rightval.type == EMPTY) return 0;
else return -ascending;
if (rightval.type == EMPTY) return ascending;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
switch (leftval.type)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* STRING */ /*{{{*/
case STRING:
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (rightval.type == STRING)
return signum(strcmp(leftval.u.string, rightval.u.string)) * ascending;
return NOT_COMPARABLE;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* FLOAT */ /*{{{*/
case FLOAT:
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (rightval.type == FLOAT)
return cmpflt(leftval.u.flt, rightval.u.flt)*ascending;
if (rightval.type == INT)
return cmpflint(leftval.u.flt, rightval.u.integer)*ascending;
return NOT_COMPARABLE;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* INT */ /*{{{*/
case INT:
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (rightval.type == INT)
return cmpint(leftval.u.integer, rightval.u.integer)*ascending;
if (rightval.type == FLOAT)
return cmpintfl(leftval.u.integer, rightval.u.flt)*ascending;
return NOT_COMPARABLE;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
default: return NOT_COMPARABLE;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/* initialize_sheet -- fill in fields of a newly allocated Sheet */ /*{{{*/
void initialize_sheet(Sheet *sheet) {
assert(sheet != (Sheet*)0);
OLOCATION(sheet->cur);
sheet->offx = sheet->offy = 0;
sheet->dimx = sheet->dimy = sheet->dimz = 0;
sheet->sheet = (Cell**)0;
sheet->column = (int*)0;
sheet->orix = sheet->oriy = 0;
sheet->maxx = sheet->maxy = 0;
sheet->name = (char*)0;
OLOCATION(sheet->mark1);
OLOCATION(sheet->mark2);
sheet->marking = UNMARKED;
sheet->changed = 0;
sheet->moveonly = 0;
sheet->clk = 0;
(void)memset(sheet->labelcache, 0, sizeof(sheet->labelcache));
Prepare for per-cell configurable colors So far, this jst consists of initializing color in curses mode, and making the display start and end part of intializing and freeing a sheet (so that it can control allocating the palette, for example, where the data structure used depends on what kind of display it is. Next up will be to allocate and destroy the color palette, and set up the default colors for cells to use (0 for foreground, TEAPOT_WHITE for background.) The outline beyond that is to allow setting of the cell colors, then actually display those colors, and finally edit the palette.
2019-07-29 15:35:10 +00:00
sheet->max_colors = MAX_MAX_COLORS;
/* start display of this sheet */
if (!batch) {
display_init(sheet, always_redraw);
line_msg((const char*)0,"");
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
}
/*}}}*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* resize -- check if sheet needs to be resized in any dimension */ /*{{{*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
void resize(Sheet *sheet, int x, int y, int z, bool *qextended)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
assert(x>=0);
assert(y>=0);
assert(z>=0);
assert(sheet!=(Sheet*)0);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (!LOC_WITHINC(sheet,x,y,z))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Sheet dummy;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet->changed = true;
if (qextended) *qextended = true;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
dummy.dimx = (x >= sheet->dimx ? x+1 : sheet->dimx);
dummy.dimy = (y >= sheet->dimy ? y+1 : sheet->dimy);
dummy.dimz = (z >= sheet->dimz ? z+1 : sheet->dimz);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* allocate new sheet */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
dummy.sheet = malloc(dummy.dimx*dummy.dimy*dummy.dimz*sizeof(Cell*));
for (ALL_COORDS_IN_SHEETC(&dummy,x,y,z))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (LOC_WITHINC(sheet,x,y,z)) CELL_ATC(&dummy,x,y,z) = CELL_ATC(sheet,x,y,z);
else CELL_ATC(&dummy,x,y,z) = NULLCELL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
if (sheet->sheet!=(Cell**)0) free(sheet->sheet);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sheet->sheet = dummy.sheet;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* allocate new columns */ /*{{{*/
if (x>sheet->dimx || z>=sheet->dimz)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
dummy.column=malloc(dummy.dimx*dummy.dimz*sizeof(int));
for (x=0; x<dummy.dimx; ++x) for (z=0; z<dummy.dimz; ++z)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (x<sheet->dimx && z<sheet->dimz)
*(dummy.column+x*dummy.dimz+z)=*(sheet->column+x*sheet->dimz+z);
else *(dummy.column+x*dummy.dimz+z)=DEF_COLUMNWIDTH;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
if (sheet->column!=(int*)0) free(sheet->column);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sheet->column = dummy.column;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sheet->dimx = dummy.dimx;
sheet->dimy = dummy.dimy;
sheet->dimz = dummy.dimz;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
/* initcellofsheet -- initialise new cell, if it does not exist yet */ /*{{{*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
Cell *initcellofsheet(Sheet *sheet, const Location at, bool *qnew)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
Cell *nc;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(IN_OCTANT(at));
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet, at[X], at[Y], at[Z], qnew);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
nc = CELL_AT(sheet,at);
if (nc == NULLCELL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet->changed = true;
if (qnew) *qnew = true;
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
nc = malloc(sizeof(Cell));
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_AT(sheet,at) = nc;
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
initcellcontents(nc);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
return nc;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/* safe_cell_at -- returns pointer to the cell at the given location,
even in the face of various edge conditions.
*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Cell *safe_cell_atc(Sheet *sheet, int x, int y, int z)
{
if (sheet == (Sheet*)0) return NULLCELL;
if (sheet->sheet == (Cell **)0) return NULLCELL;
if (!LOC_WITHINC(sheet,x,y,z)) return NULLCELL;
return CELL_ATC(sheet,x,y,z);
}
/* safe_cell_at -- returns pointer to the cell at the given location,
even in the face of various edge conditions.
*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell *safe_cell_at(Sheet *sheet, const Location at)
{
if (sheet == (Sheet*)0) return NULLCELL;
if (sheet->sheet == (Cell **)0) return NULLCELL;
if (!LOC_WITHIN(sheet, at)) return NULLCELL;
return CELL_AT(sheet, at);
}
/* curcell -- return pointer to the current cell */ /*{{{*/
Cell *curcell(Sheet *sheet)
{
return safe_cell_at(sheet, sheet->cur);
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* cachelabels -- create new label cache */ /*{{{*/
void cachelabels(Sheet *sheet)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int i;
Location w;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (sheet==(Sheet*)0) return;
for (i=0; i<LABEL_CACHE; ++i) /* free bucket */ /*{{{*/
{
struct Label *run;
for (run=sheet->labelcache[i]; run!=(struct Label*)0;)
{
struct Label *runnext;
runnext=run->next;
free(run);
run=runnext;
}
sheet->labelcache[i]=(struct Label*)0;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_LOCS_IN_SHEET(sheet,w))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* cache all labels */ /*{{{*/
{
const char *l;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
l = getlabel(CELL_AT(sheet,w));
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (*l)
{
unsigned long hx;
struct Label **run;
HASH(hx,l);
for (run=&sheet->labelcache[(unsigned int)hx]; *run!=(struct Label*)0 && strcmp(l,(*run)->label); run=&((*run)->next));
if (*run==(struct Label*)0)
{
*run=malloc(sizeof(struct Label));
(*run)->next=(struct Label*)0;
(*run)->label=l;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
LOCATION_GETS((*run)->location,w);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/* else we have a duplicate label, which _can_ happen under */
/* unfortunate conditions. Don't tell anybody. */
}
}
/*}}}*/
}
/*}}}*/
/* freesheet -- free all cells of an entire spread sheet */ /*{{{*/
void freesheet(Sheet *sheet, int all)
{
/* variables */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
assert(sheet!=(Sheet*)0);
sheet->changed=0;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_LOCS_IN_SHEET(sheet,w))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
freecellofsheet(sheet,w);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
if (all)
{
int i;
for (i=0; i<LABEL_CACHE; ++i) /* free all buckets */ /*{{{*/
{
struct Label *run;
for (run=sheet->labelcache[i]; run!=(struct Label*)0;)
{
struct Label *runnext;
runnext=run->next;
free(run);
run=runnext;
}
}
/*}}}*/
if (sheet->sheet) free(sheet->sheet);
if (sheet->column) free(sheet->column);
if (sheet->name) free(sheet->name);
Prepare for per-cell configurable colors So far, this jst consists of initializing color in curses mode, and making the display start and end part of intializing and freeing a sheet (so that it can control allocating the palette, for example, where the data structure used depends on what kind of display it is. Next up will be to allocate and destroy the color palette, and set up the default colors for cells to use (0 for foreground, TEAPOT_WHITE for background.) The outline beyond that is to allow setting of the cell colors, then actually display those colors, and finally edit the palette.
2019-07-29 15:35:10 +00:00
if (!batch)
{
display_end(sheet);
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
else
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int x,z;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
for (x=0; x<sheet->dimx; ++x) for (z=0; z<sheet->dimz; ++z)
{
*(sheet->column+x*sheet->dimz+z)=DEF_COLUMNWIDTH;
}
cachelabels(sheet);
forceupdate(sheet);
}
}
/*}}}*/
/* forceupdate -- clear all clock and update flags */ /*{{{*/
void forceupdate(Sheet *sheet)
{
int i;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell *cell;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
assert(sheet!=(Sheet*)0);
Prevent some near-startup coredumps. Fixes #38.
2019-07-28 03:54:16 +00:00
if (sheet->sheet == (Cell **)0) return;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_CELLS_IN_SHEET(sheet,i,cell))
if (cell != NULLCELL)
cell->updated = cell->clock_t0 = cell->clock_t1 = cell->clock_t2 = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
update(sheet);
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
/* freecellofsheet -- free one cell */ /*{{{*/
void freecellofsheet(Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
if (sheet->sheet != (Cell**)0 && CELL_IS_GOOD(sheet,at))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
Cell *c = CELL_AT(sheet,at);
freecellcontents(c);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
free(c);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_AT(sheet,at) = NULLCELL;
sheet->changed = 1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* columnwidth -- get width of column */ /*{{{*/
int columnwidth(Sheet *sheet, int x, int z)
{
assert(sheet!=(Sheet*)0);
if (x<sheet->dimx && z<sheet->dimz) return (*(sheet->column+x*sheet->dimz+z));
else return DEF_COLUMNWIDTH;
}
/*}}}*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* setwidth -- set width of column */ /*{{{*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
bool setwidth(Sheet *sheet, int x, int z, int width)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
assert(sheet!=(Sheet*)0);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
bool isbigger = false;
resize(sheet, x, 1, z, &isbigger);
if (isbigger) sheet->changed = true;
int *storage = sheet->column + x*sheet->dimz + z;
if (*storage != width) {
*storage = width;
sheet->changed = true;
return true;
}
return isbigger;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* cellwidth -- get width of a cell */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int cellwidth(Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int width,x;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (SHADOWED(sheet,at)) return 0;
x = at[X];
width = columnwidth(sheet,x,at[Z]);
for (++x; SHADOWEDC(sheet,x,at[Y],at[Z]);
width += columnwidth(sheet,x,at[Z]), ++x);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
return width;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* putcont -- assign new contents */ /*{{{*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
void puttok(Sheet *sheet, const Location at, Token t, TokVariety v)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell *cell;
assert(sheet != (Sheet*)0);
sheet->changed = 1;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
cell = initcellofsheet(sheet, at, NULL);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
tfree(&(cell->tok[v]));
cell->tok[v] = t;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
redraw_cell(sheet, at);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
/* recompvalue -- get tcopy()ed value */ /*{{{*/
Token recompvalue(Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
Token result;
int orig_upd_clock;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell* cell;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
assert(sheet!=(Sheet*)0);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (!IN_OCTANT(at))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* return error */ /*{{{*/
{
result.type=EEK;
Implement a relative reference robust to source and reference moving The new function is X(SRC,REF), which briefly for a cell in the neighborhood of REF returns the corresponding cell in the neighborhood in SRC. For further details, see the updated documention. In developing and documenting this function, I refined some of the existing error messages (e.g. showing the coordinates when there is an attempt to obtain a cell with a negative coordinate) and improved the error propagation to increase the chance that the innermost error will percolate to the top level.
2019-07-30 19:07:54 +00:00
char *templ = _("Attempt to access cell at (%d,%d,%d) with negative index");
result.u.err = malloc(strlen(templ)+32);
sprintf(result.u.err, templ, at[X], at[Y], at[Z]);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
return result;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
result.type = EMPTY;
/* Can always short-circuit an out-of-bounds or empty cell */
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (!CELL_IS_GOOD(sheet,at)) return result;
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
/* only short-circuit on empty contents of a good cell if we are NOT
depending on this call to update the current value
*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell = CELL_AT(sheet, at);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (!upd_clock && gettok(cell, CONTINGENT).type == EMPTY) return result;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* update value of this cell if needed and return it */ /*{{{*/
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
orig_upd_clock = upd_clock;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->ignored)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
tfree(cell->tok + CURR_VAL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell->updated = 1;
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
else if (cell->updated == 0)
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
{
/* variables */ /*{{{*/
Sheet *old_sheet;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location old_l;
int old_max_eval;
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
Token oldvalue;
/*}}}*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
old_sheet = upd_sheet;
LOCATION_GETS(old_l, upd_l);
old_max_eval = max_eval;
upd_sheet = sheet;
LOCATION_GETS(upd_l, at);
max_eval = MAX_EVALNEST;
if (cell->clock_t1 == 0)
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell->updated = 1;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
oldvalue = gettok(cell, CURR_VAL);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
upd_clock = 0;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
cell->tok[CURR_VAL] = evaltoken(gettok(cell, CONTINGENT), FULL);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
tfree(&oldvalue);
}
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
else if (upd_clock)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell->updated = 1;
upd_clock = 0;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
oldvalue = gettok(cell, RES_VAL);
cell->tok[RES_VAL] = evaltoken(gettok(cell, CONTINGENT), FULL);
Prevent phantom values when clocking, resetting, and clocking again In the end it turned out that the cause of the phantom values was short-cutting in getvalue() when the contents of a cell were empty, preventing the update of the internal cache of the value of the cell. However, tracking this down (and getting the associated memory management correct) necessitated implementing a debugging mode in which I could dump the internal states of cells and print various other stuff to standard output. It also involved understanding the meaning of various pointers in the code, in the process of which I renamed some commonly used macros, particularly the former SHEET(s,x,y,z) which was not returning a Sheet at all but rather a pointer to a Cell. So this macro is now called CELL_AT. I also replaced several very repeatedly used patterns of checking the validity of locations and pointers with macros, now defined in sheet.h. Therefore, unfortunately the (relatively small in the end) bugfix for this major issue is entangled with numerous textual changes to the code made in tracking it down. Fixes #18. Closes #19.
2019-07-24 17:47:39 +00:00
tfree(&oldvalue);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
upd_sheet = old_sheet;
LOCATION_GETS(upd_l, old_l);
max_eval = old_max_eval;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (!orig_upd_clock) result = tcopy(cell->tok[CURR_VAL]);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
return result;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* update -- update all cells that need it */ /*{{{*/
void update(Sheet *sheet)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Cell *cell;
int i,kp,iterating;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
kp = 0;
iterating = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
do
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sheet->clk = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (iterating==1)
{
line_msg((const char*)0,_("Calculating running, press Escape to abort it"));
++iterating;
}
else if (iterating==0) ++iterating;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_CELLS_IN_SHEET(sheet,i,cell))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell && cell->clock_t2)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell->updated = 0;
cell->clock_t0 = 1;
cell->clock_t1 = 1;
cell->clock_t2 = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_LOCS_IN_SHEET(sheet,w))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
upd_clock = 1;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
recompvalue(sheet, w);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (ALL_CELLS_IN_SHEET(sheet,i,cell))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell && cell->clock_t1)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
tfree(&(cell->tok[CURR_VAL]));
cell->tok[CURR_VAL] = cell->tok[RES_VAL];;
cell->tok[RES_VAL].type = EMPTY;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cell->clock_t1 = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
upd_clock = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
} while (sheet->clk && !(kp=keypressed()));
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (iterating == 2) line_msg((const char*)0,kp ? _("Calculation aborted") : _("Calculation finished"));
sheet->clk = 0;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
redraw_sheet(sheet);
}
/*}}}*/
/* geterror -- get malloc()ed error string */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
char *geterror(Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Token v;
assert(sheet!=(Sheet*)0);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if ((v = recompvalue(sheet,at)).type != EEK)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
tfree(&v);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
return NULL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
else return (v.u.err);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* printvalue -- get ASCII representation of value */ /*{{{*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
void printvalue(char *s, size_t size, size_t chars, StringFormat sf,
FloatFormat ff, int precision, Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Token t = recompvalue(sheet, at);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
printtok(s, size, chars, sf, ff, precision, TRUNCATED_ERROR, &t);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
tfree(&t);
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
/* The cell field setters all have nearly identical code, so: */
#define DEFSETTER(funcname,argtype,field) bool\
funcname(Sheet *sheet, const Location at, argtype arg)\
{\
assert(sheet != (Sheet*)0);\
bool isnew = false;\
Cell* thecell = initcellofsheet(sheet, at, &isnew);\
if (isnew) sheet->changed = true;\
if (thecell->field != arg) {\
thecell->field = arg;\
sheet->changed = true;\
return true;\
}\
return isnew;\
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
DEFSETTER(setadjust, Adjust, adjust);
DEFSETTER(shadow, bool, shadowed);
DEFSETTER(bold, bool, bold);
DEFSETTER(underline, bool, underline);
DEFSETTER(lockcell, bool, locked);
DEFSETTER(maketrans, bool, transparent);
DEFSETTER(igncell, bool, ignored);
DEFSETTER(setfltformat, FloatFormat, fform);
DEFSETTER(setprecision, int, precision);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* clk -- clock cell */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void clk(Sheet *sheet, const Location at)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
assert(IN_OCTANT(at));
assert(LOC_WITHIN(sheet,at));
if (CELL_AT(sheet,at))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_AT(sheet,at)->clock_t2 = 1;
sheet->clk = 1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/* getmarkstate -- find out where in the marking cycle the sheet is */ /*{{{*/
MarkState getmarkstate(Sheet *sheet) {
if (sheet == (Sheet*)0) return MARK_CYCLE;
return sheet->marking;
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* setlabel -- set cell label */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void setlabel(Sheet *sheet, const Location at, const char *buf, int update)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell *cell;
assert(sheet != (Sheet*)0);
sheet->changed = 1;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
cell = initcellofsheet(sheet, at, NULL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->label != (char*)0) free(cell->label);
if (*buf == '\0') cell->label = (char*)0;
else cell->label = strcpy(malloc(strlen(buf)+1), buf);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (update)
{
cachelabels(sheet);
forceupdate(sheet);
}
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* findlabel -- return cell location for a given label */ /*{{{*/
Token findlabel(Sheet *sheet, const char *label)
{
/* variables */ /*{{{*/
Token result;
unsigned long hx;
struct Label *run;
/*}}}*/
assert(sheet!=(Sheet*)0);
/*
if (sheet==(Sheet*)0) run=(struct Label*)0;
else
*/
{
HASH(hx,label);
for (run=sheet->labelcache[(unsigned int)hx]; run!=(struct Label*)0 && strcmp(label,run->label); run=run->next);
}
if (run)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
result.type = LOCATION;
LOCATION_GETS(result.u.location, run->location);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
else
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
result.type = EEK;
Implement a relative reference robust to source and reference moving The new function is X(SRC,REF), which briefly for a cell in the neighborhood of REF returns the corresponding cell in the neighborhood in SRC. For further details, see the updated documention. In developing and documenting this function, I refined some of the existing error messages (e.g. showing the coordinates when there is an attempt to obtain a cell with a negative coordinate) and improved the error propagation to increase the chance that the innermost error will percolate to the top level.
2019-07-30 19:07:54 +00:00
char *pref = _("No such label: ");
result.u.err = malloc(strlen(pref) + strlen(label) + 1);
strcpy(result.u.err, pref);
strcat(result.u.err, label);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
return result;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
/* search and replace for labels in a token */
static void relabel_token(Token *tok, const char *oldl, const char *newl)
{
switch (tok->type)
{
case LIDENT:
if (strcmp(tok->u.lident, oldl) == 0)
{
tfree(tok);
tok->type = LIDENT;
tok->u.lident = strdup(newl);
}
break;
case FUNCALL:
for (int ai = 0; ai < tok->u.funcall.argc; ++ai)
relabel_token((tok->u.funcall.argv) + ai, oldl, newl);
break;
default:
break;
}
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* relabel -- search and replace for labels */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void relabel(Sheet *sheet, const Location at,
const char *oldlabel, const char *newlabel)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* asserts */ /*{{{*/
assert(oldlabel!=(const char*)0);
assert(newlabel!=(const char*)0);
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (!LOC_WITHIN(sheet, at)) return;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Cell *cell = CELL_AT(sheet, at);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell != NULLCELL)
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (TokVariety v = BASE_CONT; v <= MAX_PERSIST_TV; ++v)
relabel_token(cell->tok + v, oldlabel, newlabel);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* savetbl -- save as tbl tyble */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
const char *savetbl(Sheet *sheet, const char *name, int body,
const Location beg, const Location end,
unsigned int *count)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
FILE *fp=(FILE*)0; /* cause run time error */
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
char buf[1024];
char num[20];
char fullname[PATH_MAX];
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell *cw;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* asserts */ /*{{{*/
assert(sheet!=(Sheet*)0);
assert(name!=(const char*)0);
/*}}}*/
*count=0;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
w[X] = beg[X];
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++(w[Z]))
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++(w[Y]))
if (shadowed(CELL_AT(sheet,w))) return _("Shadowed cells in first column");
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (!body && (fp=fopen(name,"w"))==(FILE*)0) return strerror(errno);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++(w[Z]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
if (body)
/* open new file */ /*{{{*/
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sprintf(num, ".%d", w[Z]);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
fullname[sizeof(fullname)-strlen(num)-1]='\0';
(void)strncpy(fullname,name,sizeof(fullname)-strlen(num)-1);
fullname[sizeof(fullname)-1]='\0';
(void)strncat(fullname,num,sizeof(fullname)-strlen(num)-1);
fullname[sizeof(fullname)-1]='\0';
if ((fp=fopen(fullname,"w"))==(FILE*)0) return strerror(errno);
}
/*}}}*/
else if (fputs_close(".TS\n",fp)==EOF) return strerror(errno);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++(w[Y]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* print format */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[Y] > beg[Y] && fputs_close(".T&\n",fp)==EOF) return strerror(errno);
for (w[X]=beg[X]; w[X]<=end[X]; ++(w[X]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[X] > beg [X] && fputc_close(' ',fp)==EOF) return strerror(errno);
cw = CELL_AT(sheet,w);
if (shadowed(cw) && fputc_close('s',fp)==EOF) return strerror(errno);
if (isbold(cw) && fputc_close('b',fp)==EOF) return strerror(errno);
if (underlined(cw) && fputc_close('u',fp)==EOF) return strerror(errno);
switch (getadjust(cw))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
case LEFT: if (fputc_close('l',fp)==EOF) return strerror(errno); break;
case RIGHT: if (fputc_close('r',fp)==EOF) return strerror(errno); break;
case CENTER: if (fputc_close('c',fp)==EOF) return strerror(errno); break;
default: assert(0);
}
}
if (fputs_close(".\n",fp)==EOF) return strerror(errno);
/*}}}*/
/* print contents */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[X]=beg[X]; w[X]<=end[X]; ++(w[X]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
cw = CELL_AT(sheet,w);
if (!shadowed(cw))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[X] > beg[X] && fputc_close('\t',fp)==EOF) return strerror(errno);
if (cw != NULLCELL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
char *bufp;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
printvalue(buf, sizeof(buf), 0, DIRECT_STRING, getfltformat(cw),
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
getprecision(cw), sheet, w);
if (transparent(cw))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
if (fputs_close(buf,fp)==EOF) return strerror(errno);
}
else for (bufp=buf; *bufp; ++bufp) switch (*bufp)
{
case '\\':
{
if (fputc_close('\\',fp)==EOF || fputc_close('e',fp)==EOF) return strerror(errno);
break;
}
case '_':
{
if (fputc_close('\\',fp)==EOF || fputc_close('&',fp)==EOF || fputc_close('_',fp)==EOF) return strerror(errno);
break;
}
case '.':
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[X] == beg[X] && bufp==buf && (fputc_close('\\',fp)==EOF || fputc_close('&',fp)==EOF)) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (fputc_close('.',fp)==EOF) return strerror(errno);
break;
}
case '\'':
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[X] == beg[X] && bufp==buf && (fputc_close('\\',fp)==EOF || fputc_close('&',fp)==EOF)) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (fputc_close('\'',fp)==EOF) return strerror(errno);
break;
}
case '-':
{
if (*(bufp+1)=='-')
{
if (fputc_close('-',fp)==EOF) return strerror(errno);
else ++bufp;
}
else if (fputs_close("\\-",fp)==EOF) return strerror(errno);
break;
}
default: if (fputc_close(*bufp,fp)==EOF) return strerror(errno);
}
}
}
}
if (fputc_close('\n',fp)==EOF) return strerror(errno);
/*}}}*/
++*count;
}
if (!body)
{
if (fputs_close(".TE\n",fp)==EOF) return strerror(errno);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[Z] < end[Z] && fputs_close(".bp\n",fp)==EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
if (body && fclose(fp)==EOF) return strerror(errno);
}
if (!body && fclose(fp)==EOF) return strerror(errno);
return (const char*)0;
}
/*}}}*/
/* savetext -- save as text */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
const char *savetext(Sheet *sheet, const char *name,
const Location beg, const Location end,
unsigned int *count)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
FILE *fp;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Cell *cw;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* asserts */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
assert(name != (const char*)0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
*count=0;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
w[X] = beg[X];
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++(w[Z]))
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++(w[Y]))
if (shadowed(CELL_AT(sheet,w))) return _("Shadowed cells in first column");
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if ((fp=fopen(name,"w"))==(FILE*)0) return strerror(errno);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++(w[Z]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++(w[Y]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
size_t size,fill;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[X]=beg[X]; w[X]<=end[X]; ++(w[X]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
size = cellwidth(sheet,w);
cw = CELL_AT(sheet,w);
if (cw != NULLCELL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
char *buf;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
buf = malloc(size*UTF8SZ+1);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
printvalue(buf, size*UTF8SZ + 1, size, DIRECT_STRING,
getfltformat(cw), getprecision(cw), sheet, w);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
adjust(getadjust(cw), buf, size);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (fputs_close(buf,fp)==EOF)
{
free(buf);
return strerror(errno);
}
for (fill=strlen(buf); fill<size; ++fill) if (fputc_close(' ',fp)==EOF)
{
free(buf);
return strerror(errno);
}
free(buf);
}
else
{
for (fill=0; fill<size; ++fill) if (fputc_close(' ',fp)==EOF) return strerror(errno);
}
++*count;
}
if (fputc_close('\n',fp)==EOF) return strerror(errno);
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (w[Z] < end[Z] && fputs_close("\f",fp)==EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
if (fclose(fp)==EOF) return strerror(errno);
return (const char*)0;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* savecsv -- save as CSV */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
const char *savecsv(Sheet *sheet, const char *name, char sep,
const Location beg, const Location end,
unsigned int *count)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
FILE *fp;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* asserts */ /*{{{*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
assert(sheet != (Sheet*)0);
assert(name != (const char*)0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
*count = 0;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
w[X] = beg[X];
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (w[Z] = beg[Z]; w[Z] <= end[Z]; ++(w[Z]))
for (w[Y] = beg[Y]; w[Y] <= end[Y]; ++(w[Y]))
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (shadowed(CELL_AT(sheet,w))) return _("Shadowed cells in first column");
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if ((fp = fopen(name,"w")) == (FILE*)0) return strerror(errno);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++(w[Z]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++(w[Y]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[X]=beg[X]; w[X]<=end[X]; ++(w[X]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (w[X] > beg[X])
if (fputc_close(sep,fp) == EOF) return strerror(errno);
Cell *cw = CELL_AT(sheet,w);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cw != NULLCELL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
static const int bufchar = 511;
static const int bufsz = bufchar*UTF8SZ + 1;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
char *buf = malloc(bufsz);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
printvalue(buf, bufsz, bufchar, DIRECT_STRING, getfltformat(cw),
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
getprecision(cw), sheet, w);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
bool needquote = !transparent(cw)
&& (strlen(buf) != strcspn(buf, "\",\n\r"));
if (needquote && fputc_close('"',fp) == EOF)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
free(buf);
return strerror(errno);
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (const char *s = buf; *s; ++s)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (fputc_close(*s,fp) == EOF
|| (*s=='"' && needquote && fputc_close(*s,fp) == EOF))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
free(buf);
return strerror(errno);
}
}
free(buf);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (needquote && fputc_close('"', fp) == EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
++*count;
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (fputc_close('\n', fp) == EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (w[Z] < end[Z] && fputc_close('\f', fp) == EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (fclose(fp) == EOF) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
return (const char*)0;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
static const char *saveport_contentleader[] =
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
{ [BASE_CONT] = ":", [ITER_CONT] = "\\\n", [ATTR_REF] = "\\\n"
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
};
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* saveport -- save as portable text */ /*{{{*/
const char *saveport(Sheet *sheet, const char *name, unsigned int *count)
{
/* variables */ /*{{{*/
FILE *fp;
/*}}}*/
/* asserts */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
assert(sheet != (Sheet*)0);
assert(name != (const char*)0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
*count = 0;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if ((fp = fopen(name,"w")) == (FILE*)0) return strerror(errno);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
fprintf(fp,"# This is a work sheet generated with teapot %s.\n",VERSION);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (int z = sheet->dimz - 1; z >= 0; --z)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (int y = sheet->dimy - 1; y >= 0; --y)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
for (int x = sheet->dimx - 1; x >= 0; --x)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (y == 0 && columnwidth(sheet,x,z) != DEF_COLUMNWIDTH)
fprintf(fp,"W%d %d %d\n",x,z,columnwidth(sheet,x,z));
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Cell *cell = CELL_ATC(sheet,x,y,z);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell != NULLCELL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
fprintf(fp,"C%d %d %d ",x,y,z);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->adjust != AUTOADJUST)
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
fprintf(fp, "A%c ", Adjust_Char[cell->adjust]);
if (cell->label) fprintf(fp, "L%s ", cell->label);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->precision != -1)
fprintf(fp,"P%d ", cell->precision);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
if (cell->fform != DEF_FLOATFORM)
fprintf(fp,"F%c ", FloatFormat_Char[cell->fform]);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
bool needscolor = false;
for (ColorAspect a = FOREGROUND; a < NUM_COLOR_ASPECTS; ++a)
if (cell->aspect[a] != DefaultCN[a])
{ needscolor = true; break; }
if (needscolor)
{
fprintf(fp, "H%d ", NUM_COLOR_ASPECTS);
for (ColorAspect a = FOREGROUND; a < NUM_COLOR_ASPECTS; ++a)
fprintf(fp, "%d ", (int)cell->aspect[a]);
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->shadowed) fprintf(fp,"S ");
if (cell->bold) fprintf(fp,"B ");
if (cell->underline) fprintf(fp,"U ");
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (cell->locked) fprintf(fp, "K ");
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (cell->transparent) fprintf(fp,"T ");
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (cell->ignored) fprintf(fp, "I ");
TokVariety lastv = MAX_PERSIST_TV;
while (lastv > BASE_CONT && cell->tok[lastv].type == EMPTY) --lastv;
for (TokVariety v = BASE_CONT; v <= lastv; ++v)
{
if (v == BASE_CONT && cell->tok[v].type == EMPTY) continue;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
static const size_t bl = 4096;
char buf[bl];
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (fputs_close(saveport_contentleader[v], fp) == EOF)
return strerror(errno);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
size_t used = printtok(buf, bl, 0, QUOTE_STRING, FLT_HEXACT,
-1, RESTORE_ERROR, cell->tok + v);
if (used > bl - 3) return _("Internal buffer overrun in saveport");
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (fputs_close(buf, fp) == EOF) return strerror(errno);
}
if (fputc_close('\n', fp) == EOF) return strerror(errno);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
++(*count);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
}
}
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (fclose(fp) == EOF) return strerror(errno);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet->changed = false;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
return (const char*)0;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* loadport -- load from portable text */ /*{{{*/
const char *loadport(Sheet *sheet, const char *name)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Cell loaded;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
FILE *fp;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if ((fp=fopen(name,"r")) == (FILE*)0) return strerror(errno);
freesheet(sheet, 0);
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
static char errbuf[80];
const char *err = NULL;
int line = 1;
char buf[4096];
while (fgets(buf, sizeof(buf), fp) != NULL)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* remove nl */ /*{{{*/
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
int width = strlen(buf);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (width > 0 && buf[width-1] == '\n') buf[--width] = '\0';
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
switch (buf[0])
{
/* C -- parse cell */ /*{{{*/
case 'C':
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
TokVariety rv = BASE_CONT, nextrv = BASE_CONT;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
Location loc;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (width > 0 && buf[width-1]=='\\') { buf[--width]='\0'; ++nextrv; }
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
initcellcontents(&loaded);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* parse x y and z */ /*{{{*/
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
const char *os = buf + 1;
char *ns = buf + 1;
loc[X] = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for x position in line %d"),line);
err=errbuf;
goto eek;
}
while (*ns==' ') ++ns;
os=ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
loc[Y] = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for y position in line %d"),line);
err=errbuf;
goto eek;
}
while (*ns==' ') ++ns;
os=ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
loc[Z] = strtol(os, &ns, 0);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (os == ns)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
sprintf(errbuf,_("Parse error for z position in line %d"),line);
err=errbuf;
goto eek;
}
/*}}}*/
/* parse optional attributes */ /*{{{*/
do
{
while (*ns==' ') ++ns;
switch (*ns)
{
/* A -- adjustment */ /*{{{*/
case 'A':
{
++ns;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
const char *found = strchr(Adjust_Char, *ns);
if (found != NULL)
{
loaded.adjust = (Adjust)(found - Adjust_Char);
++ns;
break;
}
sprintf(errbuf, _("Parse error for adjustment in line %d"), line);
err = errbuf;
goto eek;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
/* F -- float format */ /*{{{*/
case 'F':
{
++ns;
const char* found = strchr(FloatFormat_Char, *ns);
if (found != NULL)
{
loaded.fform = (FloatFormat)(found - FloatFormat_Char);
++ns;
break;
}
sprintf(errbuf,
_("Parse error for float format in line %d"), line);
err = errbuf;
goto eek;
}
/*}}}*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* L -- label */ /*{{{*/
case 'L':
{
char buf[1024],*p;
p=buf;
++ns;
while (*ns && *ns!=' ') { *p=*ns; ++p; ++ns; }
*p='\0';
Implement a relative reference robust to source and reference moving The new function is X(SRC,REF), which briefly for a cell in the neighborhood of REF returns the corresponding cell in the neighborhood in SRC. For further details, see the updated documention. In developing and documenting this function, I refined some of the existing error messages (e.g. showing the coordinates when there is an attempt to obtain a cell with a negative coordinate) and improved the error propagation to increase the chance that the innermost error will percolate to the top level.
2019-07-30 19:07:54 +00:00
loaded.label=strdup(buf);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* P -- precision */ /*{{{*/
case 'P':
{
os=++ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
loaded.precision = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for precision in line %d"),line);
err=errbuf;
goto eek;
}
break;
}
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
/* H -- hue (colors) */ /*{{{*/
case 'H':
{
os = ++ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
size_t na = strtol(os, &ns, 0);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (os == ns || na > NUM_COLOR_ASPECTS)
{
sprintf(errbuf, _("Parse error for number of hues in line %d"),
line);
err = errbuf;
goto eek;
}
for (size_t a = 0; a < na; ++a)
{
while (*ns && (*ns == ' ')) { ++ns; }
os = ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
loaded.aspect[a] = strtol(os, &ns, 0);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (os == ns)
{
sprintf(errbuf, _("Parse error in hue %d on line %d"),
a, line);
err = errbuf;
goto eek;
}
}
break;
}
/*}}}*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* S -- shadowed */ /*{{{*/
case 'S':
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (loc[X]==0)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sprintf(errbuf,_("Trying to shadow cell (%d,%d,%d) in line %d"),
loc[X],loc[Y],loc[Z],line);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
err=errbuf;
goto eek;
}
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.shadowed=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* U -- underline */ /*{{{*/
case 'U':
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (loc[X]==0)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sprintf(errbuf,_("Trying to underline cell (%d,%d,%d) in line %d"),
loc[X],loc[Y],loc[Z],line);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
err=errbuf;
goto eek;
}
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.underline=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* B -- bold */ /*{{{*/
case 'B':
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (loc[X]==0)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
sprintf(errbuf,_("Trying to bold cell (%d,%d,%d) in line %d"),
loc[X], loc[Y], loc[Z], line);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
err=errbuf;
goto eek;
}
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.bold=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
/* E -- scientific, for backward compatibility */ /*{{{*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
case 'E':
{
++ns;
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
loaded.fform = FLT_SCIENTIFIC;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
/* K (formerly C) -- locked */ /*{{{*/
case 'C':
case 'K':
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.locked=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* T -- transparent */ /*{{{*/
case 'T':
{
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.transparent=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* I -- ignored */ /*{{{*/
case 'I':
{
++ns;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
loaded.ignored=1;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* : \0 -- do nothing */ /*{{{*/
case ':':
case '\0': break;
/*}}}*/
/* default -- error */ /*{{{*/
default: sprintf(errbuf,_("Invalid option %c in line %d"),*ns,line); err=errbuf; goto eek;
/*}}}*/
}
} while (*ns!=':' && *ns!='\0');
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (*ns) ++ns;
/* convert remaining string(s) into token sequence */ /*{{{*/
while (true)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Token **t = scan(&ns);
if (t == EMPTY_TVEC)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
sprintf(errbuf,_("Expression syntax error in line %d"),line);
err=errbuf;
goto eek;
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Token pt = eval_safe(t, LITERAL);
tvecfree(t);
if (pt.type == EEK)
{
sprintf(errbuf, _("Parse error in line %d: %s"),
line, pt.u.err);
err = errbuf;
tfree(&pt);
goto eek;
}
loaded.tok[rv] = pt;
if (nextrv == rv) break;
rv = nextrv;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (fgets(buf, sizeof(buf), fp) == (char*)0) break;
++line;
width = strlen(buf);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (width>0 && buf[width-1]=='\n') buf[width-1]='\0';
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/* More content? */
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
if (width > 0 && buf[width-1]=='\\') { buf[--width]='\0'; ++nextrv; }
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
ns = buf;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
copycelltosheet(&loaded, sheet, loc, PRESERVE_LABEL);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* W -- column width */ /*{{{*/
case 'W':
{
/* parse x and z */ /*{{{*/
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
const char *os = buf+1;
char *ns = buf+1;
int cx = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for x position in line %d"),line);
err=errbuf;
goto eek;
}
while (*ns==' ') ++ns;
os=ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
int cz = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for z position in line %d"),line);
err=errbuf;
goto eek;
}
/*}}}*/
/* parse width */ /*{{{*/
while (*ns==' ') ++ns;
os=ns;
Add type predicate The main new finction is is(), which tests the type of a value. There are also constants for each of the possible token types, and one new conversion: number(), which converts a value into the most appropriate number type. This change also reuses the functions in scanner.h for exracting ints, floats, and identifiers from strings, rather than either rewriting the functionality elsewhere, or using the general-purpose scan when only one token type was of interest.
2019-08-28 06:30:02 +00:00
width = strtol(os, &ns, 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
if (os==ns)
{
sprintf(errbuf,_("Parse error for width in line %d"),line);
err=errbuf;
goto eek;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
setwidth(sheet,cx,cz,width);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* # -- comment */ /*{{{*/
case '#': break;
/*}}}*/
/* default -- error */ /*{{{*/
default:
{
sprintf(errbuf,_("Unknown tag %c in line %d"),buf[0],line);
err=errbuf;
goto eek;
}
/*}}}*/
}
++line;
}
eek:
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
if (fclose(fp) == EOF) {
if (err == (const char*)0) err=strerror(errno);
} else {
sheet->changed = false;
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
cachelabels(sheet);
forceupdate(sheet);
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
redraw_sheet(sheet);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
return err;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* loadcsv -- load/merge CSVs */ /*{{{*/
const char *loadcsv(Sheet *sheet, const char *name)
{
/* variables */ /*{{{*/
FILE *fp;
Token **t;
const char *err;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location where;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
char ln[4096];
const char *str;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
FltT value;
IntT lvalue;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
int separator = 0;
/*}}}*/
if ((fp=fopen(name,"r"))==(FILE*)0) return strerror(errno);
err=(const char*)0;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
LOCATION_GETS(where, sheet->cur);
for (; fgets(ln,sizeof(ln),fp); ++(where[Y]))
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
const char *s;
const char *cend;
if (!separator) { /* FIXME: find a better way to autodetect */
int ccnt = 0, scnt = 0;
char *pos = ln;
while ((pos = strchr(pos, ','))) pos++, ccnt++;
pos = ln;
while ((pos = strchr(pos, ';'))) pos++, scnt++;
if (ccnt || scnt) separator = 1;
csv_setopt(scnt > ccnt);
}
s=cend=ln;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
where[X] = sheet->cur[X];
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
do
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
Token t;
t.type = EMPTY;
lvalue = csv_long(s, &cend);
if (s != cend) /* ok, it is a integer */ /*{{{*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
t.type = INT;
t.u.integer = lvalue;
puttok(sheet, where, t, BASE_CONT);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
else
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
value = csv_double(s, &cend);
if (s != cend) /* ok, it is a double */ /*{{{*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
t.type = FLOAT;
t.u.flt = value;
puttok(sheet, where, t, BASE_CONT);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
else
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
str = csv_string(s, &cend);
if (s != cend) /* ok, it is a string */ /*{{{*/
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
t.type = STRING;
t.u.string = strdup(str);
puttok(sheet, where, t, BASE_CONT);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/*}}}*/
else
{
csv_separator(s,&cend);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
while (s == cend && *s && *s!='\n')
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
err=_("unknown values ignored");
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
csv_separator(++s, &cend);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
/* else it is nothing, which does not need to be stored :) */
}
}
}
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
} while (s != cend ? s = cend, ++(where[X]), 1 : 0);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
fclose(fp);
return err;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* insertcube -- insert a block */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void insertcube(Sheet *sheet, const Location beg, const Location end,
Direction ins)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
int x,y,z;
/*}}}*/
switch (ins)
{
/* IN_X */ /*{{{*/
case IN_X:
{
int right;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
right=sheet->dimx+end[X]-beg[X];
for (z=beg[Z]; z<=end[Z]; ++z) for (y=beg[Y]; y<=end[Y]; ++y) for (x=right; x>end[X]; --x)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet, x, y, z, NULL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_ATC(sheet,x,y,z) = CELL_ATC(sheet,x-(end[X]-beg[X]+1),y,z);
CELL_ATC(sheet,x-(end[X]-beg[X]+1),y,z) = NULLCELL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
/* IN_Y */ /*{{{*/
case IN_Y:
{
int down;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
down=sheet->dimy+end[Y]-beg[Y];
for (z=beg[Z]; z<=end[Z]; ++z) for (x=beg[X]; x<=end[X]; ++x) for (y=down; y>end[Y]; --y)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet, x, y, z, NULL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_ATC(sheet,x,y,z) = CELL_ATC(sheet,x,y-(end[Y]-beg[Y]+1),z);
CELL_ATC(sheet,x,y-(end[Y]-beg[Y]+1),z) = NULLCELL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
/* IN_Z */ /*{{{*/
case IN_Z:
{
int bottom;
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
bottom=sheet->dimz+end[Z]-beg[Z];
for (y=beg[Y]; y<=end[Y]; ++y) for (x=beg[X]; x<=end[X]; ++x) for (z=bottom; z>end[Z]; --z)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet, x, y, z, NULL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_ATC(sheet,x,y,z) = CELL_ATC(sheet,x,y,z-(end[Z]-beg[Z]+1));
CELL_ATC(sheet,x,y,z-(end[Z]-beg[Z]+1)) = NULLCELL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
/* default */ /*{{{*/
default: assert(0);
/*}}}*/
}
sheet->changed=1;
cachelabels(sheet);
forceupdate(sheet);
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* deletecube -- delete a block */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void deletecube(Sheet *sheet, const Location beg, const Location end,
Direction del)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location w;
Location fm;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
/* free cells in marked block */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[X]=beg[X]; w[X]<=end[X]; ++w[X])
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++w[Y])
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++w[Z])
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
freecellofsheet(sheet, w);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
switch (del)
{
/* IN_X */ /*{{{*/
case IN_X:
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++w[Z])
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++w[Y])
for (w[X]=beg[X]; w[X]<=sheet->dimx-(end[X]-beg[X]+1); ++w[X])
{
LOCATION_GETS(fm, w);
fm[X] += end[X]-beg[X]+1;
if (LOC_WITHIN(sheet,fm))
{
CELL_AT(sheet,w) = CELL_AT(sheet,fm);
CELL_AT(sheet,fm) = NULLCELL;
}
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* IN_Y */ /*{{{*/
case IN_Y:
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Z]=beg[Z]; w[Z]<=end[Z]; ++w[Z])
for (w[X]=beg[X]; w[X]<=end[X]; ++w[X])
for (w[Y]=beg[Y]; w[Y]<=sheet->dimy-(end[Y]-beg[Y]+1); ++w[Y])
{
LOCATION_GETS(fm, w);
fm[Y] += end[Y]-beg[Y]+1;
if (LOC_WITHIN(sheet,fm))
{
CELL_AT(sheet,w) = CELL_AT(sheet,fm);
CELL_AT(sheet,fm) = NULLCELL;
}
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* IN_Z */ /*{{{*/
case IN_Z:
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
for (w[Y]=beg[Y]; w[Y]<=end[Y]; ++w[Y])
for (w[X]=beg[X]; w[X]<=end[X]; ++w[X])
for (w[Z]=beg[Z]; w[Z]<=sheet->dimz-(end[Z]-beg[Z]+1); ++w[Z])
{
LOCATION_GETS(fm, w);
fm[Z] += end[Z]-beg[Z]+1;
if (LOC_WITHIN(sheet,fm))
{
CELL_AT(sheet,w) = CELL_AT(sheet,fm);
CELL_AT(sheet,fm) = NULLCELL;
}
}
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
break;
}
/*}}}*/
/* default */ /*{{{*/
default: assert(0);
/*}}}*/
}
sheet->changed=1;
cachelabels(sheet);
forceupdate(sheet);
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* moveblock -- move a block */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void moveblock(Sheet *sheet, const Location beg, const Location end,
const Location dest, int copy)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Location wid, dir, fm, to, get, go;
Dimensions dim;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
if (SAME_LOC(beg, dest)) return;
LOCATION_GETS(wid, end);
LOCATION_SUB(wid, beg);
for (dim = X; dim < HYPER; ++dim) {
if (dest[dim] > beg[dim])
{ dir[dim] = -1; fm[dim] = wid[dim]; to[dim] = beg[dim]-1; }
else
{ dir[dim] = 1; fm[dim] = 0; to[dim] = beg[dim]+wid[dim]+1; }
}
for (get[Z] = beg[Z]+fm[Z], go[Z] = dest[Z]+fm[Z];
get[Z] != to[Z]; get[Z] += dir[Z], go[Z] += dir[Z])
for (get[Y] = beg[Y]+fm[Y], go[Y] = dest[Y]+fm[Y];
get[Y] != to[Y]; get[Y] += dir[Y], go[Y] += dir[Y])
for (get[X] = beg[X]+fm[X], go[X] = dest[X]+fm[X];
get[X] != to[X]; get[X] += dir[X], go[X] += dir[X])
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
Factor Cell into its own file THis will make adding other Cell operations easier.
2019-07-27 07:00:03 +00:00
if (!LOC_WITHIN(sheet, get)) freecellofsheet(sheet, go);
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
else if (copy)
copycelltosheet(CELL_AT(sheet, get), sheet, go, ALTER_LABEL);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
else
{
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
resize(sheet, go[X], go[Y], go[Z], NULL);
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
CELL_AT(sheet, go) = CELL_AT(sheet, get);
CELL_AT(sheet, get) = NULLCELL;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
}
Make tpa the default file format for Teapot There turned out to be one blocking technical issue for full adpotion of tpa format, and that was the fact that the printed representation of floating values inside the tpa file might not reproduce the same double when read. This change therefore introduces the "hexact" floating point format, based on the the %La format string, which produces a hex representation with exact round trips. While working on this, it was convenient to add a new representation "compact" which is basically the shorter of decimal and scientific representations, without trailing zeros. This is now the default float format, but of course one can select decimal or scientific formats to restore prior appearance. However, full-precision compact format is now the (only) format for editing cell contents, as it is accurate and efficient. Of course you can enter floating point values in any format you like when typing in a formula. The addition of several new floating point options overloaded the menus in terminal teapot, so this change also revamps those menus slightly, including eliminating the unused MenuChoice struct, and just specifying menus with an array of strings. Closes #63.
2019-08-24 16:58:46 +00:00
sheet->changed = TRUE;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
cachelabels(sheet);
forceupdate(sheet);
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* sortblock -- sort a block */ /*{{{*/
/* Notes */ /*{{{*/
/*
The idea is to sort a block of cells in one direction by swapping the
planes which are canonical to the sort key vectors. An example is to
sort a two dimensional block line-wise with one column as sort key.
You can have multiple sort keys which all have the same direction and
you can sort a cube plane-wise.
*/
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
const char *sortblock(Sheet *sheet, const Location beg, const Location end,
Direction dir, Sortkey *sk, size_t sklen)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
/* variables */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int x1, y1, z1, x2, y2, z2;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
int x,y,z;
int incx=0,incy=0,incz=0;
int distx,disty,distz;
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
int i,r=-3,work;
bool norel = false;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
/* unpack corners */
x1 = beg[X]; y1 = beg[Y]; z1 = beg[Z];
x2 = end[X]; y2 = end[Y]; z2 = end[Z];
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* asserts */ /*{{{*/
assert(sklen>0);
assert(x1>=0);
assert(x2>=0);
assert(y1>=0);
assert(y2>=0);
assert(z1>=0);
assert(z2>=0);
/*}}}*/
norel=0;
posorder(&x1,&x2);
posorder(&y1,&y2);
posorder(&z1,&z2);
distx=(x2-x1+1);
disty=(y2-y1+1);
distz=(z2-z1+1);
switch (dir)
{
case IN_X: incx=1; --x2; incy=0; incz=0; distx=1; break;
case IN_Y: incx=0; incy=1; --y2; incz=0; disty=1; break;
case IN_Z: incx=0; incy=0; incz=1; --z2; distz=1; break;
default: assert(0);
}
assert(incx || incy || incz);
do
{
work=0;
for (x=x1,y=y1,z=z1; x<=x2&&y<=y2&&z<=z2; x+=incx,y+=incy,z+=incz)
{
for (i=0; i<sklen; ++i)
{
Streamline internals of teapot The primary change is to add a “funcall” token, so that an entire expression can be encapsulated as a single token. This change is used to allow a cell to include simply a selection of appropriate semantic tokens. I.e., the content and iterative content are now each a single token like the value and the result value. Token vectors are used only as intermediate results in scanning and parsing. Not this means the cells are now in effect storing parse trees, so computation should be slightly faster and future extensions (like #56) should be facilitated. This commit also takes the opportunity while internals are being altered to add another token to a cell for future use for computed attributes, cf #22, and to change the internal numerical values from double and ints to long doubles and long longs. However, the change attempts to encapsulate that choice so it would be easy to change back or change to another representation. Note that these changes break savexdr(), as the internal binary format of a cell is now different. Rather than reimplement it, it is deprecated as the world does not need another binary spreadsheet format. Hence, the ascii format for teapot spreadsheets becomes the primary file format. Loading of old xdr files is still supported for backward compatibility. Closes #59. Also along the way, various other slight fixes and enhancements crept in, a partial but probably not exhaustive list of which follows: Fixes #31. Further revisions and improvements to documentation. Make the approximate comparison of floating point values scale more accurately with the size of the doubles being compared. Further extensions of absolute and relative cell addressing. Addition of (circle constant) tau function/constant. Modified string conversion to simply use internal printing routines, and to take "scientific" and "decimal" keywords. Allowed n() function to take a list of values, or just a single location defaulting to the current location. Added floor, ceil, trunc, and round functions, and allowed them to be keywords controlling the int() integer conversion function as well. Allowed substr() to drop its last argument to go to the end of the string. Provided an enum of built-in functions to preserve legacy function identifiers, allowing the large table inside func.c to be reorganized in a clearer fashion. Added additional annotation of properties of the built-in functions, including precedence. All operators are now also accessible as built-in functions. Made precedence of unary - lower than ^ to match python. Avoided inadvertently using FLTK @symbol abbreviations for formulas with "@" in them.
2019-08-23 19:12:06 +00:00
r = cmpcell(sheet,x+sk[i].x,y+sk[i].y,z+sk[i].z,sheet,x+incx+sk[i].x,y+incy+sk[i].y,z+incz+sk[i].z,sk[i].sortkey);
if (r == NOT_COMPARABLE) norel = true;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
else if (r==-1 || r==1) break;
else assert(r==0);
}
if (r==1)
{
swapblock(sheet,dir==IN_X ? x : x1,dir==IN_Y ? y : y1,dir==IN_Z ? z : z1,sheet,dir==IN_X ? x+incx : x1,dir==IN_Y ? y+incy : y1,dir==IN_Z ? z+incz : z1,distx,disty,distz);
work=1;
}
}
x2-=incx;
y2-=incy;
z2-=incz;
} while (work);
cachelabels(sheet);
forceupdate(sheet);
sheet->changed=1;
if (norel) return _("uncomparable elements");
else return (const char*)0;
}
/*}}}*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
/* mirrorblock -- mirror a block */ /*{{{*/
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
void mirrorblock(Sheet *sheet, const Location beg, const Location end,
Direction dir)
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
{
switch (dir)
{
case IN_X: /* left-right */ /*{{{*/
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int x,middle=(end[X]-beg[X]+1)/2;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
for (x=0; x<middle; ++x)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
swapblock(sheet,beg[X]+x,beg[Y],beg[Z],sheet,end[X]-x,beg[Y],beg[Z], 1,end[Y]-beg[Y]+1,end[Z]-beg[Z]+1);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
case IN_Y: /* upside-down */ /*{{{*/
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int y,middle=(end[Y]-beg[Y]+1)/2;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
for (y=0; y<middle; ++y)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
swapblock(sheet,beg[X],beg[Y]+y,beg[Z],sheet,beg[X],end[Y]-y,beg[Z], end[X]-beg[X]+1,1,end[Z]-beg[Z]+1);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
case IN_Z: /* front-back */ /*{{{*/
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
int z,middle=(end[Z]-beg[Z]+1)/2;
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
for (z=0; z<middle; ++z)
{
A first stab at refactoring In general, the goal of these changes is to be more type-specific wherever possible, reduce duplicated code, and avoid magic numbers. The biggest individual change is to use Location triples of integers wherever possible rather than three separate int variables. Another very helpful change for understanding what is going on in the code is introducing the MarkState enum for tracking the process of marking and unmarking blocks of cell. The MarkState change was motivated by the issues with Ctrl-C, and this commit, in fact, resolves #28. Because of the lack of a test harness, it is possible that a change of this scope has created some bugs as well, but teapot was running OK for me at the time of the commit. However, I don't know how good my coverage of the changed code was -- I certainly did not save or load a Lotus 1-2-3 file!
2019-07-27 04:14:26 +00:00
swapblock(sheet,beg[X],beg[Y],beg[Z]+z,sheet,beg[X],beg[Y],end[Z]-z, end[X]-beg[X]+1,end[Y]-beg[Y]+1,1);
Import exactly the files of the teapot-2.3.0 distribution
2019-07-22 20:32:33 +00:00
}
break;
}
/*}}}*/
default: assert(0);
}
sheet->changed=1;
cachelabels(sheet);
forceupdate(sheet);
}
/*}}}*/