1 /* $Id: out.c,v 1.87 2025/07/16 14:33:08 schwarze Exp $ */
2 /*
3 * Copyright (c) 2011, 2014, 2015, 2017, 2018, 2019, 2021, 2025
4 * Ingo Schwarze <schwarze@openbsd.org>
5 * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
6 *
7 * Permission to use, copy, modify, and distribute this software for any
8 * purpose with or without fee is hereby granted, provided that the above
9 * copyright notice and this permission notice appear in all copies.
10 *
11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 */
19 #include "config.h"
20
21 #include <sys/types.h>
22
23 #include <assert.h>
24 #include <ctype.h>
25 #include <stdint.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <time.h>
30
31 #include "mandoc_aux.h"
32 #include "mandoc.h"
33 #include "tbl.h"
34 #include "out.h"
35
36 struct tbl_colgroup {
37 struct tbl_colgroup *next;
38 size_t wanted;
39 int startcol;
40 int endcol;
41 };
42
43 static size_t tblcalc_data(struct rofftbl *, struct roffcol *,
44 const struct tbl_opts *, const struct tbl_dat *,
45 size_t);
46 static size_t tblcalc_literal(struct rofftbl *, struct roffcol *,
47 const struct tbl_dat *, size_t);
48 static size_t tblcalc_number(struct rofftbl *, struct roffcol *,
49 const struct tbl_opts *, const struct tbl_dat *);
50
51
52 /*
53 * Parse the *src string and store a scaling unit into *dst.
54 * If the string doesn't specify the unit, use the default.
55 * If no default is specified, fail.
56 * Return a pointer to the byte after the last byte used,
57 * or NULL on total failure.
58 */
59 const char *
a2roffsu(const char * src,struct roffsu * dst,enum roffscale def)60 a2roffsu(const char *src, struct roffsu *dst, enum roffscale def)
61 {
62 char *endptr;
63
64 dst->unit = def == SCALE_MAX ? SCALE_BU : def;
65 dst->scale = strtod(src, &endptr);
66 if (endptr == src)
67 return NULL;
68
69 switch (*endptr++) {
70 case 'c':
71 dst->unit = SCALE_CM;
72 break;
73 case 'i':
74 dst->unit = SCALE_IN;
75 break;
76 case 'f':
77 dst->unit = SCALE_FS;
78 break;
79 case 'M':
80 dst->unit = SCALE_MM;
81 break;
82 case 'm':
83 dst->unit = SCALE_EM;
84 break;
85 case 'n':
86 dst->unit = SCALE_EN;
87 break;
88 case 'P':
89 dst->unit = SCALE_PC;
90 break;
91 case 'p':
92 dst->unit = SCALE_PT;
93 break;
94 case 'u':
95 dst->unit = SCALE_BU;
96 break;
97 case 'v':
98 dst->unit = SCALE_VS;
99 break;
100 default:
101 endptr--;
102 if (SCALE_MAX == def)
103 return NULL;
104 dst->unit = def;
105 break;
106 }
107 return endptr;
108 }
109
110 /*
111 * Calculate the abstract widths and decimal positions of columns in a
112 * table. This routine allocates the columns structures then runs over
113 * all rows and cells in the table. The function pointers in "tbl" are
114 * used for the actual width calculations.
115 */
116 void
tblcalc(struct rofftbl * tbl,const struct tbl_span * sp_first,size_t offset,size_t rmargin)117 tblcalc(struct rofftbl *tbl, const struct tbl_span *sp_first,
118 size_t offset, size_t rmargin)
119 {
120 const struct tbl_opts *opts;
121 const struct tbl_span *sp;
122 const struct tbl_dat *dp;
123 struct roffcol *col;
124 struct tbl_colgroup *first_group, **gp, *g;
125
126 /* Widths in basic units. */
127 size_t *colwidth; /* Widths of all columns. */
128 size_t min1; /* Width of the narrowest column. */
129 size_t min2; /* Width of the second narrowest column. */
130 size_t wanted; /* For any of the narrowest columns. */
131 size_t xwidth; /* Total width of columns not to expand. */
132 size_t ewidth; /* Width of widest column to equalize. */
133 size_t width; /* Width of the data in basic units. */
134 size_t enw; /* Width of one EN unit. */
135
136 int icol; /* Column number, starting at zero. */
137 int maxcol; /* Number of last column. */
138 int necol; /* Number of columns to equalize. */
139 int nxcol; /* Number of columns to expand. */
140 int done; /* Boolean: this group is wide enough. */
141 int quirkcol;
142
143 /*
144 * Allocate the master column specifiers. These will hold the
145 * widths and decimal positions for all cells in the column. It
146 * must be freed and nullified by the caller.
147 */
148
149 assert(tbl->cols == NULL);
150 tbl->cols = mandoc_calloc((size_t)sp_first->opts->cols,
151 sizeof(struct roffcol));
152 opts = sp_first->opts;
153
154 maxcol = -1;
155 first_group = NULL;
156 enw = (*tbl->len)(1, tbl->arg);
157 for (sp = sp_first; sp != NULL; sp = sp->next) {
158 if (sp->pos != TBL_SPAN_DATA)
159 continue;
160
161 /*
162 * Account for the data cells in the layout, matching it
163 * to data cells in the data section.
164 */
165
166 for (dp = sp->first; dp != NULL; dp = dp->next) {
167 icol = dp->layout->col;
168 while (maxcol < icol + dp->hspans)
169 tbl->cols[++maxcol].spacing = SIZE_MAX;
170 col = tbl->cols + icol;
171 col->flags |= dp->layout->flags;
172 if (dp->layout->flags & TBL_CELL_WIGN)
173 continue;
174
175 /* Handle explicit width specifications. */
176 if (col->width < dp->layout->width)
177 col->width = dp->layout->width;
178 if (dp->layout->spacing != SIZE_MAX &&
179 (col->spacing == SIZE_MAX ||
180 col->spacing < dp->layout->spacing))
181 col->spacing = dp->layout->spacing;
182
183 /*
184 * Calculate an automatic width.
185 * Except for spanning cells, apply it.
186 */
187
188 width = tblcalc_data(tbl,
189 dp->hspans == 0 ? col : NULL,
190 opts, dp,
191 dp->block == 0 ? 0 :
192 dp->layout->width ? dp->layout->width :
193 rmargin ? (rmargin / enw + sp->opts->cols / 2) /
194 (sp->opts->cols + 1) * enw : 0);
195 if (dp->hspans == 0)
196 continue;
197
198 /*
199 * Build a singly linked list
200 * of all groups of columns joined by spans,
201 * recording the minimum width for each group.
202 */
203
204 gp = &first_group;
205 while (*gp != NULL && ((*gp)->startcol != icol ||
206 (*gp)->endcol != icol + dp->hspans))
207 gp = &(*gp)->next;
208 if (*gp == NULL) {
209 g = mandoc_malloc(sizeof(*g));
210 g->next = *gp;
211 g->wanted = width;
212 g->startcol = icol;
213 g->endcol = icol + dp->hspans;
214 *gp = g;
215 } else if ((*gp)->wanted < width)
216 (*gp)->wanted = width;
217 }
218 }
219
220 /*
221 * The minimum width of columns explicitly specified
222 * in the layout is 1n.
223 */
224
225 if (maxcol < sp_first->opts->cols - 1)
226 maxcol = sp_first->opts->cols - 1;
227 for (icol = 0; icol <= maxcol; icol++) {
228 col = tbl->cols + icol;
229 if (col->width < enw)
230 col->width = enw;
231
232 /*
233 * Column spacings are needed for span width
234 * calculations, so set the default values now.
235 */
236
237 if (col->spacing == SIZE_MAX || icol == maxcol)
238 col->spacing = 3;
239 }
240
241 /*
242 * Replace the minimum widths with the missing widths,
243 * and dismiss groups that are already wide enough.
244 */
245
246 gp = &first_group;
247 while ((g = *gp) != NULL) {
248 done = 0;
249 for (icol = g->startcol; icol <= g->endcol; icol++) {
250 width = tbl->cols[icol].width;
251 if (icol < g->endcol)
252 width += (*tbl->len)(tbl->cols[icol].spacing,
253 tbl->arg);
254 if (g->wanted <= width) {
255 done = 1;
256 break;
257 } else
258 g->wanted -= width;
259 }
260 if (done) {
261 *gp = g->next;
262 free(g);
263 } else
264 gp = &g->next;
265 }
266
267 colwidth = mandoc_reallocarray(NULL, maxcol + 1, sizeof(*colwidth));
268 while (first_group != NULL) {
269
270 /*
271 * Rebuild the array of the widths of all columns
272 * participating in spans that require expansion.
273 */
274
275 for (icol = 0; icol <= maxcol; icol++)
276 colwidth[icol] = SIZE_MAX;
277 for (g = first_group; g != NULL; g = g->next)
278 for (icol = g->startcol; icol <= g->endcol; icol++)
279 colwidth[icol] = tbl->cols[icol].width;
280
281 /*
282 * Find the smallest and second smallest column width
283 * among the columns which may need expamsion.
284 */
285
286 min1 = min2 = SIZE_MAX;
287 for (icol = 0; icol <= maxcol; icol++) {
288 width = colwidth[icol];
289 if (min1 > width) {
290 min2 = min1;
291 min1 = width;
292 } else if (min1 < width && min2 > width)
293 min2 = width;
294 }
295
296 /*
297 * Find the minimum wanted width
298 * for any one of the narrowest columns,
299 * and mark the columns wanting that width.
300 */
301
302 wanted = min2;
303 for (g = first_group; g != NULL; g = g->next) {
304 necol = 0;
305 for (icol = g->startcol; icol <= g->endcol; icol++)
306 if (colwidth[icol] == min1)
307 necol++;
308 if (necol == 0)
309 continue;
310 width = min1 + (g->wanted - 1) / necol + 1;
311 if (width > min2)
312 width = min2;
313 if (wanted > width)
314 wanted = width;
315 }
316
317 /* Record the effect of the widening. */
318
319 gp = &first_group;
320 while ((g = *gp) != NULL) {
321 done = 0;
322 for (icol = g->startcol; icol <= g->endcol; icol++) {
323 if (colwidth[icol] != min1)
324 continue;
325 if (g->wanted <= wanted - min1) {
326 tbl->cols[icol].width += g->wanted;
327 done = 1;
328 break;
329 }
330 tbl->cols[icol].width = wanted;
331 g->wanted -= wanted - min1;
332 }
333 if (done) {
334 *gp = g->next;
335 free(g);
336 } else
337 gp = &g->next;
338 }
339 }
340 free(colwidth);
341
342 /*
343 * Align numbers with text.
344 * Count columns to equalize and columns to maximize.
345 * Find maximum width of the columns to equalize.
346 * Find total width of the columns *not* to maximize.
347 */
348
349 necol = nxcol = 0;
350 ewidth = xwidth = 0;
351 for (icol = 0; icol <= maxcol; icol++) {
352 col = tbl->cols + icol;
353 if (col->width > col->nwidth)
354 col->decimal += (col->width - col->nwidth) / 2;
355 if (col->flags & TBL_CELL_EQUAL) {
356 necol++;
357 if (ewidth < col->width)
358 ewidth = col->width;
359 }
360 if (col->flags & TBL_CELL_WMAX)
361 nxcol++;
362 else
363 xwidth += col->width;
364 }
365
366 /*
367 * Equalize columns, if requested for any of them.
368 * Update total width of the columns not to maximize.
369 */
370
371 if (necol) {
372 for (icol = 0; icol <= maxcol; icol++) {
373 col = tbl->cols + icol;
374 if ( ! (col->flags & TBL_CELL_EQUAL))
375 continue;
376 if (col->width == ewidth)
377 continue;
378 if (nxcol && rmargin)
379 xwidth += ewidth - col->width;
380 col->width = ewidth;
381 }
382 }
383
384 /*
385 * If there are any columns to maximize, find the total
386 * available width, deducting 3n margins between columns.
387 * Distribute the available width evenly.
388 */
389
390 if (nxcol && rmargin) {
391 xwidth += (*tbl->len)(3 * maxcol +
392 (opts->opts & (TBL_OPT_BOX | TBL_OPT_DBOX) ?
393 2 : !!opts->lvert + !!opts->rvert), tbl->arg);
394 if (rmargin <= offset + xwidth)
395 return;
396 xwidth = rmargin - offset - xwidth;
397
398 /*
399 * Emulate a bug in GNU tbl width calculation that
400 * manifests itself for large numbers of x-columns.
401 * Emulating it for 5 x-columns gives identical
402 * behaviour for up to 6 x-columns.
403 */
404
405 if (nxcol == 5) {
406 quirkcol = xwidth / enw % nxcol + 2;
407 if (quirkcol != 3 && quirkcol != 4)
408 quirkcol = -1;
409 } else
410 quirkcol = -1;
411
412 necol = 0;
413 ewidth = 0;
414 for (icol = 0; icol <= maxcol; icol++) {
415 col = tbl->cols + icol;
416 if ( ! (col->flags & TBL_CELL_WMAX))
417 continue;
418 col->width = (double)xwidth * ++necol / nxcol
419 - ewidth + 0.4995;
420 if (necol == quirkcol)
421 col->width -= enw;
422 ewidth += col->width;
423 }
424 }
425 }
426
427 static size_t
tblcalc_data(struct rofftbl * tbl,struct roffcol * col,const struct tbl_opts * opts,const struct tbl_dat * dp,size_t mw)428 tblcalc_data(struct rofftbl *tbl, struct roffcol *col,
429 const struct tbl_opts *opts, const struct tbl_dat *dp, size_t mw)
430 {
431 size_t sz;
432
433 /* Branch down into data sub-types. */
434
435 switch (dp->layout->pos) {
436 case TBL_CELL_HORIZ:
437 case TBL_CELL_DHORIZ:
438 sz = (*tbl->len)(1, tbl->arg);
439 if (col != NULL && col->width < sz)
440 col->width = sz;
441 return sz;
442 case TBL_CELL_LONG:
443 case TBL_CELL_CENTRE:
444 case TBL_CELL_LEFT:
445 case TBL_CELL_RIGHT:
446 return tblcalc_literal(tbl, col, dp, mw);
447 case TBL_CELL_NUMBER:
448 return tblcalc_number(tbl, col, opts, dp);
449 case TBL_CELL_DOWN:
450 return 0;
451 default:
452 abort();
453 }
454 }
455
456 static size_t
tblcalc_literal(struct rofftbl * tbl,struct roffcol * col,const struct tbl_dat * dp,size_t mw)457 tblcalc_literal(struct rofftbl *tbl, struct roffcol *col,
458 const struct tbl_dat *dp, size_t mw)
459 {
460 const char *str; /* Beginning of the first line. */
461 const char *beg; /* Beginning of the current line. */
462 char *end; /* End of the current line. */
463
464 /* Widths in basic units. */
465 size_t lsz; /* Of the current line. */
466 size_t wsz; /* Of the current word. */
467 size_t msz; /* Of the longest line. */
468 size_t enw; /* Of one EN unit. */
469
470 if (dp->string == NULL || *dp->string == '\0')
471 return 0;
472 str = mw ? mandoc_strdup(dp->string) : dp->string;
473 msz = lsz = 0;
474 for (beg = str; beg != NULL && *beg != '\0'; beg = end) {
475 end = mw ? strchr(beg, ' ') : NULL;
476 if (end != NULL) {
477 *end++ = '\0';
478 while (*end == ' ')
479 end++;
480 }
481 wsz = (*tbl->slen)(beg, tbl->arg);
482 enw = (*tbl->len)(1, tbl->arg);
483 if (mw && lsz && lsz + enw + wsz <= mw)
484 lsz += enw + wsz;
485 else
486 lsz = wsz;
487 if (msz < lsz)
488 msz = lsz;
489 }
490 if (mw)
491 free((void *)str);
492 if (col != NULL && col->width < msz)
493 col->width = msz;
494 return msz;
495 }
496
497 static size_t
tblcalc_number(struct rofftbl * tbl,struct roffcol * col,const struct tbl_opts * opts,const struct tbl_dat * dp)498 tblcalc_number(struct rofftbl *tbl, struct roffcol *col,
499 const struct tbl_opts *opts, const struct tbl_dat *dp)
500 {
501 const char *cp, *lastdigit, *lastpoint;
502 size_t totsz; /* Total width of the number in basic units. */
503 size_t intsz; /* Width of the integer part in basic units. */
504 char buf[2];
505
506 if (dp->string == NULL || *dp->string == '\0')
507 return 0;
508
509 totsz = (*tbl->slen)(dp->string, tbl->arg);
510 if (col == NULL)
511 return totsz;
512
513 /*
514 * Find the last digit and
515 * the last decimal point that is adjacent to a digit.
516 * The alignment indicator "\&" overrides everything.
517 */
518
519 lastdigit = lastpoint = NULL;
520 for (cp = dp->string; cp[0] != '\0'; cp++) {
521 if (cp[0] == '\\' && cp[1] == '&') {
522 lastdigit = lastpoint = cp;
523 break;
524 } else if (cp[0] == opts->decimal &&
525 (isdigit((unsigned char)cp[1]) ||
526 (cp > dp->string && isdigit((unsigned char)cp[-1]))))
527 lastpoint = cp;
528 else if (isdigit((unsigned char)cp[0]))
529 lastdigit = cp;
530 }
531
532 /* Not a number, treat as a literal string. */
533
534 if (lastdigit == NULL) {
535 if (col != NULL && col->width < totsz)
536 col->width = totsz;
537 return totsz;
538 }
539
540 /* Measure the width of the integer part. */
541
542 if (lastpoint == NULL)
543 lastpoint = lastdigit + 1;
544 intsz = 0;
545 buf[1] = '\0';
546 for (cp = dp->string; cp < lastpoint; cp++) {
547 buf[0] = cp[0];
548 intsz += (*tbl->slen)(buf, tbl->arg);
549 }
550
551 /*
552 * If this number has more integer digits than all numbers
553 * seen on earlier lines, shift them all to the right.
554 * If it has fewer, shift this number to the right.
555 */
556
557 if (intsz > col->decimal) {
558 col->nwidth += intsz - col->decimal;
559 col->decimal = intsz;
560 } else
561 totsz += col->decimal - intsz;
562
563 /* Update the maximum total width seen so far. */
564
565 if (totsz > col->nwidth)
566 col->nwidth = totsz;
567 if (col->nwidth > col->width)
568 col->width = col->nwidth;
569 return totsz;
570 }
571