1 /* $Id: out.c,v 1.86 2025/01/05 18:14:39 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2011, 2014, 2015, 2017, 2018, 2019, 2021 5 * Ingo Schwarze <schwarze@openbsd.org> 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 * 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 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 size_t *colwidth; 126 size_t ewidth, min1, min2, wanted, width, xwidth; 127 int done, icol, maxcol, necol, nxcol, quirkcol; 128 129 /* 130 * Allocate the master column specifiers. These will hold the 131 * widths and decimal positions for all cells in the column. It 132 * must be freed and nullified by the caller. 133 */ 134 135 assert(tbl->cols == NULL); 136 tbl->cols = mandoc_calloc((size_t)sp_first->opts->cols, 137 sizeof(struct roffcol)); 138 opts = sp_first->opts; 139 140 maxcol = -1; 141 first_group = NULL; 142 for (sp = sp_first; sp != NULL; sp = sp->next) { 143 if (sp->pos != TBL_SPAN_DATA) 144 continue; 145 146 /* 147 * Account for the data cells in the layout, matching it 148 * to data cells in the data section. 149 */ 150 151 for (dp = sp->first; dp != NULL; dp = dp->next) { 152 icol = dp->layout->col; 153 while (maxcol < icol + dp->hspans) 154 tbl->cols[++maxcol].spacing = SIZE_MAX; 155 col = tbl->cols + icol; 156 col->flags |= dp->layout->flags; 157 if (dp->layout->flags & TBL_CELL_WIGN) 158 continue; 159 160 /* Handle explicit width specifications. */ 161 if (col->width < dp->layout->width) 162 col->width = dp->layout->width; 163 if (dp->layout->spacing != SIZE_MAX && 164 (col->spacing == SIZE_MAX || 165 col->spacing < dp->layout->spacing)) 166 col->spacing = dp->layout->spacing; 167 168 /* 169 * Calculate an automatic width. 170 * Except for spanning cells, apply it. 171 */ 172 173 width = tblcalc_data(tbl, 174 dp->hspans == 0 ? col : NULL, 175 opts, dp, 176 dp->block == 0 ? 0 : 177 dp->layout->width ? dp->layout->width : 178 rmargin ? (rmargin + sp->opts->cols / 2) 179 / (sp->opts->cols + 1) : 0); 180 if (dp->hspans == 0) 181 continue; 182 183 /* 184 * Build a singly linked list 185 * of all groups of columns joined by spans, 186 * recording the minimum width for each group. 187 */ 188 189 gp = &first_group; 190 while (*gp != NULL && ((*gp)->startcol != icol || 191 (*gp)->endcol != icol + dp->hspans)) 192 gp = &(*gp)->next; 193 if (*gp == NULL) { 194 g = mandoc_malloc(sizeof(*g)); 195 g->next = *gp; 196 g->wanted = width; 197 g->startcol = icol; 198 g->endcol = icol + dp->hspans; 199 *gp = g; 200 } else if ((*gp)->wanted < width) 201 (*gp)->wanted = width; 202 } 203 } 204 205 /* 206 * The minimum width of columns explicitly specified 207 * in the layout is 1n. 208 */ 209 210 if (maxcol < sp_first->opts->cols - 1) 211 maxcol = sp_first->opts->cols - 1; 212 for (icol = 0; icol <= maxcol; icol++) { 213 col = tbl->cols + icol; 214 if (col->width < 1) 215 col->width = 1; 216 217 /* 218 * Column spacings are needed for span width 219 * calculations, so set the default values now. 220 */ 221 222 if (col->spacing == SIZE_MAX || icol == maxcol) 223 col->spacing = 3; 224 } 225 226 /* 227 * Replace the minimum widths with the missing widths, 228 * and dismiss groups that are already wide enough. 229 */ 230 231 gp = &first_group; 232 while ((g = *gp) != NULL) { 233 done = 0; 234 for (icol = g->startcol; icol <= g->endcol; icol++) { 235 width = tbl->cols[icol].width; 236 if (icol < g->endcol) 237 width += tbl->cols[icol].spacing; 238 if (g->wanted <= width) { 239 done = 1; 240 break; 241 } else 242 g->wanted -= width; 243 } 244 if (done) { 245 *gp = g->next; 246 free(g); 247 } else 248 gp = &g->next; 249 } 250 251 colwidth = mandoc_reallocarray(NULL, maxcol + 1, sizeof(*colwidth)); 252 while (first_group != NULL) { 253 254 /* 255 * Rebuild the array of the widths of all columns 256 * participating in spans that require expansion. 257 */ 258 259 for (icol = 0; icol <= maxcol; icol++) 260 colwidth[icol] = SIZE_MAX; 261 for (g = first_group; g != NULL; g = g->next) 262 for (icol = g->startcol; icol <= g->endcol; icol++) 263 colwidth[icol] = tbl->cols[icol].width; 264 265 /* 266 * Find the smallest and second smallest column width 267 * among the columns which may need expamsion. 268 */ 269 270 min1 = min2 = SIZE_MAX; 271 for (icol = 0; icol <= maxcol; icol++) { 272 width = colwidth[icol]; 273 if (min1 > width) { 274 min2 = min1; 275 min1 = width; 276 } else if (min1 < width && min2 > width) 277 min2 = width; 278 } 279 280 /* 281 * Find the minimum wanted width 282 * for any one of the narrowest columns, 283 * and mark the columns wanting that width. 284 */ 285 286 wanted = min2; 287 for (g = first_group; g != NULL; g = g->next) { 288 necol = 0; 289 for (icol = g->startcol; icol <= g->endcol; icol++) 290 if (colwidth[icol] == min1) 291 necol++; 292 if (necol == 0) 293 continue; 294 width = min1 + (g->wanted - 1) / necol + 1; 295 if (width > min2) 296 width = min2; 297 if (wanted > width) 298 wanted = width; 299 } 300 301 /* Record the effect of the widening. */ 302 303 gp = &first_group; 304 while ((g = *gp) != NULL) { 305 done = 0; 306 for (icol = g->startcol; icol <= g->endcol; icol++) { 307 if (colwidth[icol] != min1) 308 continue; 309 if (g->wanted <= wanted - min1) { 310 tbl->cols[icol].width += g->wanted; 311 done = 1; 312 break; 313 } 314 tbl->cols[icol].width = wanted; 315 g->wanted -= wanted - min1; 316 } 317 if (done) { 318 *gp = g->next; 319 free(g); 320 } else 321 gp = &g->next; 322 } 323 } 324 free(colwidth); 325 326 /* 327 * Align numbers with text. 328 * Count columns to equalize and columns to maximize. 329 * Find maximum width of the columns to equalize. 330 * Find total width of the columns *not* to maximize. 331 */ 332 333 necol = nxcol = 0; 334 ewidth = xwidth = 0; 335 for (icol = 0; icol <= maxcol; icol++) { 336 col = tbl->cols + icol; 337 if (col->width > col->nwidth) 338 col->decimal += (col->width - col->nwidth) / 2; 339 if (col->flags & TBL_CELL_EQUAL) { 340 necol++; 341 if (ewidth < col->width) 342 ewidth = col->width; 343 } 344 if (col->flags & TBL_CELL_WMAX) 345 nxcol++; 346 else 347 xwidth += col->width; 348 } 349 350 /* 351 * Equalize columns, if requested for any of them. 352 * Update total width of the columns not to maximize. 353 */ 354 355 if (necol) { 356 for (icol = 0; icol <= maxcol; icol++) { 357 col = tbl->cols + icol; 358 if ( ! (col->flags & TBL_CELL_EQUAL)) 359 continue; 360 if (col->width == ewidth) 361 continue; 362 if (nxcol && rmargin) 363 xwidth += ewidth - col->width; 364 col->width = ewidth; 365 } 366 } 367 368 /* 369 * If there are any columns to maximize, find the total 370 * available width, deducting 3n margins between columns. 371 * Distribute the available width evenly. 372 */ 373 374 if (nxcol && rmargin) { 375 xwidth += 3*maxcol + 376 (opts->opts & (TBL_OPT_BOX | TBL_OPT_DBOX) ? 377 2 : !!opts->lvert + !!opts->rvert); 378 if (rmargin <= offset + xwidth) 379 return; 380 xwidth = rmargin - offset - xwidth; 381 382 /* 383 * Emulate a bug in GNU tbl width calculation that 384 * manifests itself for large numbers of x-columns. 385 * Emulating it for 5 x-columns gives identical 386 * behaviour for up to 6 x-columns. 387 */ 388 389 if (nxcol == 5) { 390 quirkcol = xwidth % nxcol + 2; 391 if (quirkcol != 3 && quirkcol != 4) 392 quirkcol = -1; 393 } else 394 quirkcol = -1; 395 396 necol = 0; 397 ewidth = 0; 398 for (icol = 0; icol <= maxcol; icol++) { 399 col = tbl->cols + icol; 400 if ( ! (col->flags & TBL_CELL_WMAX)) 401 continue; 402 col->width = (double)xwidth * ++necol / nxcol 403 - ewidth + 0.4995; 404 if (necol == quirkcol) 405 col->width--; 406 ewidth += col->width; 407 } 408 } 409 } 410 411 static size_t 412 tblcalc_data(struct rofftbl *tbl, struct roffcol *col, 413 const struct tbl_opts *opts, const struct tbl_dat *dp, size_t mw) 414 { 415 size_t sz; 416 417 /* Branch down into data sub-types. */ 418 419 switch (dp->layout->pos) { 420 case TBL_CELL_HORIZ: 421 case TBL_CELL_DHORIZ: 422 sz = (*tbl->len)(1, tbl->arg); 423 if (col != NULL && col->width < sz) 424 col->width = sz; 425 return sz; 426 case TBL_CELL_LONG: 427 case TBL_CELL_CENTRE: 428 case TBL_CELL_LEFT: 429 case TBL_CELL_RIGHT: 430 return tblcalc_literal(tbl, col, dp, mw); 431 case TBL_CELL_NUMBER: 432 return tblcalc_number(tbl, col, opts, dp); 433 case TBL_CELL_DOWN: 434 return 0; 435 default: 436 abort(); 437 } 438 } 439 440 static size_t 441 tblcalc_literal(struct rofftbl *tbl, struct roffcol *col, 442 const struct tbl_dat *dp, size_t mw) 443 { 444 const char *str; /* Beginning of the first line. */ 445 const char *beg; /* Beginning of the current line. */ 446 char *end; /* End of the current line. */ 447 size_t lsz; /* Length of the current line. */ 448 size_t wsz; /* Length of the current word. */ 449 size_t msz; /* Length of the longest line. */ 450 451 if (dp->string == NULL || *dp->string == '\0') 452 return 0; 453 str = mw ? mandoc_strdup(dp->string) : dp->string; 454 msz = lsz = 0; 455 for (beg = str; beg != NULL && *beg != '\0'; beg = end) { 456 end = mw ? strchr(beg, ' ') : NULL; 457 if (end != NULL) { 458 *end++ = '\0'; 459 while (*end == ' ') 460 end++; 461 } 462 wsz = (*tbl->slen)(beg, tbl->arg); 463 if (mw && lsz && lsz + 1 + wsz <= mw) 464 lsz += 1 + wsz; 465 else 466 lsz = wsz; 467 if (msz < lsz) 468 msz = lsz; 469 } 470 if (mw) 471 free((void *)str); 472 if (col != NULL && col->width < msz) 473 col->width = msz; 474 return msz; 475 } 476 477 static size_t 478 tblcalc_number(struct rofftbl *tbl, struct roffcol *col, 479 const struct tbl_opts *opts, const struct tbl_dat *dp) 480 { 481 const char *cp, *lastdigit, *lastpoint; 482 size_t intsz, totsz; 483 char buf[2]; 484 485 if (dp->string == NULL || *dp->string == '\0') 486 return 0; 487 488 totsz = (*tbl->slen)(dp->string, tbl->arg); 489 if (col == NULL) 490 return totsz; 491 492 /* 493 * Find the last digit and 494 * the last decimal point that is adjacent to a digit. 495 * The alignment indicator "\&" overrides everything. 496 */ 497 498 lastdigit = lastpoint = NULL; 499 for (cp = dp->string; cp[0] != '\0'; cp++) { 500 if (cp[0] == '\\' && cp[1] == '&') { 501 lastdigit = lastpoint = cp; 502 break; 503 } else if (cp[0] == opts->decimal && 504 (isdigit((unsigned char)cp[1]) || 505 (cp > dp->string && isdigit((unsigned char)cp[-1])))) 506 lastpoint = cp; 507 else if (isdigit((unsigned char)cp[0])) 508 lastdigit = cp; 509 } 510 511 /* Not a number, treat as a literal string. */ 512 513 if (lastdigit == NULL) { 514 if (col != NULL && col->width < totsz) 515 col->width = totsz; 516 return totsz; 517 } 518 519 /* Measure the width of the integer part. */ 520 521 if (lastpoint == NULL) 522 lastpoint = lastdigit + 1; 523 intsz = 0; 524 buf[1] = '\0'; 525 for (cp = dp->string; cp < lastpoint; cp++) { 526 buf[0] = cp[0]; 527 intsz += (*tbl->slen)(buf, tbl->arg); 528 } 529 530 /* 531 * If this number has more integer digits than all numbers 532 * seen on earlier lines, shift them all to the right. 533 * If it has fewer, shift this number to the right. 534 */ 535 536 if (intsz > col->decimal) { 537 col->nwidth += intsz - col->decimal; 538 col->decimal = intsz; 539 } else 540 totsz += col->decimal - intsz; 541 542 /* Update the maximum total width seen so far. */ 543 544 if (totsz > col->nwidth) 545 col->nwidth = totsz; 546 if (col->nwidth > col->width) 547 col->width = col->nwidth; 548 return totsz; 549 } 550