1 /* $OpenBSD: out.c,v 1.57 2021/10/17 21:03:05 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 <sys/types.h> 20 21 #include <assert.h> 22 #include <ctype.h> 23 #include <stdint.h> 24 #include <stdio.h> 25 #include <stdlib.h> 26 #include <string.h> 27 #include <time.h> 28 29 #include "mandoc_aux.h" 30 #include "mandoc.h" 31 #include "tbl.h" 32 #include "out.h" 33 34 struct tbl_colgroup { 35 struct tbl_colgroup *next; 36 size_t wanted; 37 int startcol; 38 int endcol; 39 }; 40 41 static size_t tblcalc_data(struct rofftbl *, struct roffcol *, 42 const struct tbl_opts *, const struct tbl_dat *, 43 size_t); 44 static size_t tblcalc_literal(struct rofftbl *, struct roffcol *, 45 const struct tbl_dat *, size_t); 46 static size_t tblcalc_number(struct rofftbl *, struct roffcol *, 47 const struct tbl_opts *, const struct tbl_dat *); 48 49 50 /* 51 * Parse the *src string and store a scaling unit into *dst. 52 * If the string doesn't specify the unit, use the default. 53 * If no default is specified, fail. 54 * Return a pointer to the byte after the last byte used, 55 * or NULL on total failure. 56 */ 57 const char * 58 a2roffsu(const char *src, struct roffsu *dst, enum roffscale def) 59 { 60 char *endptr; 61 62 dst->unit = def == SCALE_MAX ? SCALE_BU : def; 63 dst->scale = strtod(src, &endptr); 64 if (endptr == src) 65 return NULL; 66 67 switch (*endptr++) { 68 case 'c': 69 dst->unit = SCALE_CM; 70 break; 71 case 'i': 72 dst->unit = SCALE_IN; 73 break; 74 case 'f': 75 dst->unit = SCALE_FS; 76 break; 77 case 'M': 78 dst->unit = SCALE_MM; 79 break; 80 case 'm': 81 dst->unit = SCALE_EM; 82 break; 83 case 'n': 84 dst->unit = SCALE_EN; 85 break; 86 case 'P': 87 dst->unit = SCALE_PC; 88 break; 89 case 'p': 90 dst->unit = SCALE_PT; 91 break; 92 case 'u': 93 dst->unit = SCALE_BU; 94 break; 95 case 'v': 96 dst->unit = SCALE_VS; 97 break; 98 default: 99 endptr--; 100 if (SCALE_MAX == def) 101 return NULL; 102 dst->unit = def; 103 break; 104 } 105 return endptr; 106 } 107 108 /* 109 * Calculate the abstract widths and decimal positions of columns in a 110 * table. This routine allocates the columns structures then runs over 111 * all rows and cells in the table. The function pointers in "tbl" are 112 * used for the actual width calculations. 113 */ 114 void 115 tblcalc(struct rofftbl *tbl, const struct tbl_span *sp_first, 116 size_t offset, size_t rmargin) 117 { 118 struct roffsu su; 119 const struct tbl_opts *opts; 120 const struct tbl_span *sp; 121 const struct tbl_dat *dp; 122 struct roffcol *col; 123 struct tbl_colgroup *first_group, **gp, *g; 124 size_t *colwidth; 125 size_t ewidth, min1, min2, wanted, width, xwidth; 126 int done, icol, maxcol, necol, nxcol, quirkcol; 127 128 /* 129 * Allocate the master column specifiers. These will hold the 130 * widths and decimal positions for all cells in the column. It 131 * must be freed and nullified by the caller. 132 */ 133 134 assert(tbl->cols == NULL); 135 tbl->cols = mandoc_calloc((size_t)sp_first->opts->cols, 136 sizeof(struct roffcol)); 137 opts = sp_first->opts; 138 139 maxcol = -1; 140 first_group = NULL; 141 for (sp = sp_first; sp != NULL; sp = sp->next) { 142 if (sp->pos != TBL_SPAN_DATA) 143 continue; 144 145 /* 146 * Account for the data cells in the layout, matching it 147 * to data cells in the data section. 148 */ 149 150 for (dp = sp->first; dp != NULL; dp = dp->next) { 151 icol = dp->layout->col; 152 while (maxcol < icol + dp->hspans) 153 tbl->cols[++maxcol].spacing = SIZE_MAX; 154 col = tbl->cols + icol; 155 col->flags |= dp->layout->flags; 156 if (dp->layout->flags & TBL_CELL_WIGN) 157 continue; 158 159 /* Handle explicit width specifications. */ 160 161 if (dp->layout->wstr != NULL && 162 dp->layout->width == 0 && 163 a2roffsu(dp->layout->wstr, &su, SCALE_EN) 164 != NULL) 165 dp->layout->width = 166 (*tbl->sulen)(&su, tbl->arg); 167 if (col->width < dp->layout->width) 168 col->width = dp->layout->width; 169 if (dp->layout->spacing != SIZE_MAX && 170 (col->spacing == SIZE_MAX || 171 col->spacing < dp->layout->spacing)) 172 col->spacing = dp->layout->spacing; 173 174 /* 175 * Calculate an automatic width. 176 * Except for spanning cells, apply it. 177 */ 178 179 width = tblcalc_data(tbl, 180 dp->hspans == 0 ? col : NULL, 181 opts, dp, 182 dp->block == 0 ? 0 : 183 dp->layout->width ? dp->layout->width : 184 rmargin ? (rmargin + sp->opts->cols / 2) 185 / (sp->opts->cols + 1) : 0); 186 if (dp->hspans == 0) 187 continue; 188 189 /* 190 * Build a singly linked list 191 * of all groups of columns joined by spans, 192 * recording the minimum width for each group. 193 */ 194 195 gp = &first_group; 196 while (*gp != NULL && ((*gp)->startcol != icol || 197 (*gp)->endcol != icol + dp->hspans)) 198 gp = &(*gp)->next; 199 if (*gp == NULL) { 200 g = mandoc_malloc(sizeof(*g)); 201 g->next = *gp; 202 g->wanted = width; 203 g->startcol = icol; 204 g->endcol = icol + dp->hspans; 205 *gp = g; 206 } else if ((*gp)->wanted < width) 207 (*gp)->wanted = width; 208 } 209 } 210 211 /* 212 * The minimum width of columns explicitly specified 213 * in the layout is 1n. 214 */ 215 216 if (maxcol < sp_first->opts->cols - 1) 217 maxcol = sp_first->opts->cols - 1; 218 for (icol = 0; icol <= maxcol; icol++) { 219 col = tbl->cols + icol; 220 if (col->width < 1) 221 col->width = 1; 222 223 /* 224 * Column spacings are needed for span width 225 * calculations, so set the default values now. 226 */ 227 228 if (col->spacing == SIZE_MAX || icol == maxcol) 229 col->spacing = 3; 230 } 231 232 /* 233 * Replace the minimum widths with the missing widths, 234 * and dismiss groups that are already wide enough. 235 */ 236 237 gp = &first_group; 238 while ((g = *gp) != NULL) { 239 done = 0; 240 for (icol = g->startcol; icol <= g->endcol; icol++) { 241 width = tbl->cols[icol].width; 242 if (icol < g->endcol) 243 width += tbl->cols[icol].spacing; 244 if (g->wanted <= width) { 245 done = 1; 246 break; 247 } else 248 g->wanted -= width; 249 } 250 if (done) { 251 *gp = g->next; 252 free(g); 253 } else 254 gp = &g->next; 255 } 256 257 colwidth = mandoc_reallocarray(NULL, maxcol + 1, sizeof(*colwidth)); 258 while (first_group != NULL) { 259 260 /* 261 * Rebuild the array of the widths of all columns 262 * participating in spans that require expansion. 263 */ 264 265 for (icol = 0; icol <= maxcol; icol++) 266 colwidth[icol] = SIZE_MAX; 267 for (g = first_group; g != NULL; g = g->next) 268 for (icol = g->startcol; icol <= g->endcol; icol++) 269 colwidth[icol] = tbl->cols[icol].width; 270 271 /* 272 * Find the smallest and second smallest column width 273 * among the columns which may need expamsion. 274 */ 275 276 min1 = min2 = SIZE_MAX; 277 for (icol = 0; icol <= maxcol; icol++) { 278 width = colwidth[icol]; 279 if (min1 > width) { 280 min2 = min1; 281 min1 = width; 282 } else if (min1 < width && min2 > width) 283 min2 = width; 284 } 285 286 /* 287 * Find the minimum wanted width 288 * for any one of the narrowest columns, 289 * and mark the columns wanting that width. 290 */ 291 292 wanted = min2; 293 for (g = first_group; g != NULL; g = g->next) { 294 necol = 0; 295 for (icol = g->startcol; icol <= g->endcol; icol++) 296 if (colwidth[icol] == min1) 297 necol++; 298 if (necol == 0) 299 continue; 300 width = min1 + (g->wanted - 1) / necol + 1; 301 if (width > min2) 302 width = min2; 303 if (wanted > width) 304 wanted = width; 305 } 306 307 /* Record the effect of the widening. */ 308 309 gp = &first_group; 310 while ((g = *gp) != NULL) { 311 done = 0; 312 for (icol = g->startcol; icol <= g->endcol; icol++) { 313 if (colwidth[icol] != min1) 314 continue; 315 if (g->wanted <= wanted - min1) { 316 tbl->cols[icol].width += g->wanted; 317 done = 1; 318 break; 319 } 320 tbl->cols[icol].width = wanted; 321 g->wanted -= wanted - min1; 322 } 323 if (done) { 324 *gp = g->next; 325 free(g); 326 } else 327 gp = &g->next; 328 } 329 } 330 free(colwidth); 331 332 /* 333 * Align numbers with text. 334 * Count columns to equalize and columns to maximize. 335 * Find maximum width of the columns to equalize. 336 * Find total width of the columns *not* to maximize. 337 */ 338 339 necol = nxcol = 0; 340 ewidth = xwidth = 0; 341 for (icol = 0; icol <= maxcol; icol++) { 342 col = tbl->cols + icol; 343 if (col->width > col->nwidth) 344 col->decimal += (col->width - col->nwidth) / 2; 345 if (col->flags & TBL_CELL_EQUAL) { 346 necol++; 347 if (ewidth < col->width) 348 ewidth = col->width; 349 } 350 if (col->flags & TBL_CELL_WMAX) 351 nxcol++; 352 else 353 xwidth += col->width; 354 } 355 356 /* 357 * Equalize columns, if requested for any of them. 358 * Update total width of the columns not to maximize. 359 */ 360 361 if (necol) { 362 for (icol = 0; icol <= maxcol; icol++) { 363 col = tbl->cols + icol; 364 if ( ! (col->flags & TBL_CELL_EQUAL)) 365 continue; 366 if (col->width == ewidth) 367 continue; 368 if (nxcol && rmargin) 369 xwidth += ewidth - col->width; 370 col->width = ewidth; 371 } 372 } 373 374 /* 375 * If there are any columns to maximize, find the total 376 * available width, deducting 3n margins between columns. 377 * Distribute the available width evenly. 378 */ 379 380 if (nxcol && rmargin) { 381 xwidth += 3*maxcol + 382 (opts->opts & (TBL_OPT_BOX | TBL_OPT_DBOX) ? 383 2 : !!opts->lvert + !!opts->rvert); 384 if (rmargin <= offset + xwidth) 385 return; 386 xwidth = rmargin - offset - xwidth; 387 388 /* 389 * Emulate a bug in GNU tbl width calculation that 390 * manifests itself for large numbers of x-columns. 391 * Emulating it for 5 x-columns gives identical 392 * behaviour for up to 6 x-columns. 393 */ 394 395 if (nxcol == 5) { 396 quirkcol = xwidth % nxcol + 2; 397 if (quirkcol != 3 && quirkcol != 4) 398 quirkcol = -1; 399 } else 400 quirkcol = -1; 401 402 necol = 0; 403 ewidth = 0; 404 for (icol = 0; icol <= maxcol; icol++) { 405 col = tbl->cols + icol; 406 if ( ! (col->flags & TBL_CELL_WMAX)) 407 continue; 408 col->width = (double)xwidth * ++necol / nxcol 409 - ewidth + 0.4995; 410 if (necol == quirkcol) 411 col->width--; 412 ewidth += col->width; 413 } 414 } 415 } 416 417 static size_t 418 tblcalc_data(struct rofftbl *tbl, struct roffcol *col, 419 const struct tbl_opts *opts, const struct tbl_dat *dp, size_t mw) 420 { 421 size_t sz; 422 423 /* Branch down into data sub-types. */ 424 425 switch (dp->layout->pos) { 426 case TBL_CELL_HORIZ: 427 case TBL_CELL_DHORIZ: 428 sz = (*tbl->len)(1, tbl->arg); 429 if (col != NULL && col->width < sz) 430 col->width = sz; 431 return sz; 432 case TBL_CELL_LONG: 433 case TBL_CELL_CENTRE: 434 case TBL_CELL_LEFT: 435 case TBL_CELL_RIGHT: 436 return tblcalc_literal(tbl, col, dp, mw); 437 case TBL_CELL_NUMBER: 438 return tblcalc_number(tbl, col, opts, dp); 439 case TBL_CELL_DOWN: 440 return 0; 441 default: 442 abort(); 443 } 444 } 445 446 static size_t 447 tblcalc_literal(struct rofftbl *tbl, struct roffcol *col, 448 const struct tbl_dat *dp, size_t mw) 449 { 450 const char *str; /* Beginning of the first line. */ 451 const char *beg; /* Beginning of the current line. */ 452 char *end; /* End of the current line. */ 453 size_t lsz; /* Length of the current line. */ 454 size_t wsz; /* Length of the current word. */ 455 size_t msz; /* Length of the longest line. */ 456 457 if (dp->string == NULL || *dp->string == '\0') 458 return 0; 459 str = mw ? mandoc_strdup(dp->string) : dp->string; 460 msz = lsz = 0; 461 for (beg = str; beg != NULL && *beg != '\0'; beg = end) { 462 end = mw ? strchr(beg, ' ') : NULL; 463 if (end != NULL) { 464 *end++ = '\0'; 465 while (*end == ' ') 466 end++; 467 } 468 wsz = (*tbl->slen)(beg, tbl->arg); 469 if (mw && lsz && lsz + 1 + wsz <= mw) 470 lsz += 1 + wsz; 471 else 472 lsz = wsz; 473 if (msz < lsz) 474 msz = lsz; 475 } 476 if (mw) 477 free((void *)str); 478 if (col != NULL && col->width < msz) 479 col->width = msz; 480 return msz; 481 } 482 483 static size_t 484 tblcalc_number(struct rofftbl *tbl, struct roffcol *col, 485 const struct tbl_opts *opts, const struct tbl_dat *dp) 486 { 487 const char *cp, *lastdigit, *lastpoint; 488 size_t intsz, totsz; 489 char buf[2]; 490 491 if (dp->string == NULL || *dp->string == '\0') 492 return 0; 493 494 totsz = (*tbl->slen)(dp->string, tbl->arg); 495 if (col == NULL) 496 return totsz; 497 498 /* 499 * Find the last digit and 500 * the last decimal point that is adjacent to a digit. 501 * The alignment indicator "\&" overrides everything. 502 */ 503 504 lastdigit = lastpoint = NULL; 505 for (cp = dp->string; cp[0] != '\0'; cp++) { 506 if (cp[0] == '\\' && cp[1] == '&') { 507 lastdigit = lastpoint = cp; 508 break; 509 } else if (cp[0] == opts->decimal && 510 (isdigit((unsigned char)cp[1]) || 511 (cp > dp->string && isdigit((unsigned char)cp[-1])))) 512 lastpoint = cp; 513 else if (isdigit((unsigned char)cp[0])) 514 lastdigit = cp; 515 } 516 517 /* Not a number, treat as a literal string. */ 518 519 if (lastdigit == NULL) { 520 if (col != NULL && col->width < totsz) 521 col->width = totsz; 522 return totsz; 523 } 524 525 /* Measure the width of the integer part. */ 526 527 if (lastpoint == NULL) 528 lastpoint = lastdigit + 1; 529 intsz = 0; 530 buf[1] = '\0'; 531 for (cp = dp->string; cp < lastpoint; cp++) { 532 buf[0] = cp[0]; 533 intsz += (*tbl->slen)(buf, tbl->arg); 534 } 535 536 /* 537 * If this number has more integer digits than all numbers 538 * seen on earlier lines, shift them all to the right. 539 * If it has fewer, shift this number to the right. 540 */ 541 542 if (intsz > col->decimal) { 543 col->nwidth += intsz - col->decimal; 544 col->decimal = intsz; 545 } else 546 totsz += col->decimal - intsz; 547 548 /* Update the maximum total width seen so far. */ 549 550 if (totsz > col->nwidth) 551 col->nwidth = totsz; 552 if (col->nwidth > col->width) 553 col->width = col->nwidth; 554 return totsz; 555 } 556