1 /* $Id: term.c,v 1.214 2013/12/25 00:39:31 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010, 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 */ 18 #ifdef HAVE_CONFIG_H 19 #include "config.h" 20 #endif 21 22 #include <sys/types.h> 23 24 #include <assert.h> 25 #include <ctype.h> 26 #include <stdint.h> 27 #include <stdio.h> 28 #include <stdlib.h> 29 #include <string.h> 30 31 #include "mandoc.h" 32 #include "out.h" 33 #include "term.h" 34 #include "main.h" 35 36 static size_t cond_width(const struct termp *, int, int *); 37 static void adjbuf(struct termp *p, size_t); 38 static void bufferc(struct termp *, char); 39 static void encode(struct termp *, const char *, size_t); 40 static void encode1(struct termp *, int); 41 42 void 43 term_free(struct termp *p) 44 { 45 46 if (p->buf) 47 free(p->buf); 48 if (p->symtab) 49 mchars_free(p->symtab); 50 51 free(p); 52 } 53 54 55 void 56 term_begin(struct termp *p, term_margin head, 57 term_margin foot, const void *arg) 58 { 59 60 p->headf = head; 61 p->footf = foot; 62 p->argf = arg; 63 (*p->begin)(p); 64 } 65 66 67 void 68 term_end(struct termp *p) 69 { 70 71 (*p->end)(p); 72 } 73 74 /* 75 * Flush a line of text. A "line" is loosely defined as being something 76 * that should be followed by a newline, regardless of whether it's 77 * broken apart by newlines getting there. A line can also be a 78 * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does 79 * not have a trailing newline. 80 * 81 * The following flags may be specified: 82 * 83 * - TERMP_NOBREAK: this is the most important and is used when making 84 * columns. In short: don't print a newline and instead expect the 85 * next call to do the padding up to the start of the next column. 86 * p->trailspace may be set to 0, 1, or 2, depending on how many 87 * space characters are required at the end of the column. 88 * 89 * - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and 90 * the line is overrun, and don't pad-right if it's underrun. 91 * 92 * - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when 93 * overrunning, instead save the position and continue at that point 94 * when the next invocation. 95 * 96 * In-line line breaking: 97 * 98 * If TERMP_NOBREAK is specified and the line overruns the right 99 * margin, it will break and pad-right to the right margin after 100 * writing. If maxrmargin is violated, it will break and continue 101 * writing from the right-margin, which will lead to the above scenario 102 * upon exit. Otherwise, the line will break at the right margin. 103 */ 104 void 105 term_flushln(struct termp *p) 106 { 107 size_t i; /* current input position in p->buf */ 108 int ntab; /* number of tabs to prepend */ 109 size_t vis; /* current visual position on output */ 110 size_t vbl; /* number of blanks to prepend to output */ 111 size_t vend; /* end of word visual position on output */ 112 size_t bp; /* visual right border position */ 113 size_t dv; /* temporary for visual pos calculations */ 114 size_t j; /* temporary loop index for p->buf */ 115 size_t jhy; /* last hyph before overflow w/r/t j */ 116 size_t maxvis; /* output position of visible boundary */ 117 size_t mmax; /* used in calculating bp */ 118 119 /* 120 * First, establish the maximum columns of "visible" content. 121 * This is usually the difference between the right-margin and 122 * an indentation, but can be, for tagged lists or columns, a 123 * small set of values. 124 * 125 * The following unsigned-signed subtractions look strange, 126 * but they are actually correct. If the int p->overstep 127 * is negative, it gets sign extended. Subtracting that 128 * very large size_t effectively adds a small number to dv. 129 */ 130 assert (p->rmargin >= p->offset); 131 dv = p->rmargin - p->offset; 132 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0; 133 dv = p->maxrmargin - p->offset; 134 mmax = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0; 135 136 bp = TERMP_NOBREAK & p->flags ? mmax : maxvis; 137 138 /* 139 * Calculate the required amount of padding. 140 */ 141 vbl = p->offset + p->overstep > p->viscol ? 142 p->offset + p->overstep - p->viscol : 0; 143 144 vis = vend = 0; 145 i = 0; 146 147 while (i < p->col) { 148 /* 149 * Handle literal tab characters: collapse all 150 * subsequent tabs into a single huge set of spaces. 151 */ 152 ntab = 0; 153 while (i < p->col && '\t' == p->buf[i]) { 154 vend = (vis / p->tabwidth + 1) * p->tabwidth; 155 vbl += vend - vis; 156 vis = vend; 157 ntab++; 158 i++; 159 } 160 161 /* 162 * Count up visible word characters. Control sequences 163 * (starting with the CSI) aren't counted. A space 164 * generates a non-printing word, which is valid (the 165 * space is printed according to regular spacing rules). 166 */ 167 168 for (j = i, jhy = 0; j < p->col; j++) { 169 if (' ' == p->buf[j] || '\t' == p->buf[j]) 170 break; 171 172 /* Back over the the last printed character. */ 173 if (8 == p->buf[j]) { 174 assert(j); 175 vend -= (*p->width)(p, p->buf[j - 1]); 176 continue; 177 } 178 179 /* Regular word. */ 180 /* Break at the hyphen point if we overrun. */ 181 if (vend > vis && vend < bp && 182 ASCII_HYPH == p->buf[j]) 183 jhy = j; 184 185 vend += (*p->width)(p, p->buf[j]); 186 } 187 188 /* 189 * Find out whether we would exceed the right margin. 190 * If so, break to the next line. 191 */ 192 if (vend > bp && 0 == jhy && vis > 0) { 193 vend -= vis; 194 (*p->endline)(p); 195 p->viscol = 0; 196 if (TERMP_NOBREAK & p->flags) { 197 vbl = p->rmargin; 198 vend += p->rmargin - p->offset; 199 } else 200 vbl = p->offset; 201 202 /* use pending tabs on the new line */ 203 204 if (0 < ntab) 205 vbl += ntab * p->tabwidth; 206 207 /* 208 * Remove the p->overstep width. 209 * Again, if p->overstep is negative, 210 * sign extension does the right thing. 211 */ 212 213 bp += (size_t)p->overstep; 214 p->overstep = 0; 215 } 216 217 /* Write out the [remaining] word. */ 218 for ( ; i < p->col; i++) { 219 if (vend > bp && jhy > 0 && i > jhy) 220 break; 221 if ('\t' == p->buf[i]) 222 break; 223 if (' ' == p->buf[i]) { 224 j = i; 225 while (' ' == p->buf[i]) 226 i++; 227 dv = (i - j) * (*p->width)(p, ' '); 228 vbl += dv; 229 vend += dv; 230 break; 231 } 232 if (ASCII_NBRSP == p->buf[i]) { 233 vbl += (*p->width)(p, ' '); 234 continue; 235 } 236 237 /* 238 * Now we definitely know there will be 239 * printable characters to output, 240 * so write preceding white space now. 241 */ 242 if (vbl) { 243 (*p->advance)(p, vbl); 244 p->viscol += vbl; 245 vbl = 0; 246 } 247 248 if (ASCII_HYPH == p->buf[i]) { 249 (*p->letter)(p, '-'); 250 p->viscol += (*p->width)(p, '-'); 251 continue; 252 } 253 254 (*p->letter)(p, p->buf[i]); 255 if (8 == p->buf[i]) 256 p->viscol -= (*p->width)(p, p->buf[i-1]); 257 else 258 p->viscol += (*p->width)(p, p->buf[i]); 259 } 260 vis = vend; 261 } 262 263 /* 264 * If there was trailing white space, it was not printed; 265 * so reset the cursor position accordingly. 266 */ 267 if (vis) 268 vis -= vbl; 269 270 p->col = 0; 271 p->overstep = 0; 272 273 if ( ! (TERMP_NOBREAK & p->flags)) { 274 p->viscol = 0; 275 (*p->endline)(p); 276 return; 277 } 278 279 if (TERMP_HANG & p->flags) { 280 p->overstep = (int)(vis - maxvis + 281 p->trailspace * (*p->width)(p, ' ')); 282 283 /* 284 * If we have overstepped the margin, temporarily move 285 * it to the right and flag the rest of the line to be 286 * shorter. 287 * If there is a request to keep the columns together, 288 * allow negative overstep when the column is not full. 289 */ 290 if (p->trailspace && p->overstep < 0) 291 p->overstep = 0; 292 return; 293 294 } else if (TERMP_DANGLE & p->flags) 295 return; 296 297 /* If the column was overrun, break the line. */ 298 if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) { 299 (*p->endline)(p); 300 p->viscol = 0; 301 } 302 } 303 304 305 /* 306 * A newline only breaks an existing line; it won't assert vertical 307 * space. All data in the output buffer is flushed prior to the newline 308 * assertion. 309 */ 310 void 311 term_newln(struct termp *p) 312 { 313 314 p->flags |= TERMP_NOSPACE; 315 if (p->col || p->viscol) 316 term_flushln(p); 317 } 318 319 320 /* 321 * Asserts a vertical space (a full, empty line-break between lines). 322 * Note that if used twice, this will cause two blank spaces and so on. 323 * All data in the output buffer is flushed prior to the newline 324 * assertion. 325 */ 326 void 327 term_vspace(struct termp *p) 328 { 329 330 term_newln(p); 331 p->viscol = 0; 332 if (0 < p->skipvsp) 333 p->skipvsp--; 334 else 335 (*p->endline)(p); 336 } 337 338 void 339 term_fontlast(struct termp *p) 340 { 341 enum termfont f; 342 343 f = p->fontl; 344 p->fontl = p->fontq[p->fonti]; 345 p->fontq[p->fonti] = f; 346 } 347 348 349 void 350 term_fontrepl(struct termp *p, enum termfont f) 351 { 352 353 p->fontl = p->fontq[p->fonti]; 354 p->fontq[p->fonti] = f; 355 } 356 357 358 void 359 term_fontpush(struct termp *p, enum termfont f) 360 { 361 362 assert(p->fonti + 1 < 10); 363 p->fontl = p->fontq[p->fonti]; 364 p->fontq[++p->fonti] = f; 365 } 366 367 368 const void * 369 term_fontq(struct termp *p) 370 { 371 372 return(&p->fontq[p->fonti]); 373 } 374 375 376 enum termfont 377 term_fonttop(struct termp *p) 378 { 379 380 return(p->fontq[p->fonti]); 381 } 382 383 384 void 385 term_fontpopq(struct termp *p, const void *key) 386 { 387 388 while (p->fonti >= 0 && key < (void *)(p->fontq + p->fonti)) 389 p->fonti--; 390 assert(p->fonti >= 0); 391 } 392 393 394 void 395 term_fontpop(struct termp *p) 396 { 397 398 assert(p->fonti); 399 p->fonti--; 400 } 401 402 /* 403 * Handle pwords, partial words, which may be either a single word or a 404 * phrase that cannot be broken down (such as a literal string). This 405 * handles word styling. 406 */ 407 void 408 term_word(struct termp *p, const char *word) 409 { 410 const char nbrsp[2] = { ASCII_NBRSP, 0 }; 411 const char *seq, *cp; 412 char c; 413 int sz, uc; 414 size_t ssz; 415 enum mandoc_esc esc; 416 417 if ( ! (TERMP_NOSPACE & p->flags)) { 418 if ( ! (TERMP_KEEP & p->flags)) { 419 bufferc(p, ' '); 420 if (TERMP_SENTENCE & p->flags) 421 bufferc(p, ' '); 422 } else 423 bufferc(p, ASCII_NBRSP); 424 } 425 if (TERMP_PREKEEP & p->flags) 426 p->flags |= TERMP_KEEP; 427 428 if ( ! (p->flags & TERMP_NONOSPACE)) 429 p->flags &= ~TERMP_NOSPACE; 430 else 431 p->flags |= TERMP_NOSPACE; 432 433 p->flags &= ~TERMP_SENTENCE; 434 435 while ('\0' != *word) { 436 if ('\\' != *word) { 437 if (TERMP_SKIPCHAR & p->flags) { 438 p->flags &= ~TERMP_SKIPCHAR; 439 word++; 440 continue; 441 } 442 if (TERMP_NBRWORD & p->flags) { 443 if (' ' == *word) { 444 encode(p, nbrsp, 1); 445 word++; 446 continue; 447 } 448 ssz = strcspn(word, "\\ "); 449 } else 450 ssz = strcspn(word, "\\"); 451 encode(p, word, ssz); 452 word += (int)ssz; 453 continue; 454 } 455 456 word++; 457 esc = mandoc_escape(&word, &seq, &sz); 458 if (ESCAPE_ERROR == esc) 459 break; 460 461 if (TERMENC_ASCII != p->enc) 462 switch (esc) { 463 case (ESCAPE_UNICODE): 464 uc = mchars_num2uc(seq + 1, sz - 1); 465 if ('\0' == uc) 466 break; 467 encode1(p, uc); 468 continue; 469 case (ESCAPE_SPECIAL): 470 uc = mchars_spec2cp(p->symtab, seq, sz); 471 if (uc <= 0) 472 break; 473 encode1(p, uc); 474 continue; 475 default: 476 break; 477 } 478 479 switch (esc) { 480 case (ESCAPE_UNICODE): 481 encode1(p, '?'); 482 break; 483 case (ESCAPE_NUMBERED): 484 c = mchars_num2char(seq, sz); 485 if ('\0' != c) 486 encode(p, &c, 1); 487 break; 488 case (ESCAPE_SPECIAL): 489 cp = mchars_spec2str(p->symtab, seq, sz, &ssz); 490 if (NULL != cp) 491 encode(p, cp, ssz); 492 else if (1 == ssz) 493 encode(p, seq, sz); 494 break; 495 case (ESCAPE_FONTBOLD): 496 term_fontrepl(p, TERMFONT_BOLD); 497 break; 498 case (ESCAPE_FONTITALIC): 499 term_fontrepl(p, TERMFONT_UNDER); 500 break; 501 case (ESCAPE_FONTBI): 502 term_fontrepl(p, TERMFONT_BI); 503 break; 504 case (ESCAPE_FONT): 505 /* FALLTHROUGH */ 506 case (ESCAPE_FONTROMAN): 507 term_fontrepl(p, TERMFONT_NONE); 508 break; 509 case (ESCAPE_FONTPREV): 510 term_fontlast(p); 511 break; 512 case (ESCAPE_NOSPACE): 513 if (TERMP_SKIPCHAR & p->flags) 514 p->flags &= ~TERMP_SKIPCHAR; 515 else if ('\0' == *word) 516 p->flags |= TERMP_NOSPACE; 517 break; 518 case (ESCAPE_SKIPCHAR): 519 p->flags |= TERMP_SKIPCHAR; 520 break; 521 default: 522 break; 523 } 524 } 525 p->flags &= ~TERMP_NBRWORD; 526 } 527 528 static void 529 adjbuf(struct termp *p, size_t sz) 530 { 531 532 if (0 == p->maxcols) 533 p->maxcols = 1024; 534 while (sz >= p->maxcols) 535 p->maxcols <<= 2; 536 537 p->buf = mandoc_realloc(p->buf, sizeof(int) * p->maxcols); 538 } 539 540 static void 541 bufferc(struct termp *p, char c) 542 { 543 544 if (p->col + 1 >= p->maxcols) 545 adjbuf(p, p->col + 1); 546 547 p->buf[p->col++] = c; 548 } 549 550 /* 551 * See encode(). 552 * Do this for a single (probably unicode) value. 553 * Does not check for non-decorated glyphs. 554 */ 555 static void 556 encode1(struct termp *p, int c) 557 { 558 enum termfont f; 559 560 if (TERMP_SKIPCHAR & p->flags) { 561 p->flags &= ~TERMP_SKIPCHAR; 562 return; 563 } 564 565 if (p->col + 6 >= p->maxcols) 566 adjbuf(p, p->col + 6); 567 568 f = term_fonttop(p); 569 570 if (TERMFONT_UNDER == f || TERMFONT_BI == f) { 571 p->buf[p->col++] = '_'; 572 p->buf[p->col++] = 8; 573 } 574 if (TERMFONT_BOLD == f || TERMFONT_BI == f) { 575 if (ASCII_HYPH == c) 576 p->buf[p->col++] = '-'; 577 else 578 p->buf[p->col++] = c; 579 p->buf[p->col++] = 8; 580 } 581 p->buf[p->col++] = c; 582 } 583 584 static void 585 encode(struct termp *p, const char *word, size_t sz) 586 { 587 size_t i; 588 589 if (TERMP_SKIPCHAR & p->flags) { 590 p->flags &= ~TERMP_SKIPCHAR; 591 return; 592 } 593 594 /* 595 * Encode and buffer a string of characters. If the current 596 * font mode is unset, buffer directly, else encode then buffer 597 * character by character. 598 */ 599 600 if (TERMFONT_NONE == term_fonttop(p)) { 601 if (p->col + sz >= p->maxcols) 602 adjbuf(p, p->col + sz); 603 for (i = 0; i < sz; i++) 604 p->buf[p->col++] = word[i]; 605 return; 606 } 607 608 /* Pre-buffer, assuming worst-case. */ 609 610 if (p->col + 1 + (sz * 5) >= p->maxcols) 611 adjbuf(p, p->col + 1 + (sz * 5)); 612 613 for (i = 0; i < sz; i++) { 614 if (ASCII_HYPH == word[i] || 615 isgraph((unsigned char)word[i])) 616 encode1(p, word[i]); 617 else 618 p->buf[p->col++] = word[i]; 619 } 620 } 621 622 size_t 623 term_len(const struct termp *p, size_t sz) 624 { 625 626 return((*p->width)(p, ' ') * sz); 627 } 628 629 static size_t 630 cond_width(const struct termp *p, int c, int *skip) 631 { 632 633 if (*skip) { 634 (*skip) = 0; 635 return(0); 636 } else 637 return((*p->width)(p, c)); 638 } 639 640 size_t 641 term_strlen(const struct termp *p, const char *cp) 642 { 643 size_t sz, rsz, i; 644 int ssz, skip, c; 645 const char *seq, *rhs; 646 enum mandoc_esc esc; 647 static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' }; 648 649 /* 650 * Account for escaped sequences within string length 651 * calculations. This follows the logic in term_word() as we 652 * must calculate the width of produced strings. 653 */ 654 655 sz = 0; 656 skip = 0; 657 while ('\0' != *cp) { 658 rsz = strcspn(cp, rej); 659 for (i = 0; i < rsz; i++) 660 sz += cond_width(p, *cp++, &skip); 661 662 c = 0; 663 switch (*cp) { 664 case ('\\'): 665 cp++; 666 esc = mandoc_escape(&cp, &seq, &ssz); 667 if (ESCAPE_ERROR == esc) 668 return(sz); 669 670 if (TERMENC_ASCII != p->enc) 671 switch (esc) { 672 case (ESCAPE_UNICODE): 673 c = mchars_num2uc 674 (seq + 1, ssz - 1); 675 if ('\0' == c) 676 break; 677 sz += cond_width(p, c, &skip); 678 continue; 679 case (ESCAPE_SPECIAL): 680 c = mchars_spec2cp 681 (p->symtab, seq, ssz); 682 if (c <= 0) 683 break; 684 sz += cond_width(p, c, &skip); 685 continue; 686 default: 687 break; 688 } 689 690 rhs = NULL; 691 692 switch (esc) { 693 case (ESCAPE_UNICODE): 694 sz += cond_width(p, '?', &skip); 695 break; 696 case (ESCAPE_NUMBERED): 697 c = mchars_num2char(seq, ssz); 698 if ('\0' != c) 699 sz += cond_width(p, c, &skip); 700 break; 701 case (ESCAPE_SPECIAL): 702 rhs = mchars_spec2str 703 (p->symtab, seq, ssz, &rsz); 704 705 if (ssz != 1 || rhs) 706 break; 707 708 rhs = seq; 709 rsz = ssz; 710 break; 711 case (ESCAPE_SKIPCHAR): 712 skip = 1; 713 break; 714 default: 715 break; 716 } 717 718 if (NULL == rhs) 719 break; 720 721 if (skip) { 722 skip = 0; 723 break; 724 } 725 726 for (i = 0; i < rsz; i++) 727 sz += (*p->width)(p, *rhs++); 728 break; 729 case (ASCII_NBRSP): 730 sz += cond_width(p, ' ', &skip); 731 cp++; 732 break; 733 case (ASCII_HYPH): 734 sz += cond_width(p, '-', &skip); 735 cp++; 736 break; 737 default: 738 break; 739 } 740 } 741 742 return(sz); 743 } 744 745 /* ARGSUSED */ 746 size_t 747 term_vspan(const struct termp *p, const struct roffsu *su) 748 { 749 double r; 750 751 switch (su->unit) { 752 case (SCALE_CM): 753 r = su->scale * 2; 754 break; 755 case (SCALE_IN): 756 r = su->scale * 6; 757 break; 758 case (SCALE_PC): 759 r = su->scale; 760 break; 761 case (SCALE_PT): 762 r = su->scale / 8; 763 break; 764 case (SCALE_MM): 765 r = su->scale / 1000; 766 break; 767 case (SCALE_VS): 768 r = su->scale; 769 break; 770 default: 771 r = su->scale - 1; 772 break; 773 } 774 775 if (r < 0.0) 776 r = 0.0; 777 return(/* LINTED */(size_t) 778 r); 779 } 780 781 size_t 782 term_hspan(const struct termp *p, const struct roffsu *su) 783 { 784 double v; 785 786 v = ((*p->hspan)(p, su)); 787 if (v < 0.0) 788 v = 0.0; 789 return((size_t) /* LINTED */ 790 v); 791 } 792