]> git.cameronkatri.com Git - mandoc.git/blob - term.c
Properly reinitialize roffce_node between parses,
[mandoc.git] / term.c
1 /* $Id: term.c,v 1.268 2017/06/08 12:54:58 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2017 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 AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 #include "config.h"
19
20 #include <sys/types.h>
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27
28 #include "mandoc.h"
29 #include "mandoc_aux.h"
30 #include "out.h"
31 #include "term.h"
32 #include "main.h"
33
34 static size_t cond_width(const struct termp *, int, int *);
35 static void adjbuf(struct termp_col *, size_t);
36 static void bufferc(struct termp *, char);
37 static void encode(struct termp *, const char *, size_t);
38 static void encode1(struct termp *, int);
39 static void endline(struct termp *);
40
41
42 void
43 term_free(struct termp *p)
44 {
45 for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++)
46 free(p->tcol->buf);
47 free(p->tcols);
48 free(p->fontq);
49 free(p);
50 }
51
52 void
53 term_begin(struct termp *p, term_margin head,
54 term_margin foot, const struct roff_meta *arg)
55 {
56
57 p->headf = head;
58 p->footf = foot;
59 p->argf = arg;
60 (*p->begin)(p);
61 }
62
63 void
64 term_end(struct termp *p)
65 {
66
67 (*p->end)(p);
68 }
69
70 /*
71 * Flush a chunk of text. By default, break the output line each time
72 * the right margin is reached, and continue output on the next line
73 * at the same offset as the chunk itself. By default, also break the
74 * output line at the end of the chunk.
75 * The following flags may be specified:
76 *
77 * - TERMP_NOBREAK: Do not break the output line at the right margin,
78 * but only at the max right margin. Also, do not break the output
79 * line at the end of the chunk, such that the next call can pad to
80 * the next column. However, if less than p->trailspace blanks,
81 * which can be 0, 1, or 2, remain to the right margin, the line
82 * will be broken.
83 * - TERMP_BRTRSP: Consider trailing whitespace significant
84 * when deciding whether the chunk fits or not.
85 * - TERMP_BRIND: If the chunk does not fit and the output line has
86 * to be broken, start the next line at the right margin instead
87 * of at the offset. Used together with TERMP_NOBREAK for the tags
88 * in various kinds of tagged lists.
89 * - TERMP_HANG: Do not break the output line at the right margin,
90 * append the next chunk after it even if this one is too long.
91 * To be used together with TERMP_NOBREAK.
92 * - TERMP_NOPAD: Start writing at the current position,
93 * do not pad with blank characters up to the offset.
94 */
95 void
96 term_flushln(struct termp *p)
97 {
98 size_t vis; /* current visual position on output */
99 size_t vbl; /* number of blanks to prepend to output */
100 size_t vend; /* end of word visual position on output */
101 size_t bp; /* visual right border position */
102 size_t dv; /* temporary for visual pos calculations */
103 size_t j; /* temporary loop index for p->tcol->buf */
104 size_t jhy; /* last hyph before overflow w/r/t j */
105 size_t maxvis; /* output position of visible boundary */
106 int ntab; /* number of tabs to prepend */
107
108 vbl = (p->flags & TERMP_NOPAD) || p->tcol->offset < p->viscol ?
109 0 : p->tcol->offset - p->viscol;
110 if (p->minbl && vbl < p->minbl)
111 vbl = p->minbl;
112 maxvis = p->tcol->rmargin > p->viscol + vbl ?
113 p->tcol->rmargin - p->viscol - vbl : 0;
114 bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
115 p->maxrmargin > p->viscol + vbl ?
116 p->maxrmargin - p->viscol - vbl : 0;
117 vis = vend = 0;
118
119 if (p->lasttcol == 0)
120 p->tcol->col = 0;
121 while (p->tcol->col < p->lastcol) {
122
123 /*
124 * Handle literal tab characters: collapse all
125 * subsequent tabs into a single huge set of spaces.
126 */
127
128 ntab = 0;
129 while (p->tcol->col < p->lastcol &&
130 p->tcol->buf[p->tcol->col] == '\t') {
131 vend = term_tab_next(vis);
132 vbl += vend - vis;
133 vis = vend;
134 ntab++;
135 p->tcol->col++;
136 }
137
138 /*
139 * Count up visible word characters. Control sequences
140 * (starting with the CSI) aren't counted. A space
141 * generates a non-printing word, which is valid (the
142 * space is printed according to regular spacing rules).
143 */
144
145 jhy = 0;
146 for (j = p->tcol->col; j < p->lastcol; j++) {
147 if (p->tcol->buf[j] == ' ' || p->tcol->buf[j] == '\t')
148 break;
149
150 /* Back over the last printed character. */
151 if (p->tcol->buf[j] == '\b') {
152 assert(j);
153 vend -= (*p->width)(p, p->tcol->buf[j - 1]);
154 continue;
155 }
156
157 /* Regular word. */
158 /* Break at the hyphen point if we overrun. */
159 if (vend > vis && vend < bp &&
160 (p->tcol->buf[j] == ASCII_HYPH||
161 p->tcol->buf[j] == ASCII_BREAK))
162 jhy = j;
163
164 /*
165 * Hyphenation now decided, put back a real
166 * hyphen such that we get the correct width.
167 */
168 if (p->tcol->buf[j] == ASCII_HYPH)
169 p->tcol->buf[j] = '-';
170
171 vend += (*p->width)(p, p->tcol->buf[j]);
172 }
173
174 /*
175 * Find out whether we would exceed the right margin.
176 * If so, break to the next line.
177 */
178
179 if (vend > bp && jhy == 0 && vis > 0 &&
180 (p->flags & TERMP_BRNEVER) == 0) {
181 if (p->lasttcol)
182 return;
183
184 endline(p);
185 vend -= vis;
186
187 /* Use pending tabs on the new line. */
188
189 vbl = 0;
190 while (ntab--)
191 vbl = term_tab_next(vbl);
192
193 /* Re-establish indentation. */
194
195 if (p->flags & TERMP_BRIND)
196 vbl += p->tcol->rmargin;
197 else
198 vbl += p->tcol->offset;
199 maxvis = p->tcol->rmargin > vbl ?
200 p->tcol->rmargin - vbl : 0;
201 bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
202 p->maxrmargin > vbl ? p->maxrmargin - vbl : 0;
203 }
204
205 /*
206 * Write out the rest of the word.
207 */
208
209 for ( ; p->tcol->col < p->lastcol; p->tcol->col++) {
210 if (vend > bp && jhy > 0 && p->tcol->col > jhy)
211 break;
212 if (p->tcol->buf[p->tcol->col] == '\t')
213 break;
214 if (p->tcol->buf[p->tcol->col] == ' ') {
215 j = p->tcol->col;
216 while (p->tcol->col < p->lastcol &&
217 p->tcol->buf[p->tcol->col] == ' ')
218 p->tcol->col++;
219 dv = (p->tcol->col - j) * (*p->width)(p, ' ');
220 vbl += dv;
221 vend += dv;
222 break;
223 }
224 if (p->tcol->buf[p->tcol->col] == ASCII_NBRSP) {
225 vbl += (*p->width)(p, ' ');
226 continue;
227 }
228 if (p->tcol->buf[p->tcol->col] == ASCII_BREAK)
229 continue;
230
231 /*
232 * Now we definitely know there will be
233 * printable characters to output,
234 * so write preceding white space now.
235 */
236 if (vbl) {
237 (*p->advance)(p, vbl);
238 p->viscol += vbl;
239 vbl = 0;
240 }
241
242 (*p->letter)(p, p->tcol->buf[p->tcol->col]);
243 if (p->tcol->buf[p->tcol->col] == '\b')
244 p->viscol -= (*p->width)(p,
245 p->tcol->buf[p->tcol->col - 1]);
246 else
247 p->viscol += (*p->width)(p,
248 p->tcol->buf[p->tcol->col]);
249 }
250 vis = vend;
251 }
252
253 /*
254 * If there was trailing white space, it was not printed;
255 * so reset the cursor position accordingly.
256 */
257
258 if (vis > vbl)
259 vis -= vbl;
260 else
261 vis = 0;
262
263 p->col = p->lastcol = 0;
264 p->minbl = p->trailspace;
265 p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
266
267 /* Trailing whitespace is significant in some columns. */
268
269 if (vis && vbl && (TERMP_BRTRSP & p->flags))
270 vis += vbl;
271
272 /* If the column was overrun, break the line. */
273 if ((p->flags & TERMP_NOBREAK) == 0 ||
274 ((p->flags & TERMP_HANG) == 0 &&
275 vis + p->trailspace * (*p->width)(p, ' ') > maxvis))
276 endline(p);
277 }
278
279 static void
280 endline(struct termp *p)
281 {
282 if ((p->flags & (TERMP_NEWMC | TERMP_ENDMC)) == TERMP_ENDMC) {
283 p->mc = NULL;
284 p->flags &= ~TERMP_ENDMC;
285 }
286 if (p->mc != NULL) {
287 if (p->viscol && p->maxrmargin >= p->viscol)
288 (*p->advance)(p, p->maxrmargin - p->viscol + 1);
289 p->flags |= TERMP_NOBUF | TERMP_NOSPACE;
290 term_word(p, p->mc);
291 p->flags &= ~(TERMP_NOBUF | TERMP_NEWMC);
292 }
293 p->viscol = 0;
294 p->minbl = 0;
295 (*p->endline)(p);
296 }
297
298 /*
299 * A newline only breaks an existing line; it won't assert vertical
300 * space. All data in the output buffer is flushed prior to the newline
301 * assertion.
302 */
303 void
304 term_newln(struct termp *p)
305 {
306
307 p->flags |= TERMP_NOSPACE;
308 if (p->lastcol || p->viscol)
309 term_flushln(p);
310 }
311
312 /*
313 * Asserts a vertical space (a full, empty line-break between lines).
314 * Note that if used twice, this will cause two blank spaces and so on.
315 * All data in the output buffer is flushed prior to the newline
316 * assertion.
317 */
318 void
319 term_vspace(struct termp *p)
320 {
321
322 term_newln(p);
323 p->viscol = 0;
324 p->minbl = 0;
325 if (0 < p->skipvsp)
326 p->skipvsp--;
327 else
328 (*p->endline)(p);
329 }
330
331 /* Swap current and previous font; for \fP and .ft P */
332 void
333 term_fontlast(struct termp *p)
334 {
335 enum termfont f;
336
337 f = p->fontl;
338 p->fontl = p->fontq[p->fonti];
339 p->fontq[p->fonti] = f;
340 }
341
342 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
343 void
344 term_fontrepl(struct termp *p, enum termfont f)
345 {
346
347 p->fontl = p->fontq[p->fonti];
348 p->fontq[p->fonti] = f;
349 }
350
351 /* Set font, save previous. */
352 void
353 term_fontpush(struct termp *p, enum termfont f)
354 {
355
356 p->fontl = p->fontq[p->fonti];
357 if (++p->fonti == p->fontsz) {
358 p->fontsz += 8;
359 p->fontq = mandoc_reallocarray(p->fontq,
360 p->fontsz, sizeof(*p->fontq));
361 }
362 p->fontq[p->fonti] = f;
363 }
364
365 /* Flush to make the saved pointer current again. */
366 void
367 term_fontpopq(struct termp *p, int i)
368 {
369
370 assert(i >= 0);
371 if (p->fonti > i)
372 p->fonti = i;
373 }
374
375 /* Pop one font off the stack. */
376 void
377 term_fontpop(struct termp *p)
378 {
379
380 assert(p->fonti);
381 p->fonti--;
382 }
383
384 /*
385 * Handle pwords, partial words, which may be either a single word or a
386 * phrase that cannot be broken down (such as a literal string). This
387 * handles word styling.
388 */
389 void
390 term_word(struct termp *p, const char *word)
391 {
392 struct roffsu su;
393 const char nbrsp[2] = { ASCII_NBRSP, 0 };
394 const char *seq, *cp;
395 int sz, uc;
396 size_t csz, lsz, ssz;
397 enum mandoc_esc esc;
398
399 if ((p->flags & TERMP_NOBUF) == 0) {
400 if ((p->flags & TERMP_NOSPACE) == 0) {
401 if ((p->flags & TERMP_KEEP) == 0) {
402 bufferc(p, ' ');
403 if (p->flags & TERMP_SENTENCE)
404 bufferc(p, ' ');
405 } else
406 bufferc(p, ASCII_NBRSP);
407 }
408 if (p->flags & TERMP_PREKEEP)
409 p->flags |= TERMP_KEEP;
410 if (p->flags & TERMP_NONOSPACE)
411 p->flags |= TERMP_NOSPACE;
412 else
413 p->flags &= ~TERMP_NOSPACE;
414 p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);
415 p->skipvsp = 0;
416 }
417
418 while ('\0' != *word) {
419 if ('\\' != *word) {
420 if (TERMP_NBRWORD & p->flags) {
421 if (' ' == *word) {
422 encode(p, nbrsp, 1);
423 word++;
424 continue;
425 }
426 ssz = strcspn(word, "\\ ");
427 } else
428 ssz = strcspn(word, "\\");
429 encode(p, word, ssz);
430 word += (int)ssz;
431 continue;
432 }
433
434 word++;
435 esc = mandoc_escape(&word, &seq, &sz);
436 if (ESCAPE_ERROR == esc)
437 continue;
438
439 switch (esc) {
440 case ESCAPE_UNICODE:
441 uc = mchars_num2uc(seq + 1, sz - 1);
442 break;
443 case ESCAPE_NUMBERED:
444 uc = mchars_num2char(seq, sz);
445 if (uc < 0)
446 continue;
447 break;
448 case ESCAPE_SPECIAL:
449 if (p->enc == TERMENC_ASCII) {
450 cp = mchars_spec2str(seq, sz, &ssz);
451 if (cp != NULL)
452 encode(p, cp, ssz);
453 } else {
454 uc = mchars_spec2cp(seq, sz);
455 if (uc > 0)
456 encode1(p, uc);
457 }
458 continue;
459 case ESCAPE_FONTBOLD:
460 term_fontrepl(p, TERMFONT_BOLD);
461 continue;
462 case ESCAPE_FONTITALIC:
463 term_fontrepl(p, TERMFONT_UNDER);
464 continue;
465 case ESCAPE_FONTBI:
466 term_fontrepl(p, TERMFONT_BI);
467 continue;
468 case ESCAPE_FONT:
469 case ESCAPE_FONTROMAN:
470 term_fontrepl(p, TERMFONT_NONE);
471 continue;
472 case ESCAPE_FONTPREV:
473 term_fontlast(p);
474 continue;
475 case ESCAPE_NOSPACE:
476 if (p->flags & TERMP_BACKAFTER)
477 p->flags &= ~TERMP_BACKAFTER;
478 else if (*word == '\0')
479 p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
480 continue;
481 case ESCAPE_HORIZ:
482 if (a2roffsu(seq, &su, SCALE_EM) == NULL)
483 continue;
484 uc = term_hspan(p, &su) / 24;
485 if (uc > 0)
486 while (uc-- > 0)
487 bufferc(p, ASCII_NBRSP);
488 else if (p->col > (size_t)(-uc))
489 p->col += uc;
490 else {
491 uc += p->col;
492 p->col = 0;
493 if (p->tcol->offset > (size_t)(-uc)) {
494 p->ti += uc;
495 p->tcol->offset += uc;
496 } else {
497 p->ti -= p->tcol->offset;
498 p->tcol->offset = 0;
499 }
500 }
501 continue;
502 case ESCAPE_HLINE:
503 if ((seq = a2roffsu(seq, &su, SCALE_EM)) == NULL)
504 continue;
505 uc = term_hspan(p, &su) / 24;
506 if (uc <= 0) {
507 if (p->tcol->rmargin <= p->tcol->offset)
508 continue;
509 lsz = p->tcol->rmargin - p->tcol->offset;
510 } else
511 lsz = uc;
512 if (*seq == '\0')
513 uc = -1;
514 else if (*seq == '\\') {
515 seq++;
516 esc = mandoc_escape(&seq, &cp, &sz);
517 switch (esc) {
518 case ESCAPE_UNICODE:
519 uc = mchars_num2uc(cp + 1, sz - 1);
520 break;
521 case ESCAPE_NUMBERED:
522 uc = mchars_num2char(cp, sz);
523 break;
524 case ESCAPE_SPECIAL:
525 uc = mchars_spec2cp(cp, sz);
526 break;
527 default:
528 uc = -1;
529 break;
530 }
531 } else
532 uc = *seq;
533 if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
534 uc = '_';
535 if (p->enc == TERMENC_ASCII) {
536 cp = ascii_uc2str(uc);
537 csz = term_strlen(p, cp);
538 ssz = strlen(cp);
539 } else
540 csz = (*p->width)(p, uc);
541 while (lsz >= csz) {
542 if (p->enc == TERMENC_ASCII)
543 encode(p, cp, ssz);
544 else
545 encode1(p, uc);
546 lsz -= csz;
547 }
548 continue;
549 case ESCAPE_SKIPCHAR:
550 p->flags |= TERMP_BACKAFTER;
551 continue;
552 case ESCAPE_OVERSTRIKE:
553 cp = seq + sz;
554 while (seq < cp) {
555 if (*seq == '\\') {
556 mandoc_escape(&seq, NULL, NULL);
557 continue;
558 }
559 encode1(p, *seq++);
560 if (seq < cp) {
561 if (p->flags & TERMP_BACKBEFORE)
562 p->flags |= TERMP_BACKAFTER;
563 else
564 p->flags |= TERMP_BACKBEFORE;
565 }
566 }
567 /* Trim trailing backspace/blank pair. */
568 if (p->lastcol > 2 &&
569 (p->tcol->buf[p->lastcol - 1] == ' ' ||
570 p->tcol->buf[p->lastcol - 1] == '\t'))
571 p->lastcol -= 2;
572 if (p->col > p->lastcol)
573 p->col = p->lastcol;
574 continue;
575 default:
576 continue;
577 }
578
579 /*
580 * Common handling for Unicode and numbered
581 * character escape sequences.
582 */
583
584 if (p->enc == TERMENC_ASCII) {
585 cp = ascii_uc2str(uc);
586 encode(p, cp, strlen(cp));
587 } else {
588 if ((uc < 0x20 && uc != 0x09) ||
589 (uc > 0x7E && uc < 0xA0))
590 uc = 0xFFFD;
591 encode1(p, uc);
592 }
593 }
594 p->flags &= ~TERMP_NBRWORD;
595 }
596
597 static void
598 adjbuf(struct termp_col *c, size_t sz)
599 {
600 if (c->maxcols == 0)
601 c->maxcols = 1024;
602 while (c->maxcols <= sz)
603 c->maxcols <<= 2;
604 c->buf = mandoc_reallocarray(c->buf, c->maxcols, sizeof(*c->buf));
605 }
606
607 static void
608 bufferc(struct termp *p, char c)
609 {
610 if (p->flags & TERMP_NOBUF) {
611 (*p->letter)(p, c);
612 return;
613 }
614 if (p->col + 1 >= p->tcol->maxcols)
615 adjbuf(p->tcol, p->col + 1);
616 if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
617 p->tcol->buf[p->col] = c;
618 if (p->lastcol < ++p->col)
619 p->lastcol = p->col;
620 }
621
622 /*
623 * See encode().
624 * Do this for a single (probably unicode) value.
625 * Does not check for non-decorated glyphs.
626 */
627 static void
628 encode1(struct termp *p, int c)
629 {
630 enum termfont f;
631
632 if (p->flags & TERMP_NOBUF) {
633 (*p->letter)(p, c);
634 return;
635 }
636
637 if (p->col + 7 >= p->tcol->maxcols)
638 adjbuf(p->tcol, p->col + 7);
639
640 f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
641 p->fontq[p->fonti] : TERMFONT_NONE;
642
643 if (p->flags & TERMP_BACKBEFORE) {
644 if (p->tcol->buf[p->col - 1] == ' ' ||
645 p->tcol->buf[p->col - 1] == '\t')
646 p->col--;
647 else
648 p->tcol->buf[p->col++] = '\b';
649 p->flags &= ~TERMP_BACKBEFORE;
650 }
651 if (f == TERMFONT_UNDER || f == TERMFONT_BI) {
652 p->tcol->buf[p->col++] = '_';
653 p->tcol->buf[p->col++] = '\b';
654 }
655 if (f == TERMFONT_BOLD || f == TERMFONT_BI) {
656 if (c == ASCII_HYPH)
657 p->tcol->buf[p->col++] = '-';
658 else
659 p->tcol->buf[p->col++] = c;
660 p->tcol->buf[p->col++] = '\b';
661 }
662 if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
663 p->tcol->buf[p->col] = c;
664 if (p->lastcol < ++p->col)
665 p->lastcol = p->col;
666 if (p->flags & TERMP_BACKAFTER) {
667 p->flags |= TERMP_BACKBEFORE;
668 p->flags &= ~TERMP_BACKAFTER;
669 }
670 }
671
672 static void
673 encode(struct termp *p, const char *word, size_t sz)
674 {
675 size_t i;
676
677 if (p->flags & TERMP_NOBUF) {
678 for (i = 0; i < sz; i++)
679 (*p->letter)(p, word[i]);
680 return;
681 }
682
683 if (p->col + 2 + (sz * 5) >= p->tcol->maxcols)
684 adjbuf(p->tcol, p->col + 2 + (sz * 5));
685
686 for (i = 0; i < sz; i++) {
687 if (ASCII_HYPH == word[i] ||
688 isgraph((unsigned char)word[i]))
689 encode1(p, word[i]);
690 else {
691 if (p->lastcol <= p->col ||
692 (word[i] != ' ' && word[i] != ASCII_NBRSP))
693 p->tcol->buf[p->col] = word[i];
694 p->col++;
695
696 /*
697 * Postpone the effect of \z while handling
698 * an overstrike sequence from ascii_uc2str().
699 */
700
701 if (word[i] == '\b' &&
702 (p->flags & TERMP_BACKBEFORE)) {
703 p->flags &= ~TERMP_BACKBEFORE;
704 p->flags |= TERMP_BACKAFTER;
705 }
706 }
707 }
708 if (p->lastcol < p->col)
709 p->lastcol = p->col;
710 }
711
712 void
713 term_setwidth(struct termp *p, const char *wstr)
714 {
715 struct roffsu su;
716 int iop, width;
717
718 iop = 0;
719 width = 0;
720 if (NULL != wstr) {
721 switch (*wstr) {
722 case '+':
723 iop = 1;
724 wstr++;
725 break;
726 case '-':
727 iop = -1;
728 wstr++;
729 break;
730 default:
731 break;
732 }
733 if (a2roffsu(wstr, &su, SCALE_MAX) != NULL)
734 width = term_hspan(p, &su);
735 else
736 iop = 0;
737 }
738 (*p->setwidth)(p, iop, width);
739 }
740
741 size_t
742 term_len(const struct termp *p, size_t sz)
743 {
744
745 return (*p->width)(p, ' ') * sz;
746 }
747
748 static size_t
749 cond_width(const struct termp *p, int c, int *skip)
750 {
751
752 if (*skip) {
753 (*skip) = 0;
754 return 0;
755 } else
756 return (*p->width)(p, c);
757 }
758
759 size_t
760 term_strlen(const struct termp *p, const char *cp)
761 {
762 size_t sz, rsz, i;
763 int ssz, skip, uc;
764 const char *seq, *rhs;
765 enum mandoc_esc esc;
766 static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
767 ASCII_BREAK, '\0' };
768
769 /*
770 * Account for escaped sequences within string length
771 * calculations. This follows the logic in term_word() as we
772 * must calculate the width of produced strings.
773 */
774
775 sz = 0;
776 skip = 0;
777 while ('\0' != *cp) {
778 rsz = strcspn(cp, rej);
779 for (i = 0; i < rsz; i++)
780 sz += cond_width(p, *cp++, &skip);
781
782 switch (*cp) {
783 case '\\':
784 cp++;
785 esc = mandoc_escape(&cp, &seq, &ssz);
786 if (ESCAPE_ERROR == esc)
787 continue;
788
789 rhs = NULL;
790
791 switch (esc) {
792 case ESCAPE_UNICODE:
793 uc = mchars_num2uc(seq + 1, ssz - 1);
794 break;
795 case ESCAPE_NUMBERED:
796 uc = mchars_num2char(seq, ssz);
797 if (uc < 0)
798 continue;
799 break;
800 case ESCAPE_SPECIAL:
801 if (p->enc == TERMENC_ASCII) {
802 rhs = mchars_spec2str(seq, ssz, &rsz);
803 if (rhs != NULL)
804 break;
805 } else {
806 uc = mchars_spec2cp(seq, ssz);
807 if (uc > 0)
808 sz += cond_width(p, uc, &skip);
809 }
810 continue;
811 case ESCAPE_SKIPCHAR:
812 skip = 1;
813 continue;
814 case ESCAPE_OVERSTRIKE:
815 rsz = 0;
816 rhs = seq + ssz;
817 while (seq < rhs) {
818 if (*seq == '\\') {
819 mandoc_escape(&seq, NULL, NULL);
820 continue;
821 }
822 i = (*p->width)(p, *seq++);
823 if (rsz < i)
824 rsz = i;
825 }
826 sz += rsz;
827 continue;
828 default:
829 continue;
830 }
831
832 /*
833 * Common handling for Unicode and numbered
834 * character escape sequences.
835 */
836
837 if (rhs == NULL) {
838 if (p->enc == TERMENC_ASCII) {
839 rhs = ascii_uc2str(uc);
840 rsz = strlen(rhs);
841 } else {
842 if ((uc < 0x20 && uc != 0x09) ||
843 (uc > 0x7E && uc < 0xA0))
844 uc = 0xFFFD;
845 sz += cond_width(p, uc, &skip);
846 continue;
847 }
848 }
849
850 if (skip) {
851 skip = 0;
852 break;
853 }
854
855 /*
856 * Common handling for all escape sequences
857 * printing more than one character.
858 */
859
860 for (i = 0; i < rsz; i++)
861 sz += (*p->width)(p, *rhs++);
862 break;
863 case ASCII_NBRSP:
864 sz += cond_width(p, ' ', &skip);
865 cp++;
866 break;
867 case ASCII_HYPH:
868 sz += cond_width(p, '-', &skip);
869 cp++;
870 break;
871 default:
872 break;
873 }
874 }
875
876 return sz;
877 }
878
879 int
880 term_vspan(const struct termp *p, const struct roffsu *su)
881 {
882 double r;
883 int ri;
884
885 switch (su->unit) {
886 case SCALE_BU:
887 r = su->scale / 40.0;
888 break;
889 case SCALE_CM:
890 r = su->scale * 6.0 / 2.54;
891 break;
892 case SCALE_FS:
893 r = su->scale * 65536.0 / 40.0;
894 break;
895 case SCALE_IN:
896 r = su->scale * 6.0;
897 break;
898 case SCALE_MM:
899 r = su->scale * 0.006;
900 break;
901 case SCALE_PC:
902 r = su->scale;
903 break;
904 case SCALE_PT:
905 r = su->scale / 12.0;
906 break;
907 case SCALE_EN:
908 case SCALE_EM:
909 r = su->scale * 0.6;
910 break;
911 case SCALE_VS:
912 r = su->scale;
913 break;
914 default:
915 abort();
916 }
917 ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
918 return ri < 66 ? ri : 1;
919 }
920
921 /*
922 * Convert a scaling width to basic units, rounding down.
923 */
924 int
925 term_hspan(const struct termp *p, const struct roffsu *su)
926 {
927
928 return (*p->hspan)(p, su);
929 }