]> git.cameronkatri.com Git - mandoc.git/blob - term.c
Change chars.in HTML encoding to be a Unicode codepoint (int), which is
[mandoc.git] / term.c
1 /* $Id: term.c,v 1.161 2010/07/16 22:33:30 kristaps Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010 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 "chars.h"
33 #include "out.h"
34 #include "term.h"
35 #include "main.h"
36
37 static void spec(struct termp *, const char *, size_t);
38 static void res(struct termp *, const char *, size_t);
39 static void buffera(struct termp *, const char *, size_t);
40 static void bufferc(struct termp *, char);
41 static void adjbuf(struct termp *p, size_t);
42 static void encode(struct termp *, const char *, size_t);
43
44
45 void
46 term_free(struct termp *p)
47 {
48
49 if (p->buf)
50 free(p->buf);
51 if (p->symtab)
52 chars_free(p->symtab);
53
54 free(p);
55 }
56
57
58 void
59 term_begin(struct termp *p, term_margin head,
60 term_margin foot, const void *arg)
61 {
62
63 p->headf = head;
64 p->footf = foot;
65 p->argf = arg;
66 (*p->begin)(p);
67 }
68
69
70 void
71 term_end(struct termp *p)
72 {
73
74 (*p->end)(p);
75 }
76
77
78 struct termp *
79 term_alloc(enum termenc enc)
80 {
81 struct termp *p;
82
83 p = calloc(1, sizeof(struct termp));
84 if (NULL == p) {
85 perror(NULL);
86 exit(EXIT_FAILURE);
87 }
88
89 p->enc = enc;
90 return(p);
91 }
92
93
94 /*
95 * Flush a line of text. A "line" is loosely defined as being something
96 * that should be followed by a newline, regardless of whether it's
97 * broken apart by newlines getting there. A line can also be a
98 * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
99 * not have a trailing newline.
100 *
101 * The following flags may be specified:
102 *
103 * - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
104 * offset value. This is useful when doing columnar lists where the
105 * prior column has right-padded.
106 *
107 * - TERMP_NOBREAK: this is the most important and is used when making
108 * columns. In short: don't print a newline and instead pad to the
109 * right margin. Used in conjunction with TERMP_NOLPAD.
110 *
111 * - TERMP_TWOSPACE: when padding, make sure there are at least two
112 * space characters of padding. Otherwise, rather break the line.
113 *
114 * - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
115 * the line is overrun, and don't pad-right if it's underrun.
116 *
117 * - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
118 * overruning, instead save the position and continue at that point
119 * when the next invocation.
120 *
121 * In-line line breaking:
122 *
123 * If TERMP_NOBREAK is specified and the line overruns the right
124 * margin, it will break and pad-right to the right margin after
125 * writing. If maxrmargin is violated, it will break and continue
126 * writing from the right-margin, which will lead to the above scenario
127 * upon exit. Otherwise, the line will break at the right margin.
128 */
129 void
130 term_flushln(struct termp *p)
131 {
132 int i; /* current input position in p->buf */
133 size_t vis; /* current visual position on output */
134 size_t vbl; /* number of blanks to prepend to output */
135 size_t vend; /* end of word visual position on output */
136 size_t bp; /* visual right border position */
137 int j; /* temporary loop index for p->buf */
138 int jhy; /* last hyph before overflow w/r/t j */
139 size_t maxvis; /* output position of visible boundary */
140 size_t mmax; /* used in calculating bp */
141
142 /*
143 * First, establish the maximum columns of "visible" content.
144 * This is usually the difference between the right-margin and
145 * an indentation, but can be, for tagged lists or columns, a
146 * small set of values.
147 */
148
149 assert(p->offset < p->rmargin);
150
151 maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
152 /* LINTED */
153 0 : p->rmargin - p->offset - p->overstep;
154 mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
155 /* LINTED */
156 0 : p->maxrmargin - p->offset - p->overstep;
157
158 bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
159
160 /*
161 * Indent the first line of a paragraph.
162 */
163 vbl = p->flags & TERMP_NOLPAD ? 0 : p->offset;
164
165 vis = vend = i = 0;
166
167 while (i < (int)p->col) {
168 /*
169 * Handle literal tab characters: collapse all
170 * subsequent tabs into a single huge set of spaces.
171 */
172 for (j = i; j < (int)p->col; j++) {
173 if ('\t' != p->buf[j])
174 break;
175 vend = (vis / p->tabwidth + 1) * p->tabwidth;
176 vbl += vend - vis;
177 vis = vend;
178 }
179
180 /*
181 * Count up visible word characters. Control sequences
182 * (starting with the CSI) aren't counted. A space
183 * generates a non-printing word, which is valid (the
184 * space is printed according to regular spacing rules).
185 */
186
187 /* LINTED */
188 for (jhy = 0; j < (int)p->col; j++) {
189 if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
190 break;
191
192 /* Back over the the last printed character. */
193 if (8 == p->buf[j]) {
194 assert(j);
195 vend -= (*p->width)(p, p->buf[j - 1]);
196 continue;
197 }
198
199 /* Regular word. */
200 /* Break at the hyphen point if we overrun. */
201 if (vend > vis && vend < bp &&
202 ASCII_HYPH == p->buf[j])
203 jhy = j;
204
205 vend += (*p->width)(p, p->buf[j]);
206 }
207
208 /*
209 * Find out whether we would exceed the right margin.
210 * If so, break to the next line.
211 */
212 if (vend > bp && 0 == jhy && vis > 0) {
213 vend -= vis;
214 (*p->endline)(p);
215 if (TERMP_NOBREAK & p->flags) {
216 p->viscol = p->rmargin;
217 (*p->advance)(p, p->rmargin);
218 vend += p->rmargin - p->offset;
219 } else {
220 p->viscol = 0;
221 vbl = p->offset;
222 }
223
224 /* Remove the p->overstep width. */
225
226 bp += (int)/* LINTED */
227 p->overstep;
228 p->overstep = 0;
229 }
230
231 /*
232 * Skip leading tabs, they were handled above.
233 */
234 while (i < (int)p->col && '\t' == p->buf[i])
235 i++;
236
237 /* Write out the [remaining] word. */
238 for ( ; i < (int)p->col; i++) {
239 if (vend > bp && jhy > 0 && i > jhy)
240 break;
241 if ('\t' == p->buf[i])
242 break;
243 if (' ' == p->buf[i]) {
244 while (' ' == p->buf[i]) {
245 vbl += (*p->width)(p, p->buf[i]);
246 i++;
247 }
248 break;
249 }
250 if (ASCII_NBRSP == p->buf[i]) {
251 vbl += (*p->width)(p, ' ');
252 continue;
253 }
254
255 /*
256 * Now we definitely know there will be
257 * printable characters to output,
258 * so write preceding white space now.
259 */
260 if (vbl) {
261 (*p->advance)(p, vbl);
262 p->viscol += vbl;
263 vbl = 0;
264 }
265
266 if (ASCII_HYPH == p->buf[i]) {
267 (*p->letter)(p, '-');
268 p->viscol += (*p->width)(p, '-');
269 } else {
270 (*p->letter)(p, p->buf[i]);
271 p->viscol += (*p->width)(p, p->buf[i]);
272 }
273 }
274 vend += vbl;
275 vis = vend;
276 }
277
278 p->col = 0;
279 p->overstep = 0;
280
281 if ( ! (TERMP_NOBREAK & p->flags)) {
282 p->viscol = 0;
283 (*p->endline)(p);
284 return;
285 }
286
287 if (TERMP_HANG & p->flags) {
288 /* We need one blank after the tag. */
289 p->overstep = /* LINTED */
290 vis - maxvis + (*p->width)(p, ' ');
291
292 /*
293 * Behave exactly the same way as groff:
294 * If we have overstepped the margin, temporarily move
295 * it to the right and flag the rest of the line to be
296 * shorter.
297 * If we landed right at the margin, be happy.
298 * If we are one step before the margin, temporarily
299 * move it one step LEFT and flag the rest of the line
300 * to be longer.
301 */
302 if (p->overstep >= -1) {
303 assert((int)maxvis + p->overstep >= 0);
304 /* LINTED */
305 maxvis += p->overstep;
306 } else
307 p->overstep = 0;
308
309 } else if (TERMP_DANGLE & p->flags)
310 return;
311
312 /* Right-pad. */
313 if (maxvis > vis + /* LINTED */
314 ((TERMP_TWOSPACE & p->flags) ?
315 (*p->width)(p, ' ') : 0)) {
316 p->viscol += maxvis - vis;
317 (*p->advance)(p, maxvis - vis);
318 vis += (maxvis - vis);
319 } else { /* ...or newline break. */
320 (*p->endline)(p);
321 p->viscol = p->rmargin;
322 (*p->advance)(p, p->rmargin);
323 }
324 }
325
326
327 /*
328 * A newline only breaks an existing line; it won't assert vertical
329 * space. All data in the output buffer is flushed prior to the newline
330 * assertion.
331 */
332 void
333 term_newln(struct termp *p)
334 {
335
336 p->flags |= TERMP_NOSPACE;
337 if (0 == p->col && 0 == p->viscol) {
338 p->flags &= ~TERMP_NOLPAD;
339 return;
340 }
341 term_flushln(p);
342 p->flags &= ~TERMP_NOLPAD;
343 }
344
345
346 /*
347 * Asserts a vertical space (a full, empty line-break between lines).
348 * Note that if used twice, this will cause two blank spaces and so on.
349 * All data in the output buffer is flushed prior to the newline
350 * assertion.
351 */
352 void
353 term_vspace(struct termp *p)
354 {
355
356 term_newln(p);
357 p->viscol = 0;
358 (*p->endline)(p);
359 }
360
361
362 static void
363 spec(struct termp *p, const char *word, size_t len)
364 {
365 const char *rhs;
366 size_t sz;
367
368 rhs = chars_spec2str(p->symtab, word, len, &sz);
369 if (rhs)
370 encode(p, rhs, sz);
371 }
372
373
374 static void
375 res(struct termp *p, const char *word, size_t len)
376 {
377 const char *rhs;
378 size_t sz;
379
380 rhs = chars_res2str(p->symtab, word, len, &sz);
381 if (rhs)
382 encode(p, rhs, sz);
383 }
384
385
386 void
387 term_fontlast(struct termp *p)
388 {
389 enum termfont f;
390
391 f = p->fontl;
392 p->fontl = p->fontq[p->fonti];
393 p->fontq[p->fonti] = f;
394 }
395
396
397 void
398 term_fontrepl(struct termp *p, enum termfont f)
399 {
400
401 p->fontl = p->fontq[p->fonti];
402 p->fontq[p->fonti] = f;
403 }
404
405
406 void
407 term_fontpush(struct termp *p, enum termfont f)
408 {
409
410 assert(p->fonti + 1 < 10);
411 p->fontl = p->fontq[p->fonti];
412 p->fontq[++p->fonti] = f;
413 }
414
415
416 const void *
417 term_fontq(struct termp *p)
418 {
419
420 return(&p->fontq[p->fonti]);
421 }
422
423
424 enum termfont
425 term_fonttop(struct termp *p)
426 {
427
428 return(p->fontq[p->fonti]);
429 }
430
431
432 void
433 term_fontpopq(struct termp *p, const void *key)
434 {
435
436 while (p->fonti >= 0 && key != &p->fontq[p->fonti])
437 p->fonti--;
438 assert(p->fonti >= 0);
439 }
440
441
442 void
443 term_fontpop(struct termp *p)
444 {
445
446 assert(p->fonti);
447 p->fonti--;
448 }
449
450
451 /*
452 * Handle pwords, partial words, which may be either a single word or a
453 * phrase that cannot be broken down (such as a literal string). This
454 * handles word styling.
455 */
456 void
457 term_word(struct termp *p, const char *word)
458 {
459 const char *sv, *seq;
460 int sz;
461 size_t ssz;
462 enum roffdeco deco;
463
464 sv = word;
465
466 if (word[0] && '\0' == word[1])
467 switch (word[0]) {
468 case('.'):
469 /* FALLTHROUGH */
470 case(','):
471 /* FALLTHROUGH */
472 case(';'):
473 /* FALLTHROUGH */
474 case(':'):
475 /* FALLTHROUGH */
476 case('?'):
477 /* FALLTHROUGH */
478 case('!'):
479 /* FALLTHROUGH */
480 case(')'):
481 /* FALLTHROUGH */
482 case(']'):
483 if ( ! (TERMP_IGNDELIM & p->flags))
484 p->flags |= TERMP_NOSPACE;
485 break;
486 default:
487 break;
488 }
489
490 if ( ! (TERMP_NOSPACE & p->flags)) {
491 if ( ! (TERMP_KEEP & p->flags)) {
492 if (TERMP_PREKEEP & p->flags)
493 p->flags |= TERMP_KEEP;
494 bufferc(p, ' ');
495 if (TERMP_SENTENCE & p->flags)
496 bufferc(p, ' ');
497 } else
498 bufferc(p, ASCII_NBRSP);
499 }
500
501 if ( ! (p->flags & TERMP_NONOSPACE))
502 p->flags &= ~TERMP_NOSPACE;
503
504 p->flags &= ~TERMP_SENTENCE;
505
506 /* FIXME: use strcspn. */
507
508 while (*word) {
509 if ('\\' != *word) {
510 encode(p, word, 1);
511 word++;
512 continue;
513 }
514
515 seq = ++word;
516 sz = a2roffdeco(&deco, &seq, &ssz);
517
518 switch (deco) {
519 case (DECO_RESERVED):
520 res(p, seq, ssz);
521 break;
522 case (DECO_SPECIAL):
523 spec(p, seq, ssz);
524 break;
525 case (DECO_BOLD):
526 term_fontrepl(p, TERMFONT_BOLD);
527 break;
528 case (DECO_ITALIC):
529 term_fontrepl(p, TERMFONT_UNDER);
530 break;
531 case (DECO_ROMAN):
532 term_fontrepl(p, TERMFONT_NONE);
533 break;
534 case (DECO_PREVIOUS):
535 term_fontlast(p);
536 break;
537 default:
538 break;
539 }
540
541 word += sz;
542 if (DECO_NOSPACE == deco && '\0' == *word)
543 p->flags |= TERMP_NOSPACE;
544 }
545
546 /*
547 * Note that we don't process the pipe: the parser sees it as
548 * punctuation, but we don't in terms of typography.
549 */
550 if (sv[0] && 0 == sv[1])
551 switch (sv[0]) {
552 case('('):
553 /* FALLTHROUGH */
554 case('['):
555 p->flags |= TERMP_NOSPACE;
556 break;
557 default:
558 break;
559 }
560 }
561
562
563 static void
564 adjbuf(struct termp *p, size_t sz)
565 {
566
567 if (0 == p->maxcols)
568 p->maxcols = 1024;
569 while (sz >= p->maxcols)
570 p->maxcols <<= 2;
571
572 p->buf = realloc(p->buf, p->maxcols);
573 if (NULL == p->buf) {
574 perror(NULL);
575 exit(EXIT_FAILURE);
576 }
577 }
578
579
580 static void
581 buffera(struct termp *p, const char *word, size_t sz)
582 {
583
584 if (p->col + sz >= p->maxcols)
585 adjbuf(p, p->col + sz);
586
587 memcpy(&p->buf[(int)p->col], word, sz);
588 p->col += sz;
589 }
590
591
592 static void
593 bufferc(struct termp *p, char c)
594 {
595
596 if (p->col + 1 >= p->maxcols)
597 adjbuf(p, p->col + 1);
598
599 p->buf[(int)p->col++] = c;
600 }
601
602
603 static void
604 encode(struct termp *p, const char *word, size_t sz)
605 {
606 enum termfont f;
607 int i;
608
609 /*
610 * Encode and buffer a string of characters. If the current
611 * font mode is unset, buffer directly, else encode then buffer
612 * character by character.
613 */
614
615 if (TERMFONT_NONE == (f = term_fonttop(p))) {
616 buffera(p, word, sz);
617 return;
618 }
619
620 for (i = 0; i < (int)sz; i++) {
621 if ( ! isgraph((u_char)word[i])) {
622 bufferc(p, word[i]);
623 continue;
624 }
625
626 if (TERMFONT_UNDER == f)
627 bufferc(p, '_');
628 else
629 bufferc(p, word[i]);
630
631 bufferc(p, 8);
632 bufferc(p, word[i]);
633 }
634 }
635
636
637 size_t
638 term_len(const struct termp *p, size_t sz)
639 {
640
641 return((*p->width)(p, ' ') * sz);
642 }
643
644
645 size_t
646 term_strlen(const struct termp *p, const char *cp)
647 {
648 size_t sz;
649
650 for (sz = 0; *cp; cp++)
651 sz += (*p->width)(p, *cp);
652
653 return(sz);
654 }
655
656
657 /* ARGSUSED */
658 size_t
659 term_vspan(const struct termp *p, const struct roffsu *su)
660 {
661 double r;
662
663 switch (su->unit) {
664 case (SCALE_CM):
665 r = su->scale * 2;
666 break;
667 case (SCALE_IN):
668 r = su->scale * 6;
669 break;
670 case (SCALE_PC):
671 r = su->scale;
672 break;
673 case (SCALE_PT):
674 r = su->scale / 8;
675 break;
676 case (SCALE_MM):
677 r = su->scale / 1000;
678 break;
679 case (SCALE_VS):
680 r = su->scale;
681 break;
682 default:
683 r = su->scale - 1;
684 break;
685 }
686
687 if (r < 0.0)
688 r = 0.0;
689 return(/* LINTED */(size_t)
690 r);
691 }
692
693
694 size_t
695 term_hspan(const struct termp *p, const struct roffsu *su)
696 {
697 double v;
698
699 v = ((*p->hspan)(p, su));
700 if (v < 0.0)
701 v = 0.0;
702 return((size_t) /* LINTED */
703 v);
704 }