]> git.cameronkatri.com Git - mandoc.git/blob - term.c
Basic implementation of -Tman .Fo and .Fa;
[mandoc.git] / term.c
1 /* $Id: term.c,v 1.203 2012/05/31 22:29:13 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2011, 2012 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, int);
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 *
87 * - TERMP_TWOSPACE: make sure there is room for at least two space
88 * characters of padding. Otherwise, rather break the line.
89 *
90 * - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
91 * the line is overrun, and don't pad-right if it's underrun.
92 *
93 * - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
94 * overrunning, instead save the position and continue at that point
95 * when the next invocation.
96 *
97 * In-line line breaking:
98 *
99 * If TERMP_NOBREAK is specified and the line overruns the right
100 * margin, it will break and pad-right to the right margin after
101 * writing. If maxrmargin is violated, it will break and continue
102 * writing from the right-margin, which will lead to the above scenario
103 * upon exit. Otherwise, the line will break at the right margin.
104 */
105 void
106 term_flushln(struct termp *p)
107 {
108 int i; /* current input position in p->buf */
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 int j; /* temporary loop index for p->buf */
115 int 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 assert (p->rmargin >= p->offset);
126 dv = p->rmargin - p->offset;
127 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
128 dv = p->maxrmargin - p->offset;
129 mmax = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
130
131 bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
132
133 /*
134 * Calculate the required amount of padding.
135 */
136 vbl = p->offset + p->overstep > p->viscol ?
137 p->offset + p->overstep - p->viscol : 0;
138
139 vis = vend = 0;
140 i = 0;
141
142 while (i < p->col) {
143 /*
144 * Handle literal tab characters: collapse all
145 * subsequent tabs into a single huge set of spaces.
146 */
147 while (i < p->col && '\t' == p->buf[i]) {
148 vend = (vis / p->tabwidth + 1) * p->tabwidth;
149 vbl += vend - vis;
150 vis = vend;
151 i++;
152 }
153
154 /*
155 * Count up visible word characters. Control sequences
156 * (starting with the CSI) aren't counted. A space
157 * generates a non-printing word, which is valid (the
158 * space is printed according to regular spacing rules).
159 */
160
161 for (j = i, jhy = 0; j < p->col; j++) {
162 if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
163 break;
164
165 /* Back over the the last printed character. */
166 if (8 == p->buf[j]) {
167 assert(j);
168 vend -= (*p->width)(p, p->buf[j - 1]);
169 continue;
170 }
171
172 /* Regular word. */
173 /* Break at the hyphen point if we overrun. */
174 if (vend > vis && vend < bp &&
175 ASCII_HYPH == p->buf[j])
176 jhy = j;
177
178 vend += (*p->width)(p, p->buf[j]);
179 }
180
181 /*
182 * Find out whether we would exceed the right margin.
183 * If so, break to the next line.
184 */
185 if (vend > bp && 0 == jhy && vis > 0) {
186 vend -= vis;
187 (*p->endline)(p);
188 p->viscol = 0;
189 if (TERMP_NOBREAK & p->flags) {
190 vbl = p->rmargin;
191 vend += p->rmargin - p->offset;
192 } else
193 vbl = p->offset;
194
195 /* Remove the p->overstep width. */
196
197 bp += (size_t)p->overstep;
198 p->overstep = 0;
199 }
200
201 /* Write out the [remaining] word. */
202 for ( ; i < p->col; i++) {
203 if (vend > bp && jhy > 0 && i > jhy)
204 break;
205 if ('\t' == p->buf[i])
206 break;
207 if (' ' == p->buf[i]) {
208 j = i;
209 while (' ' == p->buf[i])
210 i++;
211 dv = (size_t)(i - j) * (*p->width)(p, ' ');
212 vbl += dv;
213 vend += dv;
214 break;
215 }
216 if (ASCII_NBRSP == p->buf[i]) {
217 vbl += (*p->width)(p, ' ');
218 continue;
219 }
220
221 /*
222 * Now we definitely know there will be
223 * printable characters to output,
224 * so write preceding white space now.
225 */
226 if (vbl) {
227 (*p->advance)(p, vbl);
228 p->viscol += vbl;
229 vbl = 0;
230 }
231
232 if (ASCII_HYPH == p->buf[i]) {
233 (*p->letter)(p, '-');
234 p->viscol += (*p->width)(p, '-');
235 continue;
236 }
237
238 (*p->letter)(p, p->buf[i]);
239 if (8 == p->buf[i])
240 p->viscol -= (*p->width)(p, p->buf[i-1]);
241 else
242 p->viscol += (*p->width)(p, p->buf[i]);
243 }
244 vis = vend;
245 }
246
247 /*
248 * If there was trailing white space, it was not printed;
249 * so reset the cursor position accordingly.
250 */
251 if (vis)
252 vis -= vbl;
253
254 p->col = 0;
255 p->overstep = 0;
256
257 if ( ! (TERMP_NOBREAK & p->flags)) {
258 p->viscol = 0;
259 (*p->endline)(p);
260 return;
261 }
262
263 if (TERMP_HANG & p->flags) {
264 /* We need one blank after the tag. */
265 p->overstep = (int)(vis - maxvis + (*p->width)(p, ' '));
266
267 /*
268 * Behave exactly the same way as groff:
269 * If we have overstepped the margin, temporarily move
270 * it to the right and flag the rest of the line to be
271 * shorter.
272 * If we landed right at the margin, be happy.
273 * If we are one step before the margin, temporarily
274 * move it one step LEFT and flag the rest of the line
275 * to be longer.
276 */
277 if (p->overstep < -1)
278 p->overstep = 0;
279 return;
280
281 } else if (TERMP_DANGLE & p->flags)
282 return;
283
284 /* If the column was overrun, break the line. */
285 if (maxvis <= vis +
286 ((TERMP_TWOSPACE & p->flags) ? (*p->width)(p, ' ') : 0)) {
287 (*p->endline)(p);
288 p->viscol = 0;
289 }
290 }
291
292
293 /*
294 * A newline only breaks an existing line; it won't assert vertical
295 * space. All data in the output buffer is flushed prior to the newline
296 * assertion.
297 */
298 void
299 term_newln(struct termp *p)
300 {
301
302 p->flags |= TERMP_NOSPACE;
303 if (p->col || p->viscol)
304 term_flushln(p);
305 }
306
307
308 /*
309 * Asserts a vertical space (a full, empty line-break between lines).
310 * Note that if used twice, this will cause two blank spaces and so on.
311 * All data in the output buffer is flushed prior to the newline
312 * assertion.
313 */
314 void
315 term_vspace(struct termp *p)
316 {
317
318 term_newln(p);
319 p->viscol = 0;
320 if (0 < p->skipvsp)
321 p->skipvsp--;
322 else
323 (*p->endline)(p);
324 }
325
326 void
327 term_fontlast(struct termp *p)
328 {
329 enum termfont f;
330
331 f = p->fontl;
332 p->fontl = p->fontq[p->fonti];
333 p->fontq[p->fonti] = f;
334 }
335
336
337 void
338 term_fontrepl(struct termp *p, enum termfont f)
339 {
340
341 p->fontl = p->fontq[p->fonti];
342 p->fontq[p->fonti] = f;
343 }
344
345
346 void
347 term_fontpush(struct termp *p, enum termfont f)
348 {
349
350 assert(p->fonti + 1 < 10);
351 p->fontl = p->fontq[p->fonti];
352 p->fontq[++p->fonti] = f;
353 }
354
355
356 const void *
357 term_fontq(struct termp *p)
358 {
359
360 return(&p->fontq[p->fonti]);
361 }
362
363
364 enum termfont
365 term_fonttop(struct termp *p)
366 {
367
368 return(p->fontq[p->fonti]);
369 }
370
371
372 void
373 term_fontpopq(struct termp *p, const void *key)
374 {
375
376 while (p->fonti >= 0 && key != &p->fontq[p->fonti])
377 p->fonti--;
378 assert(p->fonti >= 0);
379 }
380
381
382 void
383 term_fontpop(struct termp *p)
384 {
385
386 assert(p->fonti);
387 p->fonti--;
388 }
389
390 /*
391 * Handle pwords, partial words, which may be either a single word or a
392 * phrase that cannot be broken down (such as a literal string). This
393 * handles word styling.
394 */
395 void
396 term_word(struct termp *p, const char *word)
397 {
398 const char *seq, *cp;
399 char c;
400 int sz, uc;
401 size_t ssz;
402 enum mandoc_esc esc;
403
404 if ( ! (TERMP_NOSPACE & p->flags)) {
405 if ( ! (TERMP_KEEP & p->flags)) {
406 if (TERMP_PREKEEP & p->flags)
407 p->flags |= TERMP_KEEP;
408 bufferc(p, ' ');
409 if (TERMP_SENTENCE & p->flags)
410 bufferc(p, ' ');
411 } else
412 bufferc(p, ASCII_NBRSP);
413 }
414
415 if ( ! (p->flags & TERMP_NONOSPACE))
416 p->flags &= ~TERMP_NOSPACE;
417 else
418 p->flags |= TERMP_NOSPACE;
419
420 p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
421
422 while ('\0' != *word) {
423 if ('\\' != *word) {
424 if (TERMP_SKIPCHAR & p->flags) {
425 p->flags &= ~TERMP_SKIPCHAR;
426 word++;
427 continue;
428 }
429 ssz = strcspn(word, "\\");
430 encode(p, word, ssz);
431 word += (int)ssz;
432 continue;
433 }
434
435 word++;
436 esc = mandoc_escape(&word, &seq, &sz);
437 if (ESCAPE_ERROR == esc)
438 break;
439
440 if (TERMENC_ASCII != p->enc)
441 switch (esc) {
442 case (ESCAPE_UNICODE):
443 uc = mchars_num2uc(seq + 1, sz - 1);
444 if ('\0' == uc)
445 break;
446 encode1(p, uc);
447 continue;
448 case (ESCAPE_SPECIAL):
449 uc = mchars_spec2cp(p->symtab, seq, sz);
450 if (uc <= 0)
451 break;
452 encode1(p, uc);
453 continue;
454 default:
455 break;
456 }
457
458 switch (esc) {
459 case (ESCAPE_UNICODE):
460 encode1(p, '?');
461 break;
462 case (ESCAPE_NUMBERED):
463 c = mchars_num2char(seq, sz);
464 if ('\0' != c)
465 encode(p, &c, 1);
466 break;
467 case (ESCAPE_SPECIAL):
468 cp = mchars_spec2str(p->symtab, seq, sz, &ssz);
469 if (NULL != cp)
470 encode(p, cp, ssz);
471 else if (1 == ssz)
472 encode(p, seq, sz);
473 break;
474 case (ESCAPE_FONTBOLD):
475 term_fontrepl(p, TERMFONT_BOLD);
476 break;
477 case (ESCAPE_FONTITALIC):
478 term_fontrepl(p, TERMFONT_UNDER);
479 break;
480 case (ESCAPE_FONT):
481 /* FALLTHROUGH */
482 case (ESCAPE_FONTROMAN):
483 term_fontrepl(p, TERMFONT_NONE);
484 break;
485 case (ESCAPE_FONTPREV):
486 term_fontlast(p);
487 break;
488 case (ESCAPE_NOSPACE):
489 if (TERMP_SKIPCHAR & p->flags)
490 p->flags &= ~TERMP_SKIPCHAR;
491 else if ('\0' == *word)
492 p->flags |= TERMP_NOSPACE;
493 break;
494 case (ESCAPE_SKIPCHAR):
495 p->flags |= TERMP_SKIPCHAR;
496 break;
497 default:
498 break;
499 }
500 }
501 }
502
503 static void
504 adjbuf(struct termp *p, int sz)
505 {
506
507 if (0 == p->maxcols)
508 p->maxcols = 1024;
509 while (sz >= p->maxcols)
510 p->maxcols <<= 2;
511
512 p->buf = mandoc_realloc
513 (p->buf, sizeof(int) * (size_t)p->maxcols);
514 }
515
516 static void
517 bufferc(struct termp *p, char c)
518 {
519
520 if (p->col + 1 >= p->maxcols)
521 adjbuf(p, p->col + 1);
522
523 p->buf[p->col++] = c;
524 }
525
526 /*
527 * See encode().
528 * Do this for a single (probably unicode) value.
529 * Does not check for non-decorated glyphs.
530 */
531 static void
532 encode1(struct termp *p, int c)
533 {
534 enum termfont f;
535
536 if (TERMP_SKIPCHAR & p->flags) {
537 p->flags &= ~TERMP_SKIPCHAR;
538 return;
539 }
540
541 if (p->col + 4 >= p->maxcols)
542 adjbuf(p, p->col + 4);
543
544 f = term_fonttop(p);
545
546 if (TERMFONT_NONE == f) {
547 p->buf[p->col++] = c;
548 return;
549 } else if (TERMFONT_UNDER == f) {
550 p->buf[p->col++] = '_';
551 } else
552 p->buf[p->col++] = c;
553
554 p->buf[p->col++] = 8;
555 p->buf[p->col++] = c;
556 }
557
558 static void
559 encode(struct termp *p, const char *word, size_t sz)
560 {
561 enum termfont f;
562 int i, len;
563
564 if (TERMP_SKIPCHAR & p->flags) {
565 p->flags &= ~TERMP_SKIPCHAR;
566 return;
567 }
568
569 /* LINTED */
570 len = sz;
571
572 /*
573 * Encode and buffer a string of characters. If the current
574 * font mode is unset, buffer directly, else encode then buffer
575 * character by character.
576 */
577
578 if (TERMFONT_NONE == (f = term_fonttop(p))) {
579 if (p->col + len >= p->maxcols)
580 adjbuf(p, p->col + len);
581 for (i = 0; i < len; i++)
582 p->buf[p->col++] = word[i];
583 return;
584 }
585
586 /* Pre-buffer, assuming worst-case. */
587
588 if (p->col + 1 + (len * 3) >= p->maxcols)
589 adjbuf(p, p->col + 1 + (len * 3));
590
591 for (i = 0; i < len; i++) {
592 if (ASCII_HYPH != word[i] &&
593 ! isgraph((unsigned char)word[i])) {
594 p->buf[p->col++] = word[i];
595 continue;
596 }
597
598 if (TERMFONT_UNDER == f)
599 p->buf[p->col++] = '_';
600 else if (ASCII_HYPH == word[i])
601 p->buf[p->col++] = '-';
602 else
603 p->buf[p->col++] = word[i];
604
605 p->buf[p->col++] = 8;
606 p->buf[p->col++] = word[i];
607 }
608 }
609
610 size_t
611 term_len(const struct termp *p, size_t sz)
612 {
613
614 return((*p->width)(p, ' ') * sz);
615 }
616
617 static size_t
618 cond_width(const struct termp *p, int c, int *skip)
619 {
620
621 if (*skip) {
622 (*skip) = 0;
623 return(0);
624 } else
625 return((*p->width)(p, c));
626 }
627
628 size_t
629 term_strlen(const struct termp *p, const char *cp)
630 {
631 size_t sz, rsz, i;
632 int ssz, skip, c;
633 const char *seq, *rhs;
634 enum mandoc_esc esc;
635 static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' };
636
637 /*
638 * Account for escaped sequences within string length
639 * calculations. This follows the logic in term_word() as we
640 * must calculate the width of produced strings.
641 */
642
643 sz = 0;
644 skip = 0;
645 while ('\0' != *cp) {
646 rsz = strcspn(cp, rej);
647 for (i = 0; i < rsz; i++)
648 sz += cond_width(p, *cp++, &skip);
649
650 c = 0;
651 switch (*cp) {
652 case ('\\'):
653 cp++;
654 esc = mandoc_escape(&cp, &seq, &ssz);
655 if (ESCAPE_ERROR == esc)
656 return(sz);
657
658 if (TERMENC_ASCII != p->enc)
659 switch (esc) {
660 case (ESCAPE_UNICODE):
661 c = mchars_num2uc
662 (seq + 1, ssz - 1);
663 if ('\0' == c)
664 break;
665 sz += cond_width(p, c, &skip);
666 continue;
667 case (ESCAPE_SPECIAL):
668 c = mchars_spec2cp
669 (p->symtab, seq, ssz);
670 if (c <= 0)
671 break;
672 sz += cond_width(p, c, &skip);
673 continue;
674 default:
675 break;
676 }
677
678 rhs = NULL;
679
680 switch (esc) {
681 case (ESCAPE_UNICODE):
682 sz += cond_width(p, '?', &skip);
683 break;
684 case (ESCAPE_NUMBERED):
685 c = mchars_num2char(seq, ssz);
686 if ('\0' != c)
687 sz += cond_width(p, c, &skip);
688 break;
689 case (ESCAPE_SPECIAL):
690 rhs = mchars_spec2str
691 (p->symtab, seq, ssz, &rsz);
692
693 if (ssz != 1 || rhs)
694 break;
695
696 rhs = seq;
697 rsz = ssz;
698 break;
699 case (ESCAPE_SKIPCHAR):
700 skip = 1;
701 break;
702 default:
703 break;
704 }
705
706 if (NULL == rhs)
707 break;
708
709 if (skip) {
710 skip = 0;
711 break;
712 }
713
714 for (i = 0; i < rsz; i++)
715 sz += (*p->width)(p, *rhs++);
716 break;
717 case (ASCII_NBRSP):
718 sz += cond_width(p, ' ', &skip);
719 cp++;
720 break;
721 case (ASCII_HYPH):
722 sz += cond_width(p, '-', &skip);
723 cp++;
724 break;
725 default:
726 break;
727 }
728 }
729
730 return(sz);
731 }
732
733 /* ARGSUSED */
734 size_t
735 term_vspan(const struct termp *p, const struct roffsu *su)
736 {
737 double r;
738
739 switch (su->unit) {
740 case (SCALE_CM):
741 r = su->scale * 2;
742 break;
743 case (SCALE_IN):
744 r = su->scale * 6;
745 break;
746 case (SCALE_PC):
747 r = su->scale;
748 break;
749 case (SCALE_PT):
750 r = su->scale / 8;
751 break;
752 case (SCALE_MM):
753 r = su->scale / 1000;
754 break;
755 case (SCALE_VS):
756 r = su->scale;
757 break;
758 default:
759 r = su->scale - 1;
760 break;
761 }
762
763 if (r < 0.0)
764 r = 0.0;
765 return(/* LINTED */(size_t)
766 r);
767 }
768
769 size_t
770 term_hspan(const struct termp *p, const struct roffsu *su)
771 {
772 double v;
773
774 v = ((*p->hspan)(p, su));
775 if (v < 0.0)
776 v = 0.0;
777 return((size_t) /* LINTED */
778 v);
779 }