]> git.cameronkatri.com Git - mandoc.git/blob - term.c
In flush-left mode of both man(7) and mdoc(7), when an output line is broken
[mandoc.git] / term.c
1 /* $Id: term.c,v 1.205 2012/07/16 21:30:42 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 int ntab; /* number of tabs to prepend */
110 size_t vis; /* current visual position on output */
111 size_t vbl; /* number of blanks to prepend to output */
112 size_t vend; /* end of word visual position on output */
113 size_t bp; /* visual right border position */
114 size_t dv; /* temporary for visual pos calculations */
115 int j; /* temporary loop index for p->buf */
116 int jhy; /* last hyph before overflow w/r/t j */
117 size_t maxvis; /* output position of visible boundary */
118 size_t mmax; /* used in calculating bp */
119
120 /*
121 * First, establish the maximum columns of "visible" content.
122 * This is usually the difference between the right-margin and
123 * an indentation, but can be, for tagged lists or columns, a
124 * small set of values.
125 */
126 assert (p->rmargin >= p->offset);
127 dv = p->rmargin - p->offset;
128 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
129 dv = p->maxrmargin - p->offset;
130 mmax = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
131
132 bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
133
134 /*
135 * Calculate the required amount of padding.
136 */
137 vbl = p->offset + p->overstep > p->viscol ?
138 p->offset + p->overstep - p->viscol : 0;
139
140 vis = vend = 0;
141 i = 0;
142
143 while (i < p->col) {
144 /*
145 * Handle literal tab characters: collapse all
146 * subsequent tabs into a single huge set of spaces.
147 */
148 ntab = 0;
149 while (i < p->col && '\t' == p->buf[i]) {
150 vend = (vis / p->tabwidth + 1) * p->tabwidth;
151 vbl += vend - vis;
152 vis = vend;
153 ntab++;
154 i++;
155 }
156
157 /*
158 * Count up visible word characters. Control sequences
159 * (starting with the CSI) aren't counted. A space
160 * generates a non-printing word, which is valid (the
161 * space is printed according to regular spacing rules).
162 */
163
164 for (j = i, jhy = 0; j < p->col; j++) {
165 if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
166 break;
167
168 /* Back over the the last printed character. */
169 if (8 == p->buf[j]) {
170 assert(j);
171 vend -= (*p->width)(p, p->buf[j - 1]);
172 continue;
173 }
174
175 /* Regular word. */
176 /* Break at the hyphen point if we overrun. */
177 if (vend > vis && vend < bp &&
178 ASCII_HYPH == p->buf[j])
179 jhy = j;
180
181 vend += (*p->width)(p, p->buf[j]);
182 }
183
184 /*
185 * Find out whether we would exceed the right margin.
186 * If so, break to the next line.
187 */
188 if (vend > bp && 0 == jhy && vis > 0) {
189 vend -= vis;
190 (*p->endline)(p);
191 p->viscol = 0;
192 if (TERMP_NOBREAK & p->flags) {
193 vbl = p->rmargin;
194 vend += p->rmargin - p->offset;
195 } else
196 vbl = p->offset;
197
198 /* use pending tabs on the new line */
199
200 if (0 < ntab)
201 vbl += ntab * p->tabwidth;
202
203 /* Remove the p->overstep width. */
204
205 bp += (size_t)p->overstep;
206 p->overstep = 0;
207 }
208
209 /* Write out the [remaining] word. */
210 for ( ; i < p->col; i++) {
211 if (vend > bp && jhy > 0 && i > jhy)
212 break;
213 if ('\t' == p->buf[i])
214 break;
215 if (' ' == p->buf[i]) {
216 j = i;
217 while (' ' == p->buf[i])
218 i++;
219 dv = (size_t)(i - j) * (*p->width)(p, ' ');
220 vbl += dv;
221 vend += dv;
222 break;
223 }
224 if (ASCII_NBRSP == p->buf[i]) {
225 vbl += (*p->width)(p, ' ');
226 continue;
227 }
228
229 /*
230 * Now we definitely know there will be
231 * printable characters to output,
232 * so write preceding white space now.
233 */
234 if (vbl) {
235 (*p->advance)(p, vbl);
236 p->viscol += vbl;
237 vbl = 0;
238 }
239
240 if (ASCII_HYPH == p->buf[i]) {
241 (*p->letter)(p, '-');
242 p->viscol += (*p->width)(p, '-');
243 continue;
244 }
245
246 (*p->letter)(p, p->buf[i]);
247 if (8 == p->buf[i])
248 p->viscol -= (*p->width)(p, p->buf[i-1]);
249 else
250 p->viscol += (*p->width)(p, p->buf[i]);
251 }
252 vis = vend;
253 }
254
255 /*
256 * If there was trailing white space, it was not printed;
257 * so reset the cursor position accordingly.
258 */
259 if (vis)
260 vis -= vbl;
261
262 p->col = 0;
263 p->overstep = 0;
264
265 if ( ! (TERMP_NOBREAK & p->flags)) {
266 p->viscol = 0;
267 (*p->endline)(p);
268 return;
269 }
270
271 if (TERMP_HANG & p->flags) {
272 /* We need one blank after the tag. */
273 p->overstep = (int)(vis - maxvis + (*p->width)(p, ' '));
274
275 /*
276 * If we have overstepped the margin, temporarily move
277 * it to the right and flag the rest of the line to be
278 * shorter.
279 */
280 if (p->overstep < 0)
281 p->overstep = 0;
282 return;
283
284 } else if (TERMP_DANGLE & p->flags)
285 return;
286
287 /* If the column was overrun, break the line. */
288 if (maxvis <= vis +
289 ((TERMP_TWOSPACE & p->flags) ? (*p->width)(p, ' ') : 0)) {
290 (*p->endline)(p);
291 p->viscol = 0;
292 }
293 }
294
295
296 /*
297 * A newline only breaks an existing line; it won't assert vertical
298 * space. All data in the output buffer is flushed prior to the newline
299 * assertion.
300 */
301 void
302 term_newln(struct termp *p)
303 {
304
305 p->flags |= TERMP_NOSPACE;
306 if (p->col || p->viscol)
307 term_flushln(p);
308 }
309
310
311 /*
312 * Asserts a vertical space (a full, empty line-break between lines).
313 * Note that if used twice, this will cause two blank spaces and so on.
314 * All data in the output buffer is flushed prior to the newline
315 * assertion.
316 */
317 void
318 term_vspace(struct termp *p)
319 {
320
321 term_newln(p);
322 p->viscol = 0;
323 if (0 < p->skipvsp)
324 p->skipvsp--;
325 else
326 (*p->endline)(p);
327 }
328
329 void
330 term_fontlast(struct termp *p)
331 {
332 enum termfont f;
333
334 f = p->fontl;
335 p->fontl = p->fontq[p->fonti];
336 p->fontq[p->fonti] = f;
337 }
338
339
340 void
341 term_fontrepl(struct termp *p, enum termfont f)
342 {
343
344 p->fontl = p->fontq[p->fonti];
345 p->fontq[p->fonti] = f;
346 }
347
348
349 void
350 term_fontpush(struct termp *p, enum termfont f)
351 {
352
353 assert(p->fonti + 1 < 10);
354 p->fontl = p->fontq[p->fonti];
355 p->fontq[++p->fonti] = f;
356 }
357
358
359 const void *
360 term_fontq(struct termp *p)
361 {
362
363 return(&p->fontq[p->fonti]);
364 }
365
366
367 enum termfont
368 term_fonttop(struct termp *p)
369 {
370
371 return(p->fontq[p->fonti]);
372 }
373
374
375 void
376 term_fontpopq(struct termp *p, const void *key)
377 {
378
379 while (p->fonti >= 0 && key != &p->fontq[p->fonti])
380 p->fonti--;
381 assert(p->fonti >= 0);
382 }
383
384
385 void
386 term_fontpop(struct termp *p)
387 {
388
389 assert(p->fonti);
390 p->fonti--;
391 }
392
393 /*
394 * Handle pwords, partial words, which may be either a single word or a
395 * phrase that cannot be broken down (such as a literal string). This
396 * handles word styling.
397 */
398 void
399 term_word(struct termp *p, const char *word)
400 {
401 const char *seq, *cp;
402 char c;
403 int sz, uc;
404 size_t ssz;
405 enum mandoc_esc esc;
406
407 if ( ! (TERMP_NOSPACE & p->flags)) {
408 if ( ! (TERMP_KEEP & p->flags)) {
409 if (TERMP_PREKEEP & p->flags)
410 p->flags |= TERMP_KEEP;
411 bufferc(p, ' ');
412 if (TERMP_SENTENCE & p->flags)
413 bufferc(p, ' ');
414 } else
415 bufferc(p, ASCII_NBRSP);
416 }
417
418 if ( ! (p->flags & TERMP_NONOSPACE))
419 p->flags &= ~TERMP_NOSPACE;
420 else
421 p->flags |= TERMP_NOSPACE;
422
423 p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
424
425 while ('\0' != *word) {
426 if ('\\' != *word) {
427 if (TERMP_SKIPCHAR & p->flags) {
428 p->flags &= ~TERMP_SKIPCHAR;
429 word++;
430 continue;
431 }
432 ssz = strcspn(word, "\\");
433 encode(p, word, ssz);
434 word += (int)ssz;
435 continue;
436 }
437
438 word++;
439 esc = mandoc_escape(&word, &seq, &sz);
440 if (ESCAPE_ERROR == esc)
441 break;
442
443 if (TERMENC_ASCII != p->enc)
444 switch (esc) {
445 case (ESCAPE_UNICODE):
446 uc = mchars_num2uc(seq + 1, sz - 1);
447 if ('\0' == uc)
448 break;
449 encode1(p, uc);
450 continue;
451 case (ESCAPE_SPECIAL):
452 uc = mchars_spec2cp(p->symtab, seq, sz);
453 if (uc <= 0)
454 break;
455 encode1(p, uc);
456 continue;
457 default:
458 break;
459 }
460
461 switch (esc) {
462 case (ESCAPE_UNICODE):
463 encode1(p, '?');
464 break;
465 case (ESCAPE_NUMBERED):
466 c = mchars_num2char(seq, sz);
467 if ('\0' != c)
468 encode(p, &c, 1);
469 break;
470 case (ESCAPE_SPECIAL):
471 cp = mchars_spec2str(p->symtab, seq, sz, &ssz);
472 if (NULL != cp)
473 encode(p, cp, ssz);
474 else if (1 == ssz)
475 encode(p, seq, sz);
476 break;
477 case (ESCAPE_FONTBOLD):
478 term_fontrepl(p, TERMFONT_BOLD);
479 break;
480 case (ESCAPE_FONTITALIC):
481 term_fontrepl(p, TERMFONT_UNDER);
482 break;
483 case (ESCAPE_FONT):
484 /* FALLTHROUGH */
485 case (ESCAPE_FONTROMAN):
486 term_fontrepl(p, TERMFONT_NONE);
487 break;
488 case (ESCAPE_FONTPREV):
489 term_fontlast(p);
490 break;
491 case (ESCAPE_NOSPACE):
492 if (TERMP_SKIPCHAR & p->flags)
493 p->flags &= ~TERMP_SKIPCHAR;
494 else if ('\0' == *word)
495 p->flags |= TERMP_NOSPACE;
496 break;
497 case (ESCAPE_SKIPCHAR):
498 p->flags |= TERMP_SKIPCHAR;
499 break;
500 default:
501 break;
502 }
503 }
504 }
505
506 static void
507 adjbuf(struct termp *p, int sz)
508 {
509
510 if (0 == p->maxcols)
511 p->maxcols = 1024;
512 while (sz >= p->maxcols)
513 p->maxcols <<= 2;
514
515 p->buf = mandoc_realloc
516 (p->buf, sizeof(int) * (size_t)p->maxcols);
517 }
518
519 static void
520 bufferc(struct termp *p, char c)
521 {
522
523 if (p->col + 1 >= p->maxcols)
524 adjbuf(p, p->col + 1);
525
526 p->buf[p->col++] = c;
527 }
528
529 /*
530 * See encode().
531 * Do this for a single (probably unicode) value.
532 * Does not check for non-decorated glyphs.
533 */
534 static void
535 encode1(struct termp *p, int c)
536 {
537 enum termfont f;
538
539 if (TERMP_SKIPCHAR & p->flags) {
540 p->flags &= ~TERMP_SKIPCHAR;
541 return;
542 }
543
544 if (p->col + 4 >= p->maxcols)
545 adjbuf(p, p->col + 4);
546
547 f = term_fonttop(p);
548
549 if (TERMFONT_NONE == f) {
550 p->buf[p->col++] = c;
551 return;
552 } else if (TERMFONT_UNDER == f) {
553 p->buf[p->col++] = '_';
554 } else
555 p->buf[p->col++] = c;
556
557 p->buf[p->col++] = 8;
558 p->buf[p->col++] = c;
559 }
560
561 static void
562 encode(struct termp *p, const char *word, size_t sz)
563 {
564 enum termfont f;
565 int i, len;
566
567 if (TERMP_SKIPCHAR & p->flags) {
568 p->flags &= ~TERMP_SKIPCHAR;
569 return;
570 }
571
572 /* LINTED */
573 len = sz;
574
575 /*
576 * Encode and buffer a string of characters. If the current
577 * font mode is unset, buffer directly, else encode then buffer
578 * character by character.
579 */
580
581 if (TERMFONT_NONE == (f = term_fonttop(p))) {
582 if (p->col + len >= p->maxcols)
583 adjbuf(p, p->col + len);
584 for (i = 0; i < len; i++)
585 p->buf[p->col++] = word[i];
586 return;
587 }
588
589 /* Pre-buffer, assuming worst-case. */
590
591 if (p->col + 1 + (len * 3) >= p->maxcols)
592 adjbuf(p, p->col + 1 + (len * 3));
593
594 for (i = 0; i < len; i++) {
595 if (ASCII_HYPH != word[i] &&
596 ! isgraph((unsigned char)word[i])) {
597 p->buf[p->col++] = word[i];
598 continue;
599 }
600
601 if (TERMFONT_UNDER == f)
602 p->buf[p->col++] = '_';
603 else if (ASCII_HYPH == word[i])
604 p->buf[p->col++] = '-';
605 else
606 p->buf[p->col++] = word[i];
607
608 p->buf[p->col++] = 8;
609 p->buf[p->col++] = word[i];
610 }
611 }
612
613 size_t
614 term_len(const struct termp *p, size_t sz)
615 {
616
617 return((*p->width)(p, ' ') * sz);
618 }
619
620 static size_t
621 cond_width(const struct termp *p, int c, int *skip)
622 {
623
624 if (*skip) {
625 (*skip) = 0;
626 return(0);
627 } else
628 return((*p->width)(p, c));
629 }
630
631 size_t
632 term_strlen(const struct termp *p, const char *cp)
633 {
634 size_t sz, rsz, i;
635 int ssz, skip, c;
636 const char *seq, *rhs;
637 enum mandoc_esc esc;
638 static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' };
639
640 /*
641 * Account for escaped sequences within string length
642 * calculations. This follows the logic in term_word() as we
643 * must calculate the width of produced strings.
644 */
645
646 sz = 0;
647 skip = 0;
648 while ('\0' != *cp) {
649 rsz = strcspn(cp, rej);
650 for (i = 0; i < rsz; i++)
651 sz += cond_width(p, *cp++, &skip);
652
653 c = 0;
654 switch (*cp) {
655 case ('\\'):
656 cp++;
657 esc = mandoc_escape(&cp, &seq, &ssz);
658 if (ESCAPE_ERROR == esc)
659 return(sz);
660
661 if (TERMENC_ASCII != p->enc)
662 switch (esc) {
663 case (ESCAPE_UNICODE):
664 c = mchars_num2uc
665 (seq + 1, ssz - 1);
666 if ('\0' == c)
667 break;
668 sz += cond_width(p, c, &skip);
669 continue;
670 case (ESCAPE_SPECIAL):
671 c = mchars_spec2cp
672 (p->symtab, seq, ssz);
673 if (c <= 0)
674 break;
675 sz += cond_width(p, c, &skip);
676 continue;
677 default:
678 break;
679 }
680
681 rhs = NULL;
682
683 switch (esc) {
684 case (ESCAPE_UNICODE):
685 sz += cond_width(p, '?', &skip);
686 break;
687 case (ESCAPE_NUMBERED):
688 c = mchars_num2char(seq, ssz);
689 if ('\0' != c)
690 sz += cond_width(p, c, &skip);
691 break;
692 case (ESCAPE_SPECIAL):
693 rhs = mchars_spec2str
694 (p->symtab, seq, ssz, &rsz);
695
696 if (ssz != 1 || rhs)
697 break;
698
699 rhs = seq;
700 rsz = ssz;
701 break;
702 case (ESCAPE_SKIPCHAR):
703 skip = 1;
704 break;
705 default:
706 break;
707 }
708
709 if (NULL == rhs)
710 break;
711
712 if (skip) {
713 skip = 0;
714 break;
715 }
716
717 for (i = 0; i < rsz; i++)
718 sz += (*p->width)(p, *rhs++);
719 break;
720 case (ASCII_NBRSP):
721 sz += cond_width(p, ' ', &skip);
722 cp++;
723 break;
724 case (ASCII_HYPH):
725 sz += cond_width(p, '-', &skip);
726 cp++;
727 break;
728 default:
729 break;
730 }
731 }
732
733 return(sz);
734 }
735
736 /* ARGSUSED */
737 size_t
738 term_vspan(const struct termp *p, const struct roffsu *su)
739 {
740 double r;
741
742 switch (su->unit) {
743 case (SCALE_CM):
744 r = su->scale * 2;
745 break;
746 case (SCALE_IN):
747 r = su->scale * 6;
748 break;
749 case (SCALE_PC):
750 r = su->scale;
751 break;
752 case (SCALE_PT):
753 r = su->scale / 8;
754 break;
755 case (SCALE_MM):
756 r = su->scale / 1000;
757 break;
758 case (SCALE_VS):
759 r = su->scale;
760 break;
761 default:
762 r = su->scale - 1;
763 break;
764 }
765
766 if (r < 0.0)
767 r = 0.0;
768 return(/* LINTED */(size_t)
769 r);
770 }
771
772 size_t
773 term_hspan(const struct termp *p, const struct roffsu *su)
774 {
775 double v;
776
777 v = ((*p->hspan)(p, su));
778 if (v < 0.0)
779 v = 0.0;
780 return((size_t) /* LINTED */
781 v);
782 }