]> git.cameronkatri.com Git - mandoc.git/blob - term.c
When a file is given on the command line, actually exists, and its name
[mandoc.git] / term.c
1 /* $Id: term.c,v 1.242 2014/12/24 23:32:42 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2014 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 #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 *p, 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
40
41 void
42 term_free(struct termp *p)
43 {
44
45 free(p->buf);
46 free(p->fontq);
47 free(p);
48 }
49
50 void
51 term_begin(struct termp *p, term_margin head,
52 term_margin foot, const void *arg)
53 {
54
55 p->headf = head;
56 p->footf = foot;
57 p->argf = arg;
58 (*p->begin)(p);
59 }
60
61 void
62 term_end(struct termp *p)
63 {
64
65 (*p->end)(p);
66 }
67
68 /*
69 * Flush a chunk of text. By default, break the output line each time
70 * the right margin is reached, and continue output on the next line
71 * at the same offset as the chunk itself. By default, also break the
72 * output line at the end of the chunk.
73 * The following flags may be specified:
74 *
75 * - TERMP_NOBREAK: Do not break the output line at the right margin,
76 * but only at the max right margin. Also, do not break the output
77 * line at the end of the chunk, such that the next call can pad to
78 * the next column. However, if less than p->trailspace blanks,
79 * which can be 0, 1, or 2, remain to the right margin, the line
80 * will be broken.
81 * - TERMP_BRIND: If the chunk does not fit and the output line has
82 * to be broken, start the next line at the right margin instead
83 * of at the offset. Used together with TERMP_NOBREAK for the tags
84 * in various kinds of tagged lists.
85 * - TERMP_DANGLE: Do not break the output line at the right margin,
86 * append the next chunk after it even if this one is too long.
87 * To be used together with TERMP_NOBREAK.
88 * - TERMP_HANG: Like TERMP_DANGLE, and also suppress padding before
89 * the next chunk if this column is not full.
90 */
91 void
92 term_flushln(struct termp *p)
93 {
94 size_t i; /* current input position in p->buf */
95 int ntab; /* number of tabs to prepend */
96 size_t vis; /* current visual position on output */
97 size_t vbl; /* number of blanks to prepend to output */
98 size_t vend; /* end of word visual position on output */
99 size_t bp; /* visual right border position */
100 size_t dv; /* temporary for visual pos calculations */
101 size_t j; /* temporary loop index for p->buf */
102 size_t jhy; /* last hyph before overflow w/r/t j */
103 size_t maxvis; /* output position of visible boundary */
104
105 /*
106 * First, establish the maximum columns of "visible" content.
107 * This is usually the difference between the right-margin and
108 * an indentation, but can be, for tagged lists or columns, a
109 * small set of values.
110 *
111 * The following unsigned-signed subtractions look strange,
112 * but they are actually correct. If the int p->overstep
113 * is negative, it gets sign extended. Subtracting that
114 * very large size_t effectively adds a small number to dv.
115 */
116 dv = p->rmargin > p->offset ? p->rmargin - p->offset : 0;
117 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
118
119 if (p->flags & TERMP_NOBREAK) {
120 dv = p->maxrmargin > p->offset ?
121 p->maxrmargin - p->offset : 0;
122 bp = (int)dv > p->overstep ?
123 dv - (size_t)p->overstep : 0;
124 } else
125 bp = maxvis;
126
127 /*
128 * Calculate the required amount of padding.
129 */
130 vbl = p->offset + p->overstep > p->viscol ?
131 p->offset + p->overstep - p->viscol : 0;
132
133 vis = vend = 0;
134 i = 0;
135
136 while (i < p->col) {
137 /*
138 * Handle literal tab characters: collapse all
139 * subsequent tabs into a single huge set of spaces.
140 */
141 ntab = 0;
142 while (i < p->col && '\t' == p->buf[i]) {
143 vend = (vis / p->tabwidth + 1) * p->tabwidth;
144 vbl += vend - vis;
145 vis = vend;
146 ntab++;
147 i++;
148 }
149
150 /*
151 * Count up visible word characters. Control sequences
152 * (starting with the CSI) aren't counted. A space
153 * generates a non-printing word, which is valid (the
154 * space is printed according to regular spacing rules).
155 */
156
157 for (j = i, jhy = 0; j < p->col; j++) {
158 if (' ' == p->buf[j] || '\t' == p->buf[j])
159 break;
160
161 /* Back over the the last printed character. */
162 if (8 == p->buf[j]) {
163 assert(j);
164 vend -= (*p->width)(p, p->buf[j - 1]);
165 continue;
166 }
167
168 /* Regular word. */
169 /* Break at the hyphen point if we overrun. */
170 if (vend > vis && vend < bp &&
171 (ASCII_HYPH == p->buf[j] ||
172 ASCII_BREAK == p->buf[j]))
173 jhy = j;
174
175 /*
176 * Hyphenation now decided, put back a real
177 * hyphen such that we get the correct width.
178 */
179 if (ASCII_HYPH == p->buf[j])
180 p->buf[j] = '-';
181
182 vend += (*p->width)(p, p->buf[j]);
183 }
184
185 /*
186 * Find out whether we would exceed the right margin.
187 * If so, break to the next line.
188 */
189 if (vend > bp && 0 == jhy && vis > 0) {
190 vend -= vis;
191 (*p->endline)(p);
192 p->viscol = 0;
193 if (TERMP_BRIND & p->flags) {
194 vbl = p->rmargin;
195 vend += p->rmargin;
196 vend -= p->offset;
197 } else
198 vbl = p->offset;
199
200 /* use pending tabs on the new line */
201
202 if (0 < ntab)
203 vbl += ntab * p->tabwidth;
204
205 /*
206 * Remove the p->overstep width.
207 * Again, if p->overstep is negative,
208 * sign extension does the right thing.
209 */
210
211 bp += (size_t)p->overstep;
212 p->overstep = 0;
213 }
214
215 /* Write out the [remaining] word. */
216 for ( ; i < p->col; i++) {
217 if (vend > bp && jhy > 0 && i > jhy)
218 break;
219 if ('\t' == p->buf[i])
220 break;
221 if (' ' == p->buf[i]) {
222 j = i;
223 while (i < p->col && ' ' == p->buf[i])
224 i++;
225 dv = (i - j) * (*p->width)(p, ' ');
226 vbl += dv;
227 vend += dv;
228 break;
229 }
230 if (ASCII_NBRSP == p->buf[i]) {
231 vbl += (*p->width)(p, ' ');
232 continue;
233 }
234 if (ASCII_BREAK == p->buf[i])
235 continue;
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 (*p->letter)(p, p->buf[i]);
249 if (8 == p->buf[i])
250 p->viscol -= (*p->width)(p, p->buf[i-1]);
251 else
252 p->viscol += (*p->width)(p, p->buf[i]);
253 }
254 vis = vend;
255 }
256
257 /*
258 * If there was trailing white space, it was not printed;
259 * so reset the cursor position accordingly.
260 */
261 if (vis > vbl)
262 vis -= vbl;
263 else
264 vis = 0;
265
266 p->col = 0;
267 p->overstep = 0;
268
269 if ( ! (TERMP_NOBREAK & p->flags)) {
270 p->viscol = 0;
271 (*p->endline)(p);
272 return;
273 }
274
275 if (TERMP_HANG & p->flags) {
276 p->overstep += (int)(p->offset + vis - p->rmargin +
277 p->trailspace * (*p->width)(p, ' '));
278
279 /*
280 * If we have overstepped the margin, temporarily move
281 * it to the right and flag the rest of the line to be
282 * shorter.
283 * If there is a request to keep the columns together,
284 * allow negative overstep when the column is not full.
285 */
286 if (p->trailspace && p->overstep < 0)
287 p->overstep = 0;
288 return;
289
290 } else if (TERMP_DANGLE & p->flags)
291 return;
292
293 /* If the column was overrun, break the line. */
294 if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) {
295 (*p->endline)(p);
296 p->viscol = 0;
297 }
298 }
299
300 /*
301 * A newline only breaks an existing line; it won't assert vertical
302 * space. All data in the output buffer is flushed prior to the newline
303 * assertion.
304 */
305 void
306 term_newln(struct termp *p)
307 {
308
309 p->flags |= TERMP_NOSPACE;
310 if (p->col || p->viscol)
311 term_flushln(p);
312 }
313
314 /*
315 * Asserts a vertical space (a full, empty line-break between lines).
316 * Note that if used twice, this will cause two blank spaces and so on.
317 * All data in the output buffer is flushed prior to the newline
318 * assertion.
319 */
320 void
321 term_vspace(struct termp *p)
322 {
323
324 term_newln(p);
325 p->viscol = 0;
326 if (0 < p->skipvsp)
327 p->skipvsp--;
328 else
329 (*p->endline)(p);
330 }
331
332 /* Swap current and previous font; for \fP and .ft P */
333 void
334 term_fontlast(struct termp *p)
335 {
336 enum termfont f;
337
338 f = p->fontl;
339 p->fontl = p->fontq[p->fonti];
340 p->fontq[p->fonti] = f;
341 }
342
343 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
344 void
345 term_fontrepl(struct termp *p, enum termfont f)
346 {
347
348 p->fontl = p->fontq[p->fonti];
349 p->fontq[p->fonti] = f;
350 }
351
352 /* Set font, save previous. */
353 void
354 term_fontpush(struct termp *p, enum termfont f)
355 {
356
357 p->fontl = p->fontq[p->fonti];
358 if (++p->fonti == p->fontsz) {
359 p->fontsz += 8;
360 p->fontq = mandoc_reallocarray(p->fontq,
361 p->fontsz, sizeof(enum termfont *));
362 }
363 p->fontq[p->fonti] = f;
364 }
365
366 /* Retrieve pointer to current font. */
367 const enum termfont *
368 term_fontq(struct termp *p)
369 {
370
371 return(&p->fontq[p->fonti]);
372 }
373
374 /* Flush to make the saved pointer current again. */
375 void
376 term_fontpopq(struct termp *p, const enum termfont *key)
377 {
378
379 while (p->fonti >= 0 && key < p->fontq + p->fonti)
380 p->fonti--;
381 assert(p->fonti >= 0);
382 }
383
384 /* Pop one font off the stack. */
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 nbrsp[2] = { ASCII_NBRSP, 0 };
402 const char *seq, *cp;
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 bufferc(p, ' ');
410 if (TERMP_SENTENCE & p->flags)
411 bufferc(p, ' ');
412 } else
413 bufferc(p, ASCII_NBRSP);
414 }
415 if (TERMP_PREKEEP & p->flags)
416 p->flags |= TERMP_KEEP;
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_NONEWLINE);
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 if (TERMP_NBRWORD & p->flags) {
433 if (' ' == *word) {
434 encode(p, nbrsp, 1);
435 word++;
436 continue;
437 }
438 ssz = strcspn(word, "\\ ");
439 } else
440 ssz = strcspn(word, "\\");
441 encode(p, word, ssz);
442 word += (int)ssz;
443 continue;
444 }
445
446 word++;
447 esc = mandoc_escape(&word, &seq, &sz);
448 if (ESCAPE_ERROR == esc)
449 continue;
450
451 switch (esc) {
452 case ESCAPE_UNICODE:
453 uc = mchars_num2uc(seq + 1, sz - 1);
454 break;
455 case ESCAPE_NUMBERED:
456 uc = mchars_num2char(seq, sz);
457 if (uc < 0)
458 continue;
459 break;
460 case ESCAPE_SPECIAL:
461 if (p->enc == TERMENC_ASCII) {
462 cp = mchars_spec2str(p->symtab,
463 seq, sz, &ssz);
464 if (cp != NULL)
465 encode(p, cp, ssz);
466 } else {
467 uc = mchars_spec2cp(p->symtab, seq, sz);
468 if (uc > 0)
469 encode1(p, uc);
470 }
471 continue;
472 case ESCAPE_FONTBOLD:
473 term_fontrepl(p, TERMFONT_BOLD);
474 continue;
475 case ESCAPE_FONTITALIC:
476 term_fontrepl(p, TERMFONT_UNDER);
477 continue;
478 case ESCAPE_FONTBI:
479 term_fontrepl(p, TERMFONT_BI);
480 continue;
481 case ESCAPE_FONT:
482 /* FALLTHROUGH */
483 case ESCAPE_FONTROMAN:
484 term_fontrepl(p, TERMFONT_NONE);
485 continue;
486 case ESCAPE_FONTPREV:
487 term_fontlast(p);
488 continue;
489 case ESCAPE_NOSPACE:
490 if (TERMP_SKIPCHAR & p->flags)
491 p->flags &= ~TERMP_SKIPCHAR;
492 else if ('\0' == *word)
493 p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
494 continue;
495 case ESCAPE_SKIPCHAR:
496 p->flags |= TERMP_SKIPCHAR;
497 continue;
498 default:
499 continue;
500 }
501
502 /*
503 * Common handling for Unicode and numbered
504 * character escape sequences.
505 */
506
507 if (p->enc == TERMENC_ASCII) {
508 cp = ascii_uc2str(uc);
509 encode(p, cp, strlen(cp));
510 } else {
511 if ((uc < 0x20 && uc != 0x09) ||
512 (uc > 0x7E && uc < 0xA0))
513 uc = 0xFFFD;
514 encode1(p, uc);
515 }
516 }
517 p->flags &= ~TERMP_NBRWORD;
518 }
519
520 static void
521 adjbuf(struct termp *p, size_t sz)
522 {
523
524 if (0 == p->maxcols)
525 p->maxcols = 1024;
526 while (sz >= p->maxcols)
527 p->maxcols <<= 2;
528
529 p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int));
530 }
531
532 static void
533 bufferc(struct termp *p, char c)
534 {
535
536 if (p->col + 1 >= p->maxcols)
537 adjbuf(p, p->col + 1);
538
539 p->buf[p->col++] = c;
540 }
541
542 /*
543 * See encode().
544 * Do this for a single (probably unicode) value.
545 * Does not check for non-decorated glyphs.
546 */
547 static void
548 encode1(struct termp *p, int c)
549 {
550 enum termfont f;
551
552 if (TERMP_SKIPCHAR & p->flags) {
553 p->flags &= ~TERMP_SKIPCHAR;
554 return;
555 }
556
557 if (p->col + 6 >= p->maxcols)
558 adjbuf(p, p->col + 6);
559
560 f = *term_fontq(p);
561
562 if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
563 p->buf[p->col++] = '_';
564 p->buf[p->col++] = 8;
565 }
566 if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
567 if (ASCII_HYPH == c)
568 p->buf[p->col++] = '-';
569 else
570 p->buf[p->col++] = c;
571 p->buf[p->col++] = 8;
572 }
573 p->buf[p->col++] = c;
574 }
575
576 static void
577 encode(struct termp *p, const char *word, size_t sz)
578 {
579 size_t i;
580
581 if (TERMP_SKIPCHAR & p->flags) {
582 p->flags &= ~TERMP_SKIPCHAR;
583 return;
584 }
585
586 /*
587 * Encode and buffer a string of characters. If the current
588 * font mode is unset, buffer directly, else encode then buffer
589 * character by character.
590 */
591
592 if (*term_fontq(p) == TERMFONT_NONE) {
593 if (p->col + sz >= p->maxcols)
594 adjbuf(p, p->col + sz);
595 for (i = 0; i < sz; i++)
596 p->buf[p->col++] = word[i];
597 return;
598 }
599
600 /* Pre-buffer, assuming worst-case. */
601
602 if (p->col + 1 + (sz * 5) >= p->maxcols)
603 adjbuf(p, p->col + 1 + (sz * 5));
604
605 for (i = 0; i < sz; i++) {
606 if (ASCII_HYPH == word[i] ||
607 isgraph((unsigned char)word[i]))
608 encode1(p, word[i]);
609 else
610 p->buf[p->col++] = word[i];
611 }
612 }
613
614 void
615 term_setwidth(struct termp *p, const char *wstr)
616 {
617 struct roffsu su;
618 size_t width;
619 int iop;
620
621 iop = 0;
622 width = 0;
623 if (NULL != wstr) {
624 switch (*wstr) {
625 case '+':
626 iop = 1;
627 wstr++;
628 break;
629 case '-':
630 iop = -1;
631 wstr++;
632 break;
633 default:
634 break;
635 }
636 if (a2roffsu(wstr, &su, SCALE_MAX))
637 width = term_hspan(p, &su);
638 else
639 iop = 0;
640 }
641 (*p->setwidth)(p, iop, width);
642 }
643
644 size_t
645 term_len(const struct termp *p, size_t sz)
646 {
647
648 return((*p->width)(p, ' ') * sz);
649 }
650
651 static size_t
652 cond_width(const struct termp *p, int c, int *skip)
653 {
654
655 if (*skip) {
656 (*skip) = 0;
657 return(0);
658 } else
659 return((*p->width)(p, c));
660 }
661
662 size_t
663 term_strlen(const struct termp *p, const char *cp)
664 {
665 size_t sz, rsz, i;
666 int ssz, skip, uc;
667 const char *seq, *rhs;
668 enum mandoc_esc esc;
669 static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
670 ASCII_BREAK, '\0' };
671
672 /*
673 * Account for escaped sequences within string length
674 * calculations. This follows the logic in term_word() as we
675 * must calculate the width of produced strings.
676 */
677
678 sz = 0;
679 skip = 0;
680 while ('\0' != *cp) {
681 rsz = strcspn(cp, rej);
682 for (i = 0; i < rsz; i++)
683 sz += cond_width(p, *cp++, &skip);
684
685 switch (*cp) {
686 case '\\':
687 cp++;
688 esc = mandoc_escape(&cp, &seq, &ssz);
689 if (ESCAPE_ERROR == esc)
690 continue;
691
692 rhs = NULL;
693
694 switch (esc) {
695 case ESCAPE_UNICODE:
696 uc = mchars_num2uc(seq + 1, ssz - 1);
697 break;
698 case ESCAPE_NUMBERED:
699 uc = mchars_num2char(seq, ssz);
700 if (uc < 0)
701 continue;
702 break;
703 case ESCAPE_SPECIAL:
704 if (p->enc == TERMENC_ASCII) {
705 rhs = mchars_spec2str(p->symtab,
706 seq, ssz, &rsz);
707 if (rhs != NULL)
708 break;
709 } else {
710 uc = mchars_spec2cp(p->symtab,
711 seq, ssz);
712 if (uc > 0)
713 sz += cond_width(p, uc, &skip);
714 }
715 continue;
716 case ESCAPE_SKIPCHAR:
717 skip = 1;
718 continue;
719 default:
720 continue;
721 }
722
723 /*
724 * Common handling for Unicode and numbered
725 * character escape sequences.
726 */
727
728 if (rhs == NULL) {
729 if (p->enc == TERMENC_ASCII) {
730 rhs = ascii_uc2str(uc);
731 rsz = strlen(rhs);
732 } else {
733 if ((uc < 0x20 && uc != 0x09) ||
734 (uc > 0x7E && uc < 0xA0))
735 uc = 0xFFFD;
736 sz += cond_width(p, uc, &skip);
737 continue;
738 }
739 }
740
741 if (skip) {
742 skip = 0;
743 break;
744 }
745
746 /*
747 * Common handling for all escape sequences
748 * printing more than one character.
749 */
750
751 for (i = 0; i < rsz; i++)
752 sz += (*p->width)(p, *rhs++);
753 break;
754 case ASCII_NBRSP:
755 sz += cond_width(p, ' ', &skip);
756 cp++;
757 break;
758 case ASCII_HYPH:
759 sz += cond_width(p, '-', &skip);
760 cp++;
761 /* FALLTHROUGH */
762 case ASCII_BREAK:
763 break;
764 default:
765 break;
766 }
767 }
768
769 return(sz);
770 }
771
772 int
773 term_vspan(const struct termp *p, const struct roffsu *su)
774 {
775 double r;
776 int ri;
777
778 switch (su->unit) {
779 case SCALE_BU:
780 r = su->scale / 40.0;
781 break;
782 case SCALE_CM:
783 r = su->scale * 6.0 / 2.54;
784 break;
785 case SCALE_FS:
786 r = su->scale * 65536.0 / 40.0;
787 break;
788 case SCALE_IN:
789 r = su->scale * 6.0;
790 break;
791 case SCALE_MM:
792 r = su->scale * 0.006;
793 break;
794 case SCALE_PC:
795 r = su->scale;
796 break;
797 case SCALE_PT:
798 r = su->scale / 12.0;
799 break;
800 case SCALE_EN:
801 /* FALLTHROUGH */
802 case SCALE_EM:
803 r = su->scale * 0.6;
804 break;
805 case SCALE_VS:
806 r = su->scale;
807 break;
808 default:
809 abort();
810 /* NOTREACHED */
811 }
812 ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
813 return(ri < 66 ? ri : 1);
814 }
815
816 int
817 term_hspan(const struct termp *p, const struct roffsu *su)
818 {
819 double v;
820
821 v = (*p->hspan)(p, su);
822 return(v > 0.0 ? v + 0.0005 : v - 0.0005);
823 }