xref: /netbsd-src/external/bsd/mdocml/dist/term.c (revision c2f76ff004a2cb67efe5b12d97bd3ef7fe89e18d) 
1 /*	$Vendor-Id: term.c,v 1.176 2011/01/04 13:14:26 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 *, enum roffdeco,
38 				const char *, size_t);
39 static	void		  res(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((int)MANDOCLEVEL_SYSERR);
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 	size_t		 dv;    /* temporary for visual pos calculations */
138 	int		 j;     /* temporary loop index for p->buf */
139 	int		 jhy;	/* last hyph before overflow w/r/t j */
140 	size_t		 maxvis; /* output position of visible boundary */
141 	size_t		 mmax; /* used in calculating bp */
142 
143 	/*
144 	 * First, establish the maximum columns of "visible" content.
145 	 * This is usually the difference between the right-margin and
146 	 * an indentation, but can be, for tagged lists or columns, a
147 	 * small set of values.
148 	 */
149 	assert  (p->rmargin >= p->offset);
150 	dv     = p->rmargin - p->offset;
151 	maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
152 	dv     = p->maxrmargin - p->offset;
153 	mmax   = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
154 
155 	bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
156 
157 	/*
158 	 * Indent the first line of a paragraph.
159 	 */
160 	vbl = p->flags & TERMP_NOLPAD ? (size_t)0 : p->offset;
161 
162 	vis = vend = 0;
163 	i = 0;
164 
165 	while (i < (int)p->col) {
166 		/*
167 		 * Handle literal tab characters: collapse all
168 		 * subsequent tabs into a single huge set of spaces.
169 		 */
170 		while (i < (int)p->col && '\t' == p->buf[i]) {
171 			vend = (vis / p->tabwidth + 1) * p->tabwidth;
172 			vbl += vend - vis;
173 			vis = vend;
174 			i++;
175 		}
176 
177 		/*
178 		 * Count up visible word characters.  Control sequences
179 		 * (starting with the CSI) aren't counted.  A space
180 		 * generates a non-printing word, which is valid (the
181 		 * space is printed according to regular spacing rules).
182 		 */
183 
184 		for (j = i, jhy = 0; j < (int)p->col; j++) {
185 			if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
186 				break;
187 
188 			/* Back over the the last printed character. */
189 			if (8 == p->buf[j]) {
190 				assert(j);
191 				vend -= (*p->width)(p, p->buf[j - 1]);
192 				continue;
193 			}
194 
195 			/* Regular word. */
196 			/* Break at the hyphen point if we overrun. */
197 			if (vend > vis && vend < bp &&
198 					ASCII_HYPH == p->buf[j])
199 				jhy = j;
200 
201 			vend += (*p->width)(p, p->buf[j]);
202 		}
203 
204 		/*
205 		 * Find out whether we would exceed the right margin.
206 		 * If so, break to the next line.
207 		 */
208 		if (vend > bp && 0 == jhy && vis > 0) {
209 			vend -= vis;
210 			(*p->endline)(p);
211 			if (TERMP_NOBREAK & p->flags) {
212 				p->viscol = p->rmargin;
213 				(*p->advance)(p, p->rmargin);
214 				vend += p->rmargin - p->offset;
215 			} else {
216 				p->viscol = 0;
217 				vbl = p->offset;
218 			}
219 
220 			/* Remove the p->overstep width. */
221 
222 			bp += (size_t)p->overstep;
223 			p->overstep = 0;
224 		}
225 
226 		/* Write out the [remaining] word. */
227 		for ( ; i < (int)p->col; i++) {
228 			if (vend > bp && jhy > 0 && i > jhy)
229 				break;
230 			if ('\t' == p->buf[i])
231 				break;
232 			if (' ' == p->buf[i]) {
233 				j = i;
234 				while (' ' == p->buf[i])
235 					i++;
236 				dv = (size_t)(i - j) * (*p->width)(p, ' ');
237 				vbl += dv;
238 				vend += dv;
239 				break;
240 			}
241 			if (ASCII_NBRSP == p->buf[i]) {
242 				vbl += (*p->width)(p, ' ');
243 				continue;
244 			}
245 
246 			/*
247 			 * Now we definitely know there will be
248 			 * printable characters to output,
249 			 * so write preceding white space now.
250 			 */
251 			if (vbl) {
252 				(*p->advance)(p, vbl);
253 				p->viscol += vbl;
254 				vbl = 0;
255 			}
256 
257 			if (ASCII_HYPH == p->buf[i]) {
258 				(*p->letter)(p, '-');
259 				p->viscol += (*p->width)(p, '-');
260 			} else {
261 				(*p->letter)(p, p->buf[i]);
262 				p->viscol += (*p->width)(p, p->buf[i]);
263 			}
264 		}
265 		vis = vend;
266 	}
267 
268 	/*
269 	 * If there was trailing white space, it was not printed;
270 	 * so reset the cursor position accordingly.
271 	 */
272 	vis -= vbl;
273 
274 	p->col = 0;
275 	p->overstep = 0;
276 
277 	if ( ! (TERMP_NOBREAK & p->flags)) {
278 		p->viscol = 0;
279 		(*p->endline)(p);
280 		return;
281 	}
282 
283 	if (TERMP_HANG & p->flags) {
284 		/* We need one blank after the tag. */
285 		p->overstep = (int)(vis - maxvis + (*p->width)(p, ' '));
286 
287 		/*
288 		 * Behave exactly the same way as groff:
289 		 * If we have overstepped the margin, temporarily move
290 		 * it to the right and flag the rest of the line to be
291 		 * shorter.
292 		 * If we landed right at the margin, be happy.
293 		 * If we are one step before the margin, temporarily
294 		 * move it one step LEFT and flag the rest of the line
295 		 * to be longer.
296 		 */
297 		if (p->overstep >= -1) {
298 			assert((int)maxvis + p->overstep >= 0);
299 			maxvis += (size_t)p->overstep;
300 		} else
301 			p->overstep = 0;
302 
303 	} else if (TERMP_DANGLE & p->flags)
304 		return;
305 
306 	/* Right-pad. */
307 	if (maxvis > vis +
308 	    ((TERMP_TWOSPACE & p->flags) ? (*p->width)(p, ' ') : 0)) {
309 		p->viscol += maxvis - vis;
310 		(*p->advance)(p, maxvis - vis);
311 		vis += (maxvis - vis);
312 	} else {	/* ...or newline break. */
313 		(*p->endline)(p);
314 		p->viscol = p->rmargin;
315 		(*p->advance)(p, p->rmargin);
316 	}
317 }
318 
319 
320 /*
321  * A newline only breaks an existing line; it won't assert vertical
322  * space.  All data in the output buffer is flushed prior to the newline
323  * assertion.
324  */
325 void
326 term_newln(struct termp *p)
327 {
328 
329 	p->flags |= TERMP_NOSPACE;
330 	if (0 == p->col && 0 == p->viscol) {
331 		p->flags &= ~TERMP_NOLPAD;
332 		return;
333 	}
334 	term_flushln(p);
335 	p->flags &= ~TERMP_NOLPAD;
336 }
337 
338 
339 /*
340  * Asserts a vertical space (a full, empty line-break between lines).
341  * Note that if used twice, this will cause two blank spaces and so on.
342  * All data in the output buffer is flushed prior to the newline
343  * assertion.
344  */
345 void
346 term_vspace(struct termp *p)
347 {
348 
349 	term_newln(p);
350 	p->viscol = 0;
351 	(*p->endline)(p);
352 }
353 
354 
355 static void
356 spec(struct termp *p, enum roffdeco d, const char *word, size_t len)
357 {
358 	const char	*rhs;
359 	size_t		 sz;
360 
361 	rhs = chars_spec2str(p->symtab, word, len, &sz);
362 	if (rhs)
363 		encode(p, rhs, sz);
364 	else if (DECO_SSPECIAL == d)
365 		encode(p, word, len);
366 }
367 
368 
369 static void
370 res(struct termp *p, const char *word, size_t len)
371 {
372 	const char	*rhs;
373 	size_t		 sz;
374 
375 	rhs = chars_res2str(p->symtab, word, len, &sz);
376 	if (rhs)
377 		encode(p, rhs, sz);
378 }
379 
380 
381 void
382 term_fontlast(struct termp *p)
383 {
384 	enum termfont	 f;
385 
386 	f = p->fontl;
387 	p->fontl = p->fontq[p->fonti];
388 	p->fontq[p->fonti] = f;
389 }
390 
391 
392 void
393 term_fontrepl(struct termp *p, enum termfont f)
394 {
395 
396 	p->fontl = p->fontq[p->fonti];
397 	p->fontq[p->fonti] = f;
398 }
399 
400 
401 void
402 term_fontpush(struct termp *p, enum termfont f)
403 {
404 
405 	assert(p->fonti + 1 < 10);
406 	p->fontl = p->fontq[p->fonti];
407 	p->fontq[++p->fonti] = f;
408 }
409 
410 
411 const void *
412 term_fontq(struct termp *p)
413 {
414 
415 	return(&p->fontq[p->fonti]);
416 }
417 
418 
419 enum termfont
420 term_fonttop(struct termp *p)
421 {
422 
423 	return(p->fontq[p->fonti]);
424 }
425 
426 
427 void
428 term_fontpopq(struct termp *p, const void *key)
429 {
430 
431 	while (p->fonti >= 0 && key != &p->fontq[p->fonti])
432 		p->fonti--;
433 	assert(p->fonti >= 0);
434 }
435 
436 
437 void
438 term_fontpop(struct termp *p)
439 {
440 
441 	assert(p->fonti);
442 	p->fonti--;
443 }
444 
445 
446 /*
447  * Handle pwords, partial words, which may be either a single word or a
448  * phrase that cannot be broken down (such as a literal string).  This
449  * handles word styling.
450  */
451 void
452 term_word(struct termp *p, const char *word)
453 {
454 	const char	*sv, *seq;
455 	size_t		 ssz;
456 	enum roffdeco	 deco;
457 
458 	sv = word;
459 
460 	if (word[0] && '\0' == word[1])
461 		switch (word[0]) {
462 		case('.'):
463 			/* FALLTHROUGH */
464 		case(','):
465 			/* FALLTHROUGH */
466 		case(';'):
467 			/* FALLTHROUGH */
468 		case(':'):
469 			/* FALLTHROUGH */
470 		case('?'):
471 			/* FALLTHROUGH */
472 		case('!'):
473 			/* FALLTHROUGH */
474 		case(')'):
475 			/* FALLTHROUGH */
476 		case(']'):
477 			if ( ! (TERMP_IGNDELIM & p->flags))
478 				p->flags |= TERMP_NOSPACE;
479 			break;
480 		default:
481 			break;
482 		}
483 
484 	if ( ! (TERMP_NOSPACE & p->flags)) {
485 		if ( ! (TERMP_KEEP & p->flags)) {
486 			if (TERMP_PREKEEP & p->flags)
487 				p->flags |= TERMP_KEEP;
488 			bufferc(p, ' ');
489 			if (TERMP_SENTENCE & p->flags)
490 				bufferc(p, ' ');
491 		} else
492 			bufferc(p, ASCII_NBRSP);
493 	}
494 
495 	if ( ! (p->flags & TERMP_NONOSPACE))
496 		p->flags &= ~TERMP_NOSPACE;
497 	else
498 		p->flags |= TERMP_NOSPACE;
499 
500 	p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
501 
502 	while (*word) {
503 		if ((ssz = strcspn(word, "\\")) > 0)
504 			encode(p, word, ssz);
505 
506 		word += ssz;
507 		if ('\\' != *word)
508 			continue;
509 
510 		seq = ++word;
511 		word += a2roffdeco(&deco, &seq, &ssz);
512 
513 		switch (deco) {
514 		case (DECO_RESERVED):
515 			res(p, seq, ssz);
516 			break;
517 		case (DECO_SPECIAL):
518 			/* FALLTHROUGH */
519 		case (DECO_SSPECIAL):
520 			spec(p, deco, seq, ssz);
521 			break;
522 		case (DECO_BOLD):
523 			term_fontrepl(p, TERMFONT_BOLD);
524 			break;
525 		case (DECO_ITALIC):
526 			term_fontrepl(p, TERMFONT_UNDER);
527 			break;
528 		case (DECO_ROMAN):
529 			term_fontrepl(p, TERMFONT_NONE);
530 			break;
531 		case (DECO_PREVIOUS):
532 			term_fontlast(p);
533 			break;
534 		default:
535 			break;
536 		}
537 
538 		if (DECO_NOSPACE == deco && '\0' == *word)
539 			p->flags |= TERMP_NOSPACE;
540 	}
541 
542 	/*
543 	 * Note that we don't process the pipe: the parser sees it as
544 	 * punctuation, but we don't in terms of typography.
545 	 */
546 	if (sv[0] && '\0' == sv[1])
547 		switch (sv[0]) {
548 		case('('):
549 			/* FALLTHROUGH */
550 		case('['):
551 			p->flags |= TERMP_NOSPACE;
552 			break;
553 		default:
554 			break;
555 		}
556 }
557 
558 
559 static void
560 adjbuf(struct termp *p, size_t sz)
561 {
562 
563 	if (0 == p->maxcols)
564 		p->maxcols = 1024;
565 	while (sz >= p->maxcols)
566 		p->maxcols <<= 2;
567 
568 	p->buf = realloc(p->buf, p->maxcols);
569 	if (NULL == p->buf) {
570 		perror(NULL);
571 		exit((int)MANDOCLEVEL_SYSERR);
572 	}
573 }
574 
575 
576 static void
577 bufferc(struct termp *p, char c)
578 {
579 
580 	if (p->col + 1 >= p->maxcols)
581 		adjbuf(p, p->col + 1);
582 
583 	p->buf[(int)p->col++] = c;
584 }
585 
586 
587 static void
588 encode(struct termp *p, const char *word, size_t sz)
589 {
590 	enum termfont	  f;
591 	int		  i;
592 
593 	/*
594 	 * Encode and buffer a string of characters.  If the current
595 	 * font mode is unset, buffer directly, else encode then buffer
596 	 * character by character.
597 	 */
598 
599 	if (TERMFONT_NONE == (f = term_fonttop(p))) {
600 		if (p->col + sz >= p->maxcols)
601 			adjbuf(p, p->col + sz);
602 		memcpy(&p->buf[(int)p->col], word, sz);
603 		p->col += sz;
604 		return;
605 	}
606 
607 	/* Pre-buffer, assuming worst-case. */
608 
609 	if (p->col + 1 + (sz * 3) >= p->maxcols)
610 		adjbuf(p, p->col + 1 + (sz * 3));
611 
612 	for (i = 0; i < (int)sz; i++) {
613 		if ( ! isgraph((u_char)word[i])) {
614 			p->buf[(int)p->col++] = word[i];
615 			continue;
616 		}
617 
618 		if (TERMFONT_UNDER == f)
619 			p->buf[(int)p->col++] = '_';
620 		else
621 			p->buf[(int)p->col++] = word[i];
622 
623 		p->buf[(int)p->col++] = 8;
624 		p->buf[(int)p->col++] = word[i];
625 	}
626 }
627 
628 
629 size_t
630 term_len(const struct termp *p, size_t sz)
631 {
632 
633 	return((*p->width)(p, ' ') * sz);
634 }
635 
636 
637 size_t
638 term_strlen(const struct termp *p, const char *cp)
639 {
640 	size_t		 sz, ssz, rsz, i;
641 	enum roffdeco	 d;
642 	const char	*seq, *rhs;
643 
644 	for (sz = 0; '\0' != *cp; )
645 		/*
646 		 * Account for escaped sequences within string length
647 		 * calculations.  This follows the logic in term_word()
648 		 * as we must calculate the width of produced strings.
649 		 */
650 		if ('\\' == *cp) {
651 			seq = ++cp;
652 			cp += a2roffdeco(&d, &seq, &ssz);
653 
654 			switch (d) {
655 			case (DECO_RESERVED):
656 				rhs = chars_res2str
657 					(p->symtab, seq, ssz, &rsz);
658 				break;
659 			case (DECO_SPECIAL):
660 				/* FALLTHROUGH */
661 			case (DECO_SSPECIAL):
662 				rhs = chars_spec2str
663 					(p->symtab, seq, ssz, &rsz);
664 
665 				/* Allow for one-char escapes. */
666 				if (DECO_SSPECIAL != d || rhs)
667 					break;
668 
669 				rhs = seq;
670 				rsz = ssz;
671 				break;
672 			default:
673 				rhs = NULL;
674 				break;
675 			}
676 
677 			if (rhs)
678 				for (i = 0; i < rsz; i++)
679 					sz += (*p->width)(p, *rhs++);
680 		} else if (ASCII_NBRSP == *cp) {
681 			sz += (*p->width)(p, ' ');
682 			cp++;
683 		} else if (ASCII_HYPH == *cp) {
684 			sz += (*p->width)(p, '-');
685 			cp++;
686 		} else
687 			sz += (*p->width)(p, *cp++);
688 
689 	return(sz);
690 }
691 
692 
693 /* ARGSUSED */
694 size_t
695 term_vspan(const struct termp *p, const struct roffsu *su)
696 {
697 	double		 r;
698 
699 	switch (su->unit) {
700 	case (SCALE_CM):
701 		r = su->scale * 2;
702 		break;
703 	case (SCALE_IN):
704 		r = su->scale * 6;
705 		break;
706 	case (SCALE_PC):
707 		r = su->scale;
708 		break;
709 	case (SCALE_PT):
710 		r = su->scale / 8;
711 		break;
712 	case (SCALE_MM):
713 		r = su->scale / 1000;
714 		break;
715 	case (SCALE_VS):
716 		r = su->scale;
717 		break;
718 	default:
719 		r = su->scale - 1;
720 		break;
721 	}
722 
723 	if (r < 0.0)
724 		r = 0.0;
725 	return(/* LINTED */(size_t)
726 			r);
727 }
728 
729 
730 size_t
731 term_hspan(const struct termp *p, const struct roffsu *su)
732 {
733 	double		 v;
734 
735 	v = ((*p->hspan)(p, su));
736 	if (v < 0.0)
737 		v = 0.0;
738 	return((size_t) /* LINTED */
739 			v);
740 }
741