xref: /dragonfly/contrib/mdocml/term.c (revision 25a2db75) 
1 /*	$Id: term.c,v 1.201 2011/09/21 09:57:13 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010, 2011 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	void		 adjbuf(struct termp *p, int);
37 static	void		 bufferc(struct termp *, char);
38 static	void		 encode(struct termp *, const char *, size_t);
39 static	void		 encode1(struct termp *, int);
40 
41 void
42 term_free(struct termp *p)
43 {
44 
45 	if (p->buf)
46 		free(p->buf);
47 	if (p->symtab)
48 		mchars_free(p->symtab);
49 
50 	free(p);
51 }
52 
53 
54 void
55 term_begin(struct termp *p, term_margin head,
56 		term_margin foot, const void *arg)
57 {
58 
59 	p->headf = head;
60 	p->footf = foot;
61 	p->argf = arg;
62 	(*p->begin)(p);
63 }
64 
65 
66 void
67 term_end(struct termp *p)
68 {
69 
70 	(*p->end)(p);
71 }
72 
73 /*
74  * Flush a line of text.  A "line" is loosely defined as being something
75  * that should be followed by a newline, regardless of whether it's
76  * broken apart by newlines getting there.  A line can also be a
77  * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
78  * not have a trailing newline.
79  *
80  * The following flags may be specified:
81  *
82  *  - TERMP_NOBREAK: this is the most important and is used when making
83  *    columns.  In short: don't print a newline and instead expect the
84  *    next call to do the padding up to the start of the next column.
85  *
86  *  - TERMP_TWOSPACE: make sure there is room for at least two space
87  *    characters of padding.  Otherwise, rather break the line.
88  *
89  *  - TERMP_DANGLE: don't newline when TERMP_NOBREAK is specified and
90  *    the line is overrun, and don't pad-right if it's underrun.
91  *
92  *  - TERMP_HANG: like TERMP_DANGLE, but doesn't newline when
93  *    overrunning, instead save the position and continue at that point
94  *    when the next invocation.
95  *
96  *  In-line line breaking:
97  *
98  *  If TERMP_NOBREAK is specified and the line overruns the right
99  *  margin, it will break and pad-right to the right margin after
100  *  writing.  If maxrmargin is violated, it will break and continue
101  *  writing from the right-margin, which will lead to the above scenario
102  *  upon exit.  Otherwise, the line will break at the right margin.
103  */
104 void
105 term_flushln(struct termp *p)
106 {
107 	int		 i;     /* current input position in p->buf */
108 	size_t		 vis;   /* current visual position on output */
109 	size_t		 vbl;   /* number of blanks to prepend to output */
110 	size_t		 vend;	/* end of word visual position on output */
111 	size_t		 bp;    /* visual right border position */
112 	size_t		 dv;    /* temporary for visual pos calculations */
113 	int		 j;     /* temporary loop index for p->buf */
114 	int		 jhy;	/* last hyph before overflow w/r/t j */
115 	size_t		 maxvis; /* output position of visible boundary */
116 	size_t		 mmax; /* used in calculating bp */
117 
118 	/*
119 	 * First, establish the maximum columns of "visible" content.
120 	 * This is usually the difference between the right-margin and
121 	 * an indentation, but can be, for tagged lists or columns, a
122 	 * small set of values.
123 	 */
124 	assert  (p->rmargin >= p->offset);
125 	dv     = p->rmargin - p->offset;
126 	maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
127 	dv     = p->maxrmargin - p->offset;
128 	mmax   = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
129 
130 	bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
131 
132 	/*
133 	 * Calculate the required amount of padding.
134 	 */
135 	vbl = p->offset + p->overstep > p->viscol ?
136 	      p->offset + p->overstep - p->viscol : 0;
137 
138 	vis = vend = 0;
139 	i = 0;
140 
141 	while (i < p->col) {
142 		/*
143 		 * Handle literal tab characters: collapse all
144 		 * subsequent tabs into a single huge set of spaces.
145 		 */
146 		while (i < p->col && '\t' == p->buf[i]) {
147 			vend = (vis / p->tabwidth + 1) * p->tabwidth;
148 			vbl += vend - vis;
149 			vis = vend;
150 			i++;
151 		}
152 
153 		/*
154 		 * Count up visible word characters.  Control sequences
155 		 * (starting with the CSI) aren't counted.  A space
156 		 * generates a non-printing word, which is valid (the
157 		 * space is printed according to regular spacing rules).
158 		 */
159 
160 		for (j = i, jhy = 0; j < p->col; j++) {
161 			if ((j && ' ' == p->buf[j]) || '\t' == p->buf[j])
162 				break;
163 
164 			/* Back over the the last printed character. */
165 			if (8 == p->buf[j]) {
166 				assert(j);
167 				vend -= (*p->width)(p, p->buf[j - 1]);
168 				continue;
169 			}
170 
171 			/* Regular word. */
172 			/* Break at the hyphen point if we overrun. */
173 			if (vend > vis && vend < bp &&
174 					ASCII_HYPH == p->buf[j])
175 				jhy = j;
176 
177 			vend += (*p->width)(p, p->buf[j]);
178 		}
179 
180 		/*
181 		 * Find out whether we would exceed the right margin.
182 		 * If so, break to the next line.
183 		 */
184 		if (vend > bp && 0 == jhy && vis > 0) {
185 			vend -= vis;
186 			(*p->endline)(p);
187 			p->viscol = 0;
188 			if (TERMP_NOBREAK & p->flags) {
189 				vbl = p->rmargin;
190 				vend += p->rmargin - p->offset;
191 			} else
192 				vbl = p->offset;
193 
194 			/* Remove the p->overstep width. */
195 
196 			bp += (size_t)p->overstep;
197 			p->overstep = 0;
198 		}
199 
200 		/* Write out the [remaining] word. */
201 		for ( ; i < p->col; i++) {
202 			if (vend > bp && jhy > 0 && i > jhy)
203 				break;
204 			if ('\t' == p->buf[i])
205 				break;
206 			if (' ' == p->buf[i]) {
207 				j = i;
208 				while (' ' == p->buf[i])
209 					i++;
210 				dv = (size_t)(i - j) * (*p->width)(p, ' ');
211 				vbl += dv;
212 				vend += dv;
213 				break;
214 			}
215 			if (ASCII_NBRSP == p->buf[i]) {
216 				vbl += (*p->width)(p, ' ');
217 				continue;
218 			}
219 
220 			/*
221 			 * Now we definitely know there will be
222 			 * printable characters to output,
223 			 * so write preceding white space now.
224 			 */
225 			if (vbl) {
226 				(*p->advance)(p, vbl);
227 				p->viscol += vbl;
228 				vbl = 0;
229 			}
230 
231 			if (ASCII_HYPH == p->buf[i]) {
232 				(*p->letter)(p, '-');
233 				p->viscol += (*p->width)(p, '-');
234 				continue;
235 			}
236 
237 			(*p->letter)(p, p->buf[i]);
238 			if (8 == p->buf[i])
239 				p->viscol -= (*p->width)(p, p->buf[i-1]);
240 			else
241 				p->viscol += (*p->width)(p, p->buf[i]);
242 		}
243 		vis = vend;
244 	}
245 
246 	/*
247 	 * If there was trailing white space, it was not printed;
248 	 * so reset the cursor position accordingly.
249 	 */
250 	if (vis)
251 		vis -= vbl;
252 
253 	p->col = 0;
254 	p->overstep = 0;
255 
256 	if ( ! (TERMP_NOBREAK & p->flags)) {
257 		p->viscol = 0;
258 		(*p->endline)(p);
259 		return;
260 	}
261 
262 	if (TERMP_HANG & p->flags) {
263 		/* We need one blank after the tag. */
264 		p->overstep = (int)(vis - maxvis + (*p->width)(p, ' '));
265 
266 		/*
267 		 * Behave exactly the same way as groff:
268 		 * If we have overstepped the margin, temporarily move
269 		 * it to the right and flag the rest of the line to be
270 		 * shorter.
271 		 * If we landed right at the margin, be happy.
272 		 * If we are one step before the margin, temporarily
273 		 * move it one step LEFT and flag the rest of the line
274 		 * to be longer.
275 		 */
276 		if (p->overstep < -1)
277 			p->overstep = 0;
278 		return;
279 
280 	} else if (TERMP_DANGLE & p->flags)
281 		return;
282 
283 	/* If the column was overrun, break the line. */
284 	if (maxvis <= vis +
285 	    ((TERMP_TWOSPACE & p->flags) ? (*p->width)(p, ' ') : 0)) {
286 		(*p->endline)(p);
287 		p->viscol = 0;
288 	}
289 }
290 
291 
292 /*
293  * A newline only breaks an existing line; it won't assert vertical
294  * space.  All data in the output buffer is flushed prior to the newline
295  * assertion.
296  */
297 void
298 term_newln(struct termp *p)
299 {
300 
301 	p->flags |= TERMP_NOSPACE;
302 	if (p->col || p->viscol)
303 		term_flushln(p);
304 }
305 
306 
307 /*
308  * Asserts a vertical space (a full, empty line-break between lines).
309  * Note that if used twice, this will cause two blank spaces and so on.
310  * All data in the output buffer is flushed prior to the newline
311  * assertion.
312  */
313 void
314 term_vspace(struct termp *p)
315 {
316 
317 	term_newln(p);
318 	p->viscol = 0;
319 	(*p->endline)(p);
320 }
321 
322 void
323 term_fontlast(struct termp *p)
324 {
325 	enum termfont	 f;
326 
327 	f = p->fontl;
328 	p->fontl = p->fontq[p->fonti];
329 	p->fontq[p->fonti] = f;
330 }
331 
332 
333 void
334 term_fontrepl(struct termp *p, enum termfont f)
335 {
336 
337 	p->fontl = p->fontq[p->fonti];
338 	p->fontq[p->fonti] = f;
339 }
340 
341 
342 void
343 term_fontpush(struct termp *p, enum termfont f)
344 {
345 
346 	assert(p->fonti + 1 < 10);
347 	p->fontl = p->fontq[p->fonti];
348 	p->fontq[++p->fonti] = f;
349 }
350 
351 
352 const void *
353 term_fontq(struct termp *p)
354 {
355 
356 	return(&p->fontq[p->fonti]);
357 }
358 
359 
360 enum termfont
361 term_fonttop(struct termp *p)
362 {
363 
364 	return(p->fontq[p->fonti]);
365 }
366 
367 
368 void
369 term_fontpopq(struct termp *p, const void *key)
370 {
371 
372 	while (p->fonti >= 0 && key != &p->fontq[p->fonti])
373 		p->fonti--;
374 	assert(p->fonti >= 0);
375 }
376 
377 
378 void
379 term_fontpop(struct termp *p)
380 {
381 
382 	assert(p->fonti);
383 	p->fonti--;
384 }
385 
386 /*
387  * Handle pwords, partial words, which may be either a single word or a
388  * phrase that cannot be broken down (such as a literal string).  This
389  * handles word styling.
390  */
391 void
392 term_word(struct termp *p, const char *word)
393 {
394 	const char	*seq, *cp;
395 	char		 c;
396 	int		 sz, uc;
397 	size_t		 ssz;
398 	enum mandoc_esc	 esc;
399 
400 	if ( ! (TERMP_NOSPACE & p->flags)) {
401 		if ( ! (TERMP_KEEP & p->flags)) {
402 			if (TERMP_PREKEEP & p->flags)
403 				p->flags |= TERMP_KEEP;
404 			bufferc(p, ' ');
405 			if (TERMP_SENTENCE & p->flags)
406 				bufferc(p, ' ');
407 		} else
408 			bufferc(p, ASCII_NBRSP);
409 	}
410 
411 	if ( ! (p->flags & TERMP_NONOSPACE))
412 		p->flags &= ~TERMP_NOSPACE;
413 	else
414 		p->flags |= TERMP_NOSPACE;
415 
416 	p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
417 
418 	while ('\0' != *word) {
419 		if ((ssz = strcspn(word, "\\")) > 0)
420 			encode(p, word, ssz);
421 
422 		word += (int)ssz;
423 		if ('\\' != *word)
424 			continue;
425 
426 		word++;
427 		esc = mandoc_escape(&word, &seq, &sz);
428 		if (ESCAPE_ERROR == esc)
429 			break;
430 
431 		if (TERMENC_ASCII != p->enc)
432 			switch (esc) {
433 			case (ESCAPE_UNICODE):
434 				uc = mchars_num2uc(seq + 1, sz - 1);
435 				if ('\0' == uc)
436 					break;
437 				encode1(p, uc);
438 				continue;
439 			case (ESCAPE_SPECIAL):
440 				uc = mchars_spec2cp(p->symtab, seq, sz);
441 				if (uc <= 0)
442 					break;
443 				encode1(p, uc);
444 				continue;
445 			default:
446 				break;
447 			}
448 
449 		switch (esc) {
450 		case (ESCAPE_UNICODE):
451 			encode1(p, '?');
452 			break;
453 		case (ESCAPE_NUMBERED):
454 			c = mchars_num2char(seq, sz);
455 			if ('\0' != c)
456 				encode(p, &c, 1);
457 			break;
458 		case (ESCAPE_SPECIAL):
459 			cp = mchars_spec2str(p->symtab, seq, sz, &ssz);
460 			if (NULL != cp)
461 				encode(p, cp, ssz);
462 			else if (1 == ssz)
463 				encode(p, seq, sz);
464 			break;
465 		case (ESCAPE_FONTBOLD):
466 			term_fontrepl(p, TERMFONT_BOLD);
467 			break;
468 		case (ESCAPE_FONTITALIC):
469 			term_fontrepl(p, TERMFONT_UNDER);
470 			break;
471 		case (ESCAPE_FONT):
472 			/* FALLTHROUGH */
473 		case (ESCAPE_FONTROMAN):
474 			term_fontrepl(p, TERMFONT_NONE);
475 			break;
476 		case (ESCAPE_FONTPREV):
477 			term_fontlast(p);
478 			break;
479 		case (ESCAPE_NOSPACE):
480 			if ('\0' == *word)
481 				p->flags |= TERMP_NOSPACE;
482 			break;
483 		default:
484 			break;
485 		}
486 	}
487 }
488 
489 static void
490 adjbuf(struct termp *p, int sz)
491 {
492 
493 	if (0 == p->maxcols)
494 		p->maxcols = 1024;
495 	while (sz >= p->maxcols)
496 		p->maxcols <<= 2;
497 
498 	p->buf = mandoc_realloc
499 		(p->buf, sizeof(int) * (size_t)p->maxcols);
500 }
501 
502 static void
503 bufferc(struct termp *p, char c)
504 {
505 
506 	if (p->col + 1 >= p->maxcols)
507 		adjbuf(p, p->col + 1);
508 
509 	p->buf[p->col++] = c;
510 }
511 
512 /*
513  * See encode().
514  * Do this for a single (probably unicode) value.
515  * Does not check for non-decorated glyphs.
516  */
517 static void
518 encode1(struct termp *p, int c)
519 {
520 	enum termfont	  f;
521 
522 	if (p->col + 4 >= p->maxcols)
523 		adjbuf(p, p->col + 4);
524 
525 	f = term_fonttop(p);
526 
527 	if (TERMFONT_NONE == f) {
528 		p->buf[p->col++] = c;
529 		return;
530 	} else if (TERMFONT_UNDER == f) {
531 		p->buf[p->col++] = '_';
532 	} else
533 		p->buf[p->col++] = c;
534 
535 	p->buf[p->col++] = 8;
536 	p->buf[p->col++] = c;
537 }
538 
539 static void
540 encode(struct termp *p, const char *word, size_t sz)
541 {
542 	enum termfont	  f;
543 	int		  i, len;
544 
545 	/* LINTED */
546 	len = sz;
547 
548 	/*
549 	 * Encode and buffer a string of characters.  If the current
550 	 * font mode is unset, buffer directly, else encode then buffer
551 	 * character by character.
552 	 */
553 
554 	if (TERMFONT_NONE == (f = term_fonttop(p))) {
555 		if (p->col + len >= p->maxcols)
556 			adjbuf(p, p->col + len);
557 		for (i = 0; i < len; i++)
558 			p->buf[p->col++] = word[i];
559 		return;
560 	}
561 
562 	/* Pre-buffer, assuming worst-case. */
563 
564 	if (p->col + 1 + (len * 3) >= p->maxcols)
565 		adjbuf(p, p->col + 1 + (len * 3));
566 
567 	for (i = 0; i < len; i++) {
568 		if (ASCII_HYPH != word[i] &&
569 		    ! isgraph((unsigned char)word[i])) {
570 			p->buf[p->col++] = word[i];
571 			continue;
572 		}
573 
574 		if (TERMFONT_UNDER == f)
575 			p->buf[p->col++] = '_';
576 		else if (ASCII_HYPH == word[i])
577 			p->buf[p->col++] = '-';
578 		else
579 			p->buf[p->col++] = word[i];
580 
581 		p->buf[p->col++] = 8;
582 		p->buf[p->col++] = word[i];
583 	}
584 }
585 
586 size_t
587 term_len(const struct termp *p, size_t sz)
588 {
589 
590 	return((*p->width)(p, ' ') * sz);
591 }
592 
593 
594 size_t
595 term_strlen(const struct termp *p, const char *cp)
596 {
597 	size_t		 sz, rsz, i;
598 	int		 ssz, c;
599 	const char	*seq, *rhs;
600 	enum mandoc_esc	 esc;
601 	static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' };
602 
603 	/*
604 	 * Account for escaped sequences within string length
605 	 * calculations.  This follows the logic in term_word() as we
606 	 * must calculate the width of produced strings.
607 	 */
608 
609 	sz = 0;
610 	while ('\0' != *cp) {
611 		rsz = strcspn(cp, rej);
612 		for (i = 0; i < rsz; i++)
613 			sz += (*p->width)(p, *cp++);
614 
615 		c = 0;
616 		switch (*cp) {
617 		case ('\\'):
618 			cp++;
619 			esc = mandoc_escape(&cp, &seq, &ssz);
620 			if (ESCAPE_ERROR == esc)
621 				return(sz);
622 
623 			if (TERMENC_ASCII != p->enc)
624 				switch (esc) {
625 				case (ESCAPE_UNICODE):
626 					c = mchars_num2uc
627 						(seq + 1, ssz - 1);
628 					if ('\0' == c)
629 						break;
630 					sz += (*p->width)(p, c);
631 					continue;
632 				case (ESCAPE_SPECIAL):
633 					c = mchars_spec2cp
634 						(p->symtab, seq, ssz);
635 					if (c <= 0)
636 						break;
637 					sz += (*p->width)(p, c);
638 					continue;
639 				default:
640 					break;
641 				}
642 
643 			rhs = NULL;
644 
645 			switch (esc) {
646 			case (ESCAPE_UNICODE):
647 				sz += (*p->width)(p, '?');
648 				break;
649 			case (ESCAPE_NUMBERED):
650 				c = mchars_num2char(seq, ssz);
651 				if ('\0' != c)
652 					sz += (*p->width)(p, c);
653 				break;
654 			case (ESCAPE_SPECIAL):
655 				rhs = mchars_spec2str
656 					(p->symtab, seq, ssz, &rsz);
657 
658 				if (ssz != 1 || rhs)
659 					break;
660 
661 				rhs = seq;
662 				rsz = ssz;
663 				break;
664 			default:
665 				break;
666 			}
667 
668 			if (NULL == rhs)
669 				break;
670 
671 			for (i = 0; i < rsz; i++)
672 				sz += (*p->width)(p, *rhs++);
673 			break;
674 		case (ASCII_NBRSP):
675 			sz += (*p->width)(p, ' ');
676 			cp++;
677 			break;
678 		case (ASCII_HYPH):
679 			sz += (*p->width)(p, '-');
680 			cp++;
681 			break;
682 		default:
683 			break;
684 		}
685 	}
686 
687 	return(sz);
688 }
689 
690 /* ARGSUSED */
691 size_t
692 term_vspan(const struct termp *p, const struct roffsu *su)
693 {
694 	double		 r;
695 
696 	switch (su->unit) {
697 	case (SCALE_CM):
698 		r = su->scale * 2;
699 		break;
700 	case (SCALE_IN):
701 		r = su->scale * 6;
702 		break;
703 	case (SCALE_PC):
704 		r = su->scale;
705 		break;
706 	case (SCALE_PT):
707 		r = su->scale / 8;
708 		break;
709 	case (SCALE_MM):
710 		r = su->scale / 1000;
711 		break;
712 	case (SCALE_VS):
713 		r = su->scale;
714 		break;
715 	default:
716 		r = su->scale - 1;
717 		break;
718 	}
719 
720 	if (r < 0.0)
721 		r = 0.0;
722 	return(/* LINTED */(size_t)
723 			r);
724 }
725 
726 size_t
727 term_hspan(const struct termp *p, const struct roffsu *su)
728 {
729 	double		 v;
730 
731 	v = ((*p->hspan)(p, su));
732 	if (v < 0.0)
733 		v = 0.0;
734 	return((size_t) /* LINTED */
735 			v);
736 }
737