xref: /illumos-gate/usr/src/cmd/mandoc/term.c (revision 2411ecfcc7e4d3dc7fc3f78b9e50601e50a8a56c)
1 /*	$Id: term.c,v 1.274 2017/07/28 14:25:48 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010-2017 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 AUTHORS DISCLAIM ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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_col *, 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 static	void		 endline(struct termp *);
40 
41 
42 void
43 term_setcol(struct termp *p, size_t maxtcol)
44 {
45 	if (maxtcol > p->maxtcol) {
46 		p->tcols = mandoc_recallocarray(p->tcols,
47 		    p->maxtcol, maxtcol, sizeof(*p->tcols));
48 		p->maxtcol = maxtcol;
49 	}
50 	p->lasttcol = maxtcol - 1;
51 	p->tcol = p->tcols;
52 }
53 
54 void
55 term_free(struct termp *p)
56 {
57 	for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++)
58 		free(p->tcol->buf);
59 	free(p->tcols);
60 	free(p->fontq);
61 	free(p);
62 }
63 
64 void
65 term_begin(struct termp *p, term_margin head,
66 		term_margin foot, const struct roff_meta *arg)
67 {
68 
69 	p->headf = head;
70 	p->footf = foot;
71 	p->argf = arg;
72 	(*p->begin)(p);
73 }
74 
75 void
76 term_end(struct termp *p)
77 {
78 
79 	(*p->end)(p);
80 }
81 
82 /*
83  * Flush a chunk of text.  By default, break the output line each time
84  * the right margin is reached, and continue output on the next line
85  * at the same offset as the chunk itself.  By default, also break the
86  * output line at the end of the chunk.
87  * The following flags may be specified:
88  *
89  *  - TERMP_NOBREAK: Do not break the output line at the right margin,
90  *    but only at the max right margin.  Also, do not break the output
91  *    line at the end of the chunk, such that the next call can pad to
92  *    the next column.  However, if less than p->trailspace blanks,
93  *    which can be 0, 1, or 2, remain to the right margin, the line
94  *    will be broken.
95  *  - TERMP_BRTRSP: Consider trailing whitespace significant
96  *    when deciding whether the chunk fits or not.
97  *  - TERMP_BRIND: If the chunk does not fit and the output line has
98  *    to be broken, start the next line at the right margin instead
99  *    of at the offset.  Used together with TERMP_NOBREAK for the tags
100  *    in various kinds of tagged lists.
101  *  - TERMP_HANG: Do not break the output line at the right margin,
102  *    append the next chunk after it even if this one is too long.
103  *    To be used together with TERMP_NOBREAK.
104  *  - TERMP_NOPAD: Start writing at the current position,
105  *    do not pad with blank characters up to the offset.
106  */
107 void
108 term_flushln(struct termp *p)
109 {
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 	size_t		 j;     /* temporary loop index for p->tcol->buf */
116 	size_t		 jhy;	/* last hyph before overflow w/r/t j */
117 	size_t		 maxvis; /* output position of visible boundary */
118 	int		 ntab;	/* number of tabs to prepend */
119 	int		 breakline; /* after this word */
120 
121 	vbl = (p->flags & TERMP_NOPAD) || p->tcol->offset < p->viscol ?
122 	    0 : p->tcol->offset - p->viscol;
123 	if (p->minbl && vbl < p->minbl)
124 		vbl = p->minbl;
125 	maxvis = p->tcol->rmargin > p->viscol + vbl ?
126 	    p->tcol->rmargin - p->viscol - vbl : 0;
127 	bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
128 	    p->maxrmargin > p->viscol + vbl ?
129 	    p->maxrmargin - p->viscol - vbl : 0;
130 	vis = vend = 0;
131 
132 	if ((p->flags & TERMP_MULTICOL) == 0)
133 		p->tcol->col = 0;
134 	while (p->tcol->col < p->tcol->lastcol) {
135 
136 		/*
137 		 * Handle literal tab characters: collapse all
138 		 * subsequent tabs into a single huge set of spaces.
139 		 */
140 
141 		ntab = 0;
142 		while (p->tcol->col < p->tcol->lastcol &&
143 		    p->tcol->buf[p->tcol->col] == '\t') {
144 			vend = term_tab_next(vis);
145 			vbl += vend - vis;
146 			vis = vend;
147 			ntab++;
148 			p->tcol->col++;
149 		}
150 
151 		/*
152 		 * Count up visible word characters.  Control sequences
153 		 * (starting with the CSI) aren't counted.  A space
154 		 * generates a non-printing word, which is valid (the
155 		 * space is printed according to regular spacing rules).
156 		 */
157 
158 		jhy = 0;
159 		breakline = 0;
160 		for (j = p->tcol->col; j < p->tcol->lastcol; j++) {
161 			if (p->tcol->buf[j] == '\n') {
162 				if ((p->flags & TERMP_BRIND) == 0)
163 					breakline = 1;
164 				continue;
165 			}
166 			if (p->tcol->buf[j] == ' ' || p->tcol->buf[j] == '\t')
167 				break;
168 
169 			/* Back over the last printed character. */
170 			if (p->tcol->buf[j] == '\b') {
171 				assert(j);
172 				vend -= (*p->width)(p, p->tcol->buf[j - 1]);
173 				continue;
174 			}
175 
176 			/* Regular word. */
177 			/* Break at the hyphen point if we overrun. */
178 			if (vend > vis && vend < bp &&
179 			    (p->tcol->buf[j] == ASCII_HYPH||
180 			     p->tcol->buf[j] == ASCII_BREAK))
181 				jhy = j;
182 
183 			/*
184 			 * Hyphenation now decided, put back a real
185 			 * hyphen such that we get the correct width.
186 			 */
187 			if (p->tcol->buf[j] == ASCII_HYPH)
188 				p->tcol->buf[j] = '-';
189 
190 			vend += (*p->width)(p, p->tcol->buf[j]);
191 		}
192 
193 		/*
194 		 * Find out whether we would exceed the right margin.
195 		 * If so, break to the next line.
196 		 */
197 
198 		if (vend > bp && jhy == 0 && vis > 0 &&
199 		    (p->flags & TERMP_BRNEVER) == 0) {
200 			if (p->flags & TERMP_MULTICOL)
201 				return;
202 
203 			endline(p);
204 			vend -= vis;
205 
206 			/* Use pending tabs on the new line. */
207 
208 			vbl = 0;
209 			while (ntab--)
210 				vbl = term_tab_next(vbl);
211 
212 			/* Re-establish indentation. */
213 
214 			if (p->flags & TERMP_BRIND)
215 				vbl += p->tcol->rmargin;
216 			else
217 				vbl += p->tcol->offset;
218 			maxvis = p->tcol->rmargin > vbl ?
219 			    p->tcol->rmargin - vbl : 0;
220 			bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
221 			    p->maxrmargin > vbl ?  p->maxrmargin - vbl : 0;
222 		}
223 
224 		/*
225 		 * Write out the rest of the word.
226 		 */
227 
228 		for ( ; p->tcol->col < p->tcol->lastcol; p->tcol->col++) {
229 			if (vend > bp && jhy > 0 && p->tcol->col > jhy)
230 				break;
231 			if (p->tcol->buf[p->tcol->col] == '\n')
232 				continue;
233 			if (p->tcol->buf[p->tcol->col] == '\t')
234 				break;
235 			if (p->tcol->buf[p->tcol->col] == ' ') {
236 				j = p->tcol->col;
237 				while (p->tcol->col < p->tcol->lastcol &&
238 				    p->tcol->buf[p->tcol->col] == ' ')
239 					p->tcol->col++;
240 				dv = (p->tcol->col - j) * (*p->width)(p, ' ');
241 				vbl += dv;
242 				vend += dv;
243 				break;
244 			}
245 			if (p->tcol->buf[p->tcol->col] == ASCII_NBRSP) {
246 				vbl += (*p->width)(p, ' ');
247 				continue;
248 			}
249 			if (p->tcol->buf[p->tcol->col] == ASCII_BREAK)
250 				continue;
251 
252 			/*
253 			 * Now we definitely know there will be
254 			 * printable characters to output,
255 			 * so write preceding white space now.
256 			 */
257 			if (vbl) {
258 				(*p->advance)(p, vbl);
259 				p->viscol += vbl;
260 				vbl = 0;
261 			}
262 
263 			(*p->letter)(p, p->tcol->buf[p->tcol->col]);
264 			if (p->tcol->buf[p->tcol->col] == '\b')
265 				p->viscol -= (*p->width)(p,
266 				    p->tcol->buf[p->tcol->col - 1]);
267 			else
268 				p->viscol += (*p->width)(p,
269 				    p->tcol->buf[p->tcol->col]);
270 		}
271 		vis = vend;
272 
273 		if (breakline == 0)
274 			continue;
275 
276 		/* Explicitly requested output line break. */
277 
278 		if (p->flags & TERMP_MULTICOL)
279 			return;
280 
281 		endline(p);
282 		breakline = 0;
283 		vis = vend = 0;
284 
285 		/* Re-establish indentation. */
286 
287 		vbl = p->tcol->offset;
288 		maxvis = p->tcol->rmargin > vbl ?
289 		    p->tcol->rmargin - vbl : 0;
290 		bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
291 		    p->maxrmargin > vbl ?  p->maxrmargin - vbl : 0;
292 	}
293 
294 	/*
295 	 * If there was trailing white space, it was not printed;
296 	 * so reset the cursor position accordingly.
297 	 */
298 
299 	if (vis > vbl)
300 		vis -= vbl;
301 	else
302 		vis = 0;
303 
304 	p->col = p->tcol->col = p->tcol->lastcol = 0;
305 	p->minbl = p->trailspace;
306 	p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
307 
308 	if (p->flags & TERMP_MULTICOL)
309 		return;
310 
311 	/* Trailing whitespace is significant in some columns. */
312 
313 	if (vis && vbl && (TERMP_BRTRSP & p->flags))
314 		vis += vbl;
315 
316 	/* If the column was overrun, break the line. */
317 	if ((p->flags & TERMP_NOBREAK) == 0 ||
318 	    ((p->flags & TERMP_HANG) == 0 &&
319 	     vis + p->trailspace * (*p->width)(p, ' ') > maxvis))
320 		endline(p);
321 }
322 
323 static void
324 endline(struct termp *p)
325 {
326 	if ((p->flags & (TERMP_NEWMC | TERMP_ENDMC)) == TERMP_ENDMC) {
327 		p->mc = NULL;
328 		p->flags &= ~TERMP_ENDMC;
329 	}
330 	if (p->mc != NULL) {
331 		if (p->viscol && p->maxrmargin >= p->viscol)
332 			(*p->advance)(p, p->maxrmargin - p->viscol + 1);
333 		p->flags |= TERMP_NOBUF | TERMP_NOSPACE;
334 		term_word(p, p->mc);
335 		p->flags &= ~(TERMP_NOBUF | TERMP_NEWMC);
336 	}
337 	p->viscol = 0;
338 	p->minbl = 0;
339 	(*p->endline)(p);
340 }
341 
342 /*
343  * A newline only breaks an existing line; it won't assert vertical
344  * space.  All data in the output buffer is flushed prior to the newline
345  * assertion.
346  */
347 void
348 term_newln(struct termp *p)
349 {
350 
351 	p->flags |= TERMP_NOSPACE;
352 	if (p->tcol->lastcol || p->viscol)
353 		term_flushln(p);
354 }
355 
356 /*
357  * Asserts a vertical space (a full, empty line-break between lines).
358  * Note that if used twice, this will cause two blank spaces and so on.
359  * All data in the output buffer is flushed prior to the newline
360  * assertion.
361  */
362 void
363 term_vspace(struct termp *p)
364 {
365 
366 	term_newln(p);
367 	p->viscol = 0;
368 	p->minbl = 0;
369 	if (0 < p->skipvsp)
370 		p->skipvsp--;
371 	else
372 		(*p->endline)(p);
373 }
374 
375 /* Swap current and previous font; for \fP and .ft P */
376 void
377 term_fontlast(struct termp *p)
378 {
379 	enum termfont	 f;
380 
381 	f = p->fontl;
382 	p->fontl = p->fontq[p->fonti];
383 	p->fontq[p->fonti] = f;
384 }
385 
386 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
387 void
388 term_fontrepl(struct termp *p, enum termfont f)
389 {
390 
391 	p->fontl = p->fontq[p->fonti];
392 	p->fontq[p->fonti] = f;
393 }
394 
395 /* Set font, save previous. */
396 void
397 term_fontpush(struct termp *p, enum termfont f)
398 {
399 
400 	p->fontl = p->fontq[p->fonti];
401 	if (++p->fonti == p->fontsz) {
402 		p->fontsz += 8;
403 		p->fontq = mandoc_reallocarray(p->fontq,
404 		    p->fontsz, sizeof(*p->fontq));
405 	}
406 	p->fontq[p->fonti] = f;
407 }
408 
409 /* Flush to make the saved pointer current again. */
410 void
411 term_fontpopq(struct termp *p, int i)
412 {
413 
414 	assert(i >= 0);
415 	if (p->fonti > i)
416 		p->fonti = i;
417 }
418 
419 /* Pop one font off the stack. */
420 void
421 term_fontpop(struct termp *p)
422 {
423 
424 	assert(p->fonti);
425 	p->fonti--;
426 }
427 
428 /*
429  * Handle pwords, partial words, which may be either a single word or a
430  * phrase that cannot be broken down (such as a literal string).  This
431  * handles word styling.
432  */
433 void
434 term_word(struct termp *p, const char *word)
435 {
436 	struct roffsu	 su;
437 	const char	 nbrsp[2] = { ASCII_NBRSP, 0 };
438 	const char	*seq, *cp;
439 	int		 sz, uc;
440 	size_t		 csz, lsz, ssz;
441 	enum mandoc_esc	 esc;
442 
443 	if ((p->flags & TERMP_NOBUF) == 0) {
444 		if ((p->flags & TERMP_NOSPACE) == 0) {
445 			if ((p->flags & TERMP_KEEP) == 0) {
446 				bufferc(p, ' ');
447 				if (p->flags & TERMP_SENTENCE)
448 					bufferc(p, ' ');
449 			} else
450 				bufferc(p, ASCII_NBRSP);
451 		}
452 		if (p->flags & TERMP_PREKEEP)
453 			p->flags |= TERMP_KEEP;
454 		if (p->flags & TERMP_NONOSPACE)
455 			p->flags |= TERMP_NOSPACE;
456 		else
457 			p->flags &= ~TERMP_NOSPACE;
458 		p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);
459 		p->skipvsp = 0;
460 	}
461 
462 	while ('\0' != *word) {
463 		if ('\\' != *word) {
464 			if (TERMP_NBRWORD & p->flags) {
465 				if (' ' == *word) {
466 					encode(p, nbrsp, 1);
467 					word++;
468 					continue;
469 				}
470 				ssz = strcspn(word, "\\ ");
471 			} else
472 				ssz = strcspn(word, "\\");
473 			encode(p, word, ssz);
474 			word += (int)ssz;
475 			continue;
476 		}
477 
478 		word++;
479 		esc = mandoc_escape(&word, &seq, &sz);
480 		if (ESCAPE_ERROR == esc)
481 			continue;
482 
483 		switch (esc) {
484 		case ESCAPE_UNICODE:
485 			uc = mchars_num2uc(seq + 1, sz - 1);
486 			break;
487 		case ESCAPE_NUMBERED:
488 			uc = mchars_num2char(seq, sz);
489 			if (uc < 0)
490 				continue;
491 			break;
492 		case ESCAPE_SPECIAL:
493 			if (p->enc == TERMENC_ASCII) {
494 				cp = mchars_spec2str(seq, sz, &ssz);
495 				if (cp != NULL)
496 					encode(p, cp, ssz);
497 			} else {
498 				uc = mchars_spec2cp(seq, sz);
499 				if (uc > 0)
500 					encode1(p, uc);
501 			}
502 			continue;
503 		case ESCAPE_FONTBOLD:
504 			term_fontrepl(p, TERMFONT_BOLD);
505 			continue;
506 		case ESCAPE_FONTITALIC:
507 			term_fontrepl(p, TERMFONT_UNDER);
508 			continue;
509 		case ESCAPE_FONTBI:
510 			term_fontrepl(p, TERMFONT_BI);
511 			continue;
512 		case ESCAPE_FONT:
513 		case ESCAPE_FONTROMAN:
514 			term_fontrepl(p, TERMFONT_NONE);
515 			continue;
516 		case ESCAPE_FONTPREV:
517 			term_fontlast(p);
518 			continue;
519 		case ESCAPE_BREAK:
520 			bufferc(p, '\n');
521 			continue;
522 		case ESCAPE_NOSPACE:
523 			if (p->flags & TERMP_BACKAFTER)
524 				p->flags &= ~TERMP_BACKAFTER;
525 			else if (*word == '\0')
526 				p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
527 			continue;
528 		case ESCAPE_HORIZ:
529 			if (*seq == '|') {
530 				seq++;
531 				uc = -p->col;
532 			} else
533 				uc = 0;
534 			if (a2roffsu(seq, &su, SCALE_EM) == NULL)
535 				continue;
536 			uc += term_hen(p, &su);
537 			if (uc > 0)
538 				while (uc-- > 0)
539 					bufferc(p, ASCII_NBRSP);
540 			else if (p->col > (size_t)(-uc))
541 				p->col += uc;
542 			else {
543 				uc += p->col;
544 				p->col = 0;
545 				if (p->tcol->offset > (size_t)(-uc)) {
546 					p->ti += uc;
547 					p->tcol->offset += uc;
548 				} else {
549 					p->ti -= p->tcol->offset;
550 					p->tcol->offset = 0;
551 				}
552 			}
553 			continue;
554 		case ESCAPE_HLINE:
555 			if ((cp = a2roffsu(seq, &su, SCALE_EM)) == NULL)
556 				continue;
557 			uc = term_hen(p, &su);
558 			if (uc <= 0) {
559 				if (p->tcol->rmargin <= p->tcol->offset)
560 					continue;
561 				lsz = p->tcol->rmargin - p->tcol->offset;
562 			} else
563 				lsz = uc;
564 			if (*cp == seq[-1])
565 				uc = -1;
566 			else if (*cp == '\\') {
567 				seq = cp + 1;
568 				esc = mandoc_escape(&seq, &cp, &sz);
569 				switch (esc) {
570 				case ESCAPE_UNICODE:
571 					uc = mchars_num2uc(cp + 1, sz - 1);
572 					break;
573 				case ESCAPE_NUMBERED:
574 					uc = mchars_num2char(cp, sz);
575 					break;
576 				case ESCAPE_SPECIAL:
577 					uc = mchars_spec2cp(cp, sz);
578 					break;
579 				default:
580 					uc = -1;
581 					break;
582 				}
583 			} else
584 				uc = *cp;
585 			if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
586 				uc = '_';
587 			if (p->enc == TERMENC_ASCII) {
588 				cp = ascii_uc2str(uc);
589 				csz = term_strlen(p, cp);
590 				ssz = strlen(cp);
591 			} else
592 				csz = (*p->width)(p, uc);
593 			while (lsz >= csz) {
594 				if (p->enc == TERMENC_ASCII)
595 					encode(p, cp, ssz);
596 				else
597 					encode1(p, uc);
598 				lsz -= csz;
599 			}
600 			continue;
601 		case ESCAPE_SKIPCHAR:
602 			p->flags |= TERMP_BACKAFTER;
603 			continue;
604 		case ESCAPE_OVERSTRIKE:
605 			cp = seq + sz;
606 			while (seq < cp) {
607 				if (*seq == '\\') {
608 					mandoc_escape(&seq, NULL, NULL);
609 					continue;
610 				}
611 				encode1(p, *seq++);
612 				if (seq < cp) {
613 					if (p->flags & TERMP_BACKBEFORE)
614 						p->flags |= TERMP_BACKAFTER;
615 					else
616 						p->flags |= TERMP_BACKBEFORE;
617 				}
618 			}
619 			/* Trim trailing backspace/blank pair. */
620 			if (p->tcol->lastcol > 2 &&
621 			    (p->tcol->buf[p->tcol->lastcol - 1] == ' ' ||
622 			     p->tcol->buf[p->tcol->lastcol - 1] == '\t'))
623 				p->tcol->lastcol -= 2;
624 			if (p->col > p->tcol->lastcol)
625 				p->col = p->tcol->lastcol;
626 			continue;
627 		default:
628 			continue;
629 		}
630 
631 		/*
632 		 * Common handling for Unicode and numbered
633 		 * character escape sequences.
634 		 */
635 
636 		if (p->enc == TERMENC_ASCII) {
637 			cp = ascii_uc2str(uc);
638 			encode(p, cp, strlen(cp));
639 		} else {
640 			if ((uc < 0x20 && uc != 0x09) ||
641 			    (uc > 0x7E && uc < 0xA0))
642 				uc = 0xFFFD;
643 			encode1(p, uc);
644 		}
645 	}
646 	p->flags &= ~TERMP_NBRWORD;
647 }
648 
649 static void
650 adjbuf(struct termp_col *c, size_t sz)
651 {
652 	if (c->maxcols == 0)
653 		c->maxcols = 1024;
654 	while (c->maxcols <= sz)
655 		c->maxcols <<= 2;
656 	c->buf = mandoc_reallocarray(c->buf, c->maxcols, sizeof(*c->buf));
657 }
658 
659 static void
660 bufferc(struct termp *p, char c)
661 {
662 	if (p->flags & TERMP_NOBUF) {
663 		(*p->letter)(p, c);
664 		return;
665 	}
666 	if (p->col + 1 >= p->tcol->maxcols)
667 		adjbuf(p->tcol, p->col + 1);
668 	if (p->tcol->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
669 		p->tcol->buf[p->col] = c;
670 	if (p->tcol->lastcol < ++p->col)
671 		p->tcol->lastcol = p->col;
672 }
673 
674 /*
675  * See encode().
676  * Do this for a single (probably unicode) value.
677  * Does not check for non-decorated glyphs.
678  */
679 static void
680 encode1(struct termp *p, int c)
681 {
682 	enum termfont	  f;
683 
684 	if (p->flags & TERMP_NOBUF) {
685 		(*p->letter)(p, c);
686 		return;
687 	}
688 
689 	if (p->col + 7 >= p->tcol->maxcols)
690 		adjbuf(p->tcol, p->col + 7);
691 
692 	f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
693 	    p->fontq[p->fonti] : TERMFONT_NONE;
694 
695 	if (p->flags & TERMP_BACKBEFORE) {
696 		if (p->tcol->buf[p->col - 1] == ' ' ||
697 		    p->tcol->buf[p->col - 1] == '\t')
698 			p->col--;
699 		else
700 			p->tcol->buf[p->col++] = '\b';
701 		p->flags &= ~TERMP_BACKBEFORE;
702 	}
703 	if (f == TERMFONT_UNDER || f == TERMFONT_BI) {
704 		p->tcol->buf[p->col++] = '_';
705 		p->tcol->buf[p->col++] = '\b';
706 	}
707 	if (f == TERMFONT_BOLD || f == TERMFONT_BI) {
708 		if (c == ASCII_HYPH)
709 			p->tcol->buf[p->col++] = '-';
710 		else
711 			p->tcol->buf[p->col++] = c;
712 		p->tcol->buf[p->col++] = '\b';
713 	}
714 	if (p->tcol->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
715 		p->tcol->buf[p->col] = c;
716 	if (p->tcol->lastcol < ++p->col)
717 		p->tcol->lastcol = p->col;
718 	if (p->flags & TERMP_BACKAFTER) {
719 		p->flags |= TERMP_BACKBEFORE;
720 		p->flags &= ~TERMP_BACKAFTER;
721 	}
722 }
723 
724 static void
725 encode(struct termp *p, const char *word, size_t sz)
726 {
727 	size_t		  i;
728 
729 	if (p->flags & TERMP_NOBUF) {
730 		for (i = 0; i < sz; i++)
731 			(*p->letter)(p, word[i]);
732 		return;
733 	}
734 
735 	if (p->col + 2 + (sz * 5) >= p->tcol->maxcols)
736 		adjbuf(p->tcol, p->col + 2 + (sz * 5));
737 
738 	for (i = 0; i < sz; i++) {
739 		if (ASCII_HYPH == word[i] ||
740 		    isgraph((unsigned char)word[i]))
741 			encode1(p, word[i]);
742 		else {
743 			if (p->tcol->lastcol <= p->col ||
744 			    (word[i] != ' ' && word[i] != ASCII_NBRSP))
745 				p->tcol->buf[p->col] = word[i];
746 			p->col++;
747 
748 			/*
749 			 * Postpone the effect of \z while handling
750 			 * an overstrike sequence from ascii_uc2str().
751 			 */
752 
753 			if (word[i] == '\b' &&
754 			    (p->flags & TERMP_BACKBEFORE)) {
755 				p->flags &= ~TERMP_BACKBEFORE;
756 				p->flags |= TERMP_BACKAFTER;
757 			}
758 		}
759 	}
760 	if (p->tcol->lastcol < p->col)
761 		p->tcol->lastcol = p->col;
762 }
763 
764 void
765 term_setwidth(struct termp *p, const char *wstr)
766 {
767 	struct roffsu	 su;
768 	int		 iop, width;
769 
770 	iop = 0;
771 	width = 0;
772 	if (NULL != wstr) {
773 		switch (*wstr) {
774 		case '+':
775 			iop = 1;
776 			wstr++;
777 			break;
778 		case '-':
779 			iop = -1;
780 			wstr++;
781 			break;
782 		default:
783 			break;
784 		}
785 		if (a2roffsu(wstr, &su, SCALE_MAX) != NULL)
786 			width = term_hspan(p, &su);
787 		else
788 			iop = 0;
789 	}
790 	(*p->setwidth)(p, iop, width);
791 }
792 
793 size_t
794 term_len(const struct termp *p, size_t sz)
795 {
796 
797 	return (*p->width)(p, ' ') * sz;
798 }
799 
800 static size_t
801 cond_width(const struct termp *p, int c, int *skip)
802 {
803 
804 	if (*skip) {
805 		(*skip) = 0;
806 		return 0;
807 	} else
808 		return (*p->width)(p, c);
809 }
810 
811 size_t
812 term_strlen(const struct termp *p, const char *cp)
813 {
814 	size_t		 sz, rsz, i;
815 	int		 ssz, skip, uc;
816 	const char	*seq, *rhs;
817 	enum mandoc_esc	 esc;
818 	static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
819 			ASCII_BREAK, '\0' };
820 
821 	/*
822 	 * Account for escaped sequences within string length
823 	 * calculations.  This follows the logic in term_word() as we
824 	 * must calculate the width of produced strings.
825 	 */
826 
827 	sz = 0;
828 	skip = 0;
829 	while ('\0' != *cp) {
830 		rsz = strcspn(cp, rej);
831 		for (i = 0; i < rsz; i++)
832 			sz += cond_width(p, *cp++, &skip);
833 
834 		switch (*cp) {
835 		case '\\':
836 			cp++;
837 			esc = mandoc_escape(&cp, &seq, &ssz);
838 			if (ESCAPE_ERROR == esc)
839 				continue;
840 
841 			rhs = NULL;
842 
843 			switch (esc) {
844 			case ESCAPE_UNICODE:
845 				uc = mchars_num2uc(seq + 1, ssz - 1);
846 				break;
847 			case ESCAPE_NUMBERED:
848 				uc = mchars_num2char(seq, ssz);
849 				if (uc < 0)
850 					continue;
851 				break;
852 			case ESCAPE_SPECIAL:
853 				if (p->enc == TERMENC_ASCII) {
854 					rhs = mchars_spec2str(seq, ssz, &rsz);
855 					if (rhs != NULL)
856 						break;
857 				} else {
858 					uc = mchars_spec2cp(seq, ssz);
859 					if (uc > 0)
860 						sz += cond_width(p, uc, &skip);
861 				}
862 				continue;
863 			case ESCAPE_SKIPCHAR:
864 				skip = 1;
865 				continue;
866 			case ESCAPE_OVERSTRIKE:
867 				rsz = 0;
868 				rhs = seq + ssz;
869 				while (seq < rhs) {
870 					if (*seq == '\\') {
871 						mandoc_escape(&seq, NULL, NULL);
872 						continue;
873 					}
874 					i = (*p->width)(p, *seq++);
875 					if (rsz < i)
876 						rsz = i;
877 				}
878 				sz += rsz;
879 				continue;
880 			default:
881 				continue;
882 			}
883 
884 			/*
885 			 * Common handling for Unicode and numbered
886 			 * character escape sequences.
887 			 */
888 
889 			if (rhs == NULL) {
890 				if (p->enc == TERMENC_ASCII) {
891 					rhs = ascii_uc2str(uc);
892 					rsz = strlen(rhs);
893 				} else {
894 					if ((uc < 0x20 && uc != 0x09) ||
895 					    (uc > 0x7E && uc < 0xA0))
896 						uc = 0xFFFD;
897 					sz += cond_width(p, uc, &skip);
898 					continue;
899 				}
900 			}
901 
902 			if (skip) {
903 				skip = 0;
904 				break;
905 			}
906 
907 			/*
908 			 * Common handling for all escape sequences
909 			 * printing more than one character.
910 			 */
911 
912 			for (i = 0; i < rsz; i++)
913 				sz += (*p->width)(p, *rhs++);
914 			break;
915 		case ASCII_NBRSP:
916 			sz += cond_width(p, ' ', &skip);
917 			cp++;
918 			break;
919 		case ASCII_HYPH:
920 			sz += cond_width(p, '-', &skip);
921 			cp++;
922 			break;
923 		default:
924 			break;
925 		}
926 	}
927 
928 	return sz;
929 }
930 
931 int
932 term_vspan(const struct termp *p, const struct roffsu *su)
933 {
934 	double		 r;
935 	int		 ri;
936 
937 	switch (su->unit) {
938 	case SCALE_BU:
939 		r = su->scale / 40.0;
940 		break;
941 	case SCALE_CM:
942 		r = su->scale * 6.0 / 2.54;
943 		break;
944 	case SCALE_FS:
945 		r = su->scale * 65536.0 / 40.0;
946 		break;
947 	case SCALE_IN:
948 		r = su->scale * 6.0;
949 		break;
950 	case SCALE_MM:
951 		r = su->scale * 0.006;
952 		break;
953 	case SCALE_PC:
954 		r = su->scale;
955 		break;
956 	case SCALE_PT:
957 		r = su->scale / 12.0;
958 		break;
959 	case SCALE_EN:
960 	case SCALE_EM:
961 		r = su->scale * 0.6;
962 		break;
963 	case SCALE_VS:
964 		r = su->scale;
965 		break;
966 	default:
967 		abort();
968 	}
969 	ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
970 	return ri < 66 ? ri : 1;
971 }
972 
973 /*
974  * Convert a scaling width to basic units, rounding towards 0.
975  */
976 int
977 term_hspan(const struct termp *p, const struct roffsu *su)
978 {
979 
980 	return (*p->hspan)(p, su);
981 }
982 
983 /*
984  * Convert a scaling width to basic units, rounding to closest.
985  */
986 int
987 term_hen(const struct termp *p, const struct roffsu *su)
988 {
989 	int bu;
990 
991 	if ((bu = (*p->hspan)(p, su)) >= 0)
992 		return (bu + 11) / 24;
993 	else
994 		return -((-bu + 11) / 24);
995 }
996