1 /* $Id: term.c,v 1.212 2013/12/23 02:20:09 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
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.
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.
22 #include <sys/types.h>
36 static size_t cond_width(const struct termp *, int, int *);
37 static void adjbuf(struct termp *p, size_t);
38 static void bufferc(struct termp *, char);
39 static void encode(struct termp *, const char *, size_t);
40 static void encode1(struct termp *, int);
43 term_free(struct termp *p)
49 mchars_free(p->symtab);
56 term_begin(struct termp *p, term_margin head,
57 term_margin foot, const void *arg)
68 term_end(struct termp *p)
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.
81 * The following flags may be specified:
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 * p->trailspace may be set to 0, 1, or 2, depending on how many
87 * space characters are required at the end of the column.
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.
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.
96 * In-line line breaking:
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.
105 term_flushln(struct termp *p)
107 size_t i; /* current input position in p->buf */
108 int ntab; /* number of tabs to prepend */
109 size_t vis; /* current visual position on output */
110 size_t vbl; /* number of blanks to prepend to output */
111 size_t vend; /* end of word visual position on output */
112 size_t bp; /* visual right border position */
113 size_t dv; /* temporary for visual pos calculations */
114 size_t j; /* temporary loop index for p->buf */
115 size_t jhy; /* last hyph before overflow w/r/t j */
116 size_t maxvis; /* output position of visible boundary */
117 size_t mmax; /* used in calculating bp */
120 * First, establish the maximum columns of "visible" content.
121 * This is usually the difference between the right-margin and
122 * an indentation, but can be, for tagged lists or columns, a
123 * small set of values.
125 * The following unsigned-signed subtractions look strange,
126 * but they are actually correct. If the int p->overstep
127 * is negative, it gets sign extended. Subtracting that
128 * very large size_t effectively adds a small number to dv.
130 assert (p->rmargin >= p->offset);
131 dv = p->rmargin - p->offset;
132 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
133 dv = p->maxrmargin - p->offset;
134 mmax = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
136 bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
139 * Calculate the required amount of padding.
141 vbl = p->offset + p->overstep > p->viscol ?
142 p->offset + p->overstep - p->viscol : 0;
149 * Handle literal tab characters: collapse all
150 * subsequent tabs into a single huge set of spaces.
153 while (i < p->col && '\t' == p->buf[i]) {
154 vend = (vis / p->tabwidth + 1) * p->tabwidth;
162 * Count up visible word characters. Control sequences
163 * (starting with the CSI) aren't counted. A space
164 * generates a non-printing word, which is valid (the
165 * space is printed according to regular spacing rules).
168 for (j = i, jhy = 0; j < p->col; j++) {
169 if (' ' == p->buf[j] || '\t' == p->buf[j])
172 /* Back over the the last printed character. */
173 if (8 == p->buf[j]) {
175 vend -= (*p->width)(p, p->buf[j - 1]);
180 /* Break at the hyphen point if we overrun. */
181 if (vend > vis && vend < bp &&
182 ASCII_HYPH == p->buf[j])
185 vend += (*p->width)(p, p->buf[j]);
189 * Find out whether we would exceed the right margin.
190 * If so, break to the next line.
192 if (vend > bp && 0 == jhy && vis > 0) {
196 if (TERMP_NOBREAK & p->flags) {
198 vend += p->rmargin - p->offset;
202 /* use pending tabs on the new line */
205 vbl += ntab * p->tabwidth;
208 * Remove the p->overstep width.
209 * Again, if p->overstep is negative,
210 * sign extension does the right thing.
213 bp += (size_t)p->overstep;
217 /* Write out the [remaining] word. */
218 for ( ; i < p->col; i++) {
219 if (vend > bp && jhy > 0 && i > jhy)
221 if ('\t' == p->buf[i])
223 if (' ' == p->buf[i]) {
225 while (' ' == p->buf[i])
227 dv = (i - j) * (*p->width)(p, ' ');
232 if (ASCII_NBRSP == p->buf[i]) {
233 vbl += (*p->width)(p, ' ');
238 * Now we definitely know there will be
239 * printable characters to output,
240 * so write preceding white space now.
243 (*p->advance)(p, vbl);
248 if (ASCII_HYPH == p->buf[i]) {
249 (*p->letter)(p, '-');
250 p->viscol += (*p->width)(p, '-');
254 (*p->letter)(p, p->buf[i]);
256 p->viscol -= (*p->width)(p, p->buf[i-1]);
258 p->viscol += (*p->width)(p, p->buf[i]);
264 * If there was trailing white space, it was not printed;
265 * so reset the cursor position accordingly.
273 if ( ! (TERMP_NOBREAK & p->flags)) {
279 if (TERMP_HANG & p->flags) {
280 p->overstep = (int)(vis - maxvis +
281 p->trailspace * (*p->width)(p, ' '));
284 * If we have overstepped the margin, temporarily move
285 * it to the right and flag the rest of the line to be
287 * If there is a request to keep the columns together,
288 * allow negative overstep when the column is not full.
290 if (p->trailspace && p->overstep < 0)
294 } else if (TERMP_DANGLE & p->flags)
297 /* If the column was overrun, break the line. */
298 if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) {
306 * A newline only breaks an existing line; it won't assert vertical
307 * space. All data in the output buffer is flushed prior to the newline
311 term_newln(struct termp *p)
314 p->flags |= TERMP_NOSPACE;
315 if (p->col || p->viscol)
321 * Asserts a vertical space (a full, empty line-break between lines).
322 * Note that if used twice, this will cause two blank spaces and so on.
323 * All data in the output buffer is flushed prior to the newline
327 term_vspace(struct termp *p)
339 term_fontlast(struct termp *p)
344 p->fontl = p->fontq[p->fonti];
345 p->fontq[p->fonti] = f;
350 term_fontrepl(struct termp *p, enum termfont f)
353 p->fontl = p->fontq[p->fonti];
354 p->fontq[p->fonti] = f;
359 term_fontpush(struct termp *p, enum termfont f)
362 assert(p->fonti + 1 < 10);
363 p->fontl = p->fontq[p->fonti];
364 p->fontq[++p->fonti] = f;
369 term_fontq(struct termp *p)
372 return(&p->fontq[p->fonti]);
377 term_fonttop(struct termp *p)
380 return(p->fontq[p->fonti]);
385 term_fontpopq(struct termp *p, const void *key)
388 while (p->fonti >= 0 && key < (void *)(p->fontq + p->fonti))
390 assert(p->fonti >= 0);
395 term_fontpop(struct termp *p)
403 * Handle pwords, partial words, which may be either a single word or a
404 * phrase that cannot be broken down (such as a literal string). This
405 * handles word styling.
408 term_word(struct termp *p, const char *word)
410 const char *seq, *cp;
416 if ( ! (TERMP_NOSPACE & p->flags)) {
417 if ( ! (TERMP_KEEP & p->flags)) {
419 if (TERMP_SENTENCE & p->flags)
422 bufferc(p, ASCII_NBRSP);
424 if (TERMP_PREKEEP & p->flags)
425 p->flags |= TERMP_KEEP;
427 if ( ! (p->flags & TERMP_NONOSPACE))
428 p->flags &= ~TERMP_NOSPACE;
430 p->flags |= TERMP_NOSPACE;
432 p->flags &= ~(TERMP_SENTENCE | TERMP_IGNDELIM);
434 while ('\0' != *word) {
436 if (TERMP_SKIPCHAR & p->flags) {
437 p->flags &= ~TERMP_SKIPCHAR;
441 ssz = strcspn(word, "\\");
442 encode(p, word, ssz);
448 esc = mandoc_escape(&word, &seq, &sz);
449 if (ESCAPE_ERROR == esc)
452 if (TERMENC_ASCII != p->enc)
454 case (ESCAPE_UNICODE):
455 uc = mchars_num2uc(seq + 1, sz - 1);
460 case (ESCAPE_SPECIAL):
461 uc = mchars_spec2cp(p->symtab, seq, sz);
471 case (ESCAPE_UNICODE):
474 case (ESCAPE_NUMBERED):
475 c = mchars_num2char(seq, sz);
479 case (ESCAPE_SPECIAL):
480 cp = mchars_spec2str(p->symtab, seq, sz, &ssz);
486 case (ESCAPE_FONTBOLD):
487 term_fontrepl(p, TERMFONT_BOLD);
489 case (ESCAPE_FONTITALIC):
490 term_fontrepl(p, TERMFONT_UNDER);
492 case (ESCAPE_FONTBI):
493 term_fontrepl(p, TERMFONT_BI);
497 case (ESCAPE_FONTROMAN):
498 term_fontrepl(p, TERMFONT_NONE);
500 case (ESCAPE_FONTPREV):
503 case (ESCAPE_NOSPACE):
504 if (TERMP_SKIPCHAR & p->flags)
505 p->flags &= ~TERMP_SKIPCHAR;
506 else if ('\0' == *word)
507 p->flags |= TERMP_NOSPACE;
509 case (ESCAPE_SKIPCHAR):
510 p->flags |= TERMP_SKIPCHAR;
519 adjbuf(struct termp *p, size_t sz)
524 while (sz >= p->maxcols)
527 p->buf = mandoc_realloc(p->buf, sizeof(int) * p->maxcols);
531 bufferc(struct termp *p, char c)
534 if (p->col + 1 >= p->maxcols)
535 adjbuf(p, p->col + 1);
537 p->buf[p->col++] = c;
542 * Do this for a single (probably unicode) value.
543 * Does not check for non-decorated glyphs.
546 encode1(struct termp *p, int c)
550 if (TERMP_SKIPCHAR & p->flags) {
551 p->flags &= ~TERMP_SKIPCHAR;
555 if (p->col + 6 >= p->maxcols)
556 adjbuf(p, p->col + 6);
560 if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
561 p->buf[p->col++] = '_';
562 p->buf[p->col++] = 8;
564 if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
566 p->buf[p->col++] = '-';
568 p->buf[p->col++] = c;
569 p->buf[p->col++] = 8;
571 p->buf[p->col++] = c;
575 encode(struct termp *p, const char *word, size_t sz)
579 if (TERMP_SKIPCHAR & p->flags) {
580 p->flags &= ~TERMP_SKIPCHAR;
585 * Encode and buffer a string of characters. If the current
586 * font mode is unset, buffer directly, else encode then buffer
587 * character by character.
590 if (TERMFONT_NONE == term_fonttop(p)) {
591 if (p->col + sz >= p->maxcols)
592 adjbuf(p, p->col + sz);
593 for (i = 0; i < sz; i++)
594 p->buf[p->col++] = word[i];
598 /* Pre-buffer, assuming worst-case. */
600 if (p->col + 1 + (sz * 5) >= p->maxcols)
601 adjbuf(p, p->col + 1 + (sz * 5));
603 for (i = 0; i < sz; i++) {
604 if (ASCII_HYPH == word[i] ||
605 isgraph((unsigned char)word[i]))
608 p->buf[p->col++] = word[i];
613 term_len(const struct termp *p, size_t sz)
616 return((*p->width)(p, ' ') * sz);
620 cond_width(const struct termp *p, int c, int *skip)
627 return((*p->width)(p, c));
631 term_strlen(const struct termp *p, const char *cp)
635 const char *seq, *rhs;
637 static const char rej[] = { '\\', ASCII_HYPH, ASCII_NBRSP, '\0' };
640 * Account for escaped sequences within string length
641 * calculations. This follows the logic in term_word() as we
642 * must calculate the width of produced strings.
647 while ('\0' != *cp) {
648 rsz = strcspn(cp, rej);
649 for (i = 0; i < rsz; i++)
650 sz += cond_width(p, *cp++, &skip);
656 esc = mandoc_escape(&cp, &seq, &ssz);
657 if (ESCAPE_ERROR == esc)
660 if (TERMENC_ASCII != p->enc)
662 case (ESCAPE_UNICODE):
667 sz += cond_width(p, c, &skip);
669 case (ESCAPE_SPECIAL):
671 (p->symtab, seq, ssz);
674 sz += cond_width(p, c, &skip);
683 case (ESCAPE_UNICODE):
684 sz += cond_width(p, '?', &skip);
686 case (ESCAPE_NUMBERED):
687 c = mchars_num2char(seq, ssz);
689 sz += cond_width(p, c, &skip);
691 case (ESCAPE_SPECIAL):
692 rhs = mchars_spec2str
693 (p->symtab, seq, ssz, &rsz);
701 case (ESCAPE_SKIPCHAR):
716 for (i = 0; i < rsz; i++)
717 sz += (*p->width)(p, *rhs++);
720 sz += cond_width(p, ' ', &skip);
724 sz += cond_width(p, '-', &skip);
737 term_vspan(const struct termp *p, const struct roffsu *su)
755 r = su->scale / 1000;
767 return(/* LINTED */(size_t)
772 term_hspan(const struct termp *p, const struct roffsu *su)
776 v = ((*p->hspan)(p, su));
779 return((size_t) /* LINTED */