1 /* $Id: term.c,v 1.257 2016/04/12 15:30:00 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2015 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 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.
20 #include <sys/types.h>
29 #include "mandoc_aux.h"
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);
42 term_free(struct termp *p)
51 term_begin(struct termp *p, term_margin head,
52 term_margin foot, const struct roff_meta *arg)
62 term_end(struct termp *p)
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:
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
81 * - TERMP_BRTRSP: Consider trailing whitespace significant
82 * when deciding whether the chunk fits or not.
83 * - TERMP_BRIND: If the chunk does not fit and the output line has
84 * to be broken, start the next line at the right margin instead
85 * of at the offset. Used together with TERMP_NOBREAK for the tags
86 * in various kinds of tagged lists.
87 * - TERMP_DANGLE: Do not break the output line at the right margin,
88 * append the next chunk after it even if this one is too long.
89 * To be used together with TERMP_NOBREAK.
90 * - TERMP_HANG: Like TERMP_DANGLE, and also suppress padding before
91 * the next chunk if this column is not full.
94 term_flushln(struct termp *p)
96 size_t i; /* current input position in p->buf */
97 int ntab; /* number of tabs to prepend */
98 size_t vis; /* current visual position on output */
99 size_t vbl; /* number of blanks to prepend to output */
100 size_t vend; /* end of word visual position on output */
101 size_t bp; /* visual right border position */
102 size_t dv; /* temporary for visual pos calculations */
103 size_t j; /* temporary loop index for p->buf */
104 size_t jhy; /* last hyph before overflow w/r/t j */
105 size_t maxvis; /* output position of visible boundary */
108 * First, establish the maximum columns of "visible" content.
109 * This is usually the difference between the right-margin and
110 * an indentation, but can be, for tagged lists or columns, a
111 * small set of values.
113 * The following unsigned-signed subtractions look strange,
114 * but they are actually correct. If the int p->overstep
115 * is negative, it gets sign extended. Subtracting that
116 * very large size_t effectively adds a small number to dv.
118 dv = p->rmargin > p->offset ? p->rmargin - p->offset : 0;
119 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
121 if (p->flags & TERMP_NOBREAK) {
122 dv = p->maxrmargin > p->offset ?
123 p->maxrmargin - p->offset : 0;
124 bp = (int)dv > p->overstep ?
125 dv - (size_t)p->overstep : 0;
130 * Calculate the required amount of padding.
132 vbl = p->offset + p->overstep > p->viscol ?
133 p->offset + p->overstep - p->viscol : 0;
140 * Handle literal tab characters: collapse all
141 * subsequent tabs into a single huge set of spaces.
144 while (i < p->col && '\t' == p->buf[i]) {
145 vend = (vis / p->tabwidth + 1) * p->tabwidth;
153 * Count up visible word characters. Control sequences
154 * (starting with the CSI) aren't counted. A space
155 * generates a non-printing word, which is valid (the
156 * space is printed according to regular spacing rules).
159 for (j = i, jhy = 0; j < p->col; j++) {
160 if (' ' == p->buf[j] || '\t' == p->buf[j])
163 /* Back over the last printed character. */
164 if (8 == p->buf[j]) {
166 vend -= (*p->width)(p, p->buf[j - 1]);
171 /* Break at the hyphen point if we overrun. */
172 if (vend > vis && vend < bp &&
173 (ASCII_HYPH == p->buf[j] ||
174 ASCII_BREAK == p->buf[j]))
178 * Hyphenation now decided, put back a real
179 * hyphen such that we get the correct width.
181 if (ASCII_HYPH == p->buf[j])
184 vend += (*p->width)(p, p->buf[j]);
188 * Find out whether we would exceed the right margin.
189 * If so, break to the next line.
191 if (vend > bp && 0 == jhy && vis > 0) {
195 if (TERMP_BRIND & p->flags) {
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 (i < p->col && ' ' == p->buf[i])
227 dv = (i - j) * (*p->width)(p, ' ');
232 if (ASCII_NBRSP == p->buf[i]) {
233 vbl += (*p->width)(p, ' ');
236 if (ASCII_BREAK == p->buf[i])
240 * Now we definitely know there will be
241 * printable characters to output,
242 * so write preceding white space now.
245 (*p->advance)(p, vbl);
250 (*p->letter)(p, p->buf[i]);
252 p->viscol -= (*p->width)(p, p->buf[i-1]);
254 p->viscol += (*p->width)(p, p->buf[i]);
260 * If there was trailing white space, it was not printed;
261 * so reset the cursor position accordingly.
270 p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE);
272 if ( ! (TERMP_NOBREAK & p->flags)) {
278 if (TERMP_HANG & p->flags) {
279 p->overstep += (int)(p->offset + vis - p->rmargin +
280 p->trailspace * (*p->width)(p, ' '));
283 * If we have overstepped the margin, temporarily move
284 * it to the right and flag the rest of the line to be
286 * If there is a request to keep the columns together,
287 * allow negative overstep when the column is not full.
289 if (p->trailspace && p->overstep < 0)
293 } else if (TERMP_DANGLE & p->flags)
296 /* Trailing whitespace is significant in some columns. */
297 if (vis && vbl && (TERMP_BRTRSP & p->flags))
300 /* If the column was overrun, break the line. */
301 if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) {
308 * A newline only breaks an existing line; it won't assert vertical
309 * space. All data in the output buffer is flushed prior to the newline
313 term_newln(struct termp *p)
316 p->flags |= TERMP_NOSPACE;
317 if (p->col || p->viscol)
322 * Asserts a vertical space (a full, empty line-break between lines).
323 * Note that if used twice, this will cause two blank spaces and so on.
324 * All data in the output buffer is flushed prior to the newline
328 term_vspace(struct termp *p)
339 /* Swap current and previous font; for \fP and .ft P */
341 term_fontlast(struct termp *p)
346 p->fontl = p->fontq[p->fonti];
347 p->fontq[p->fonti] = f;
350 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
352 term_fontrepl(struct termp *p, enum termfont f)
355 p->fontl = p->fontq[p->fonti];
356 p->fontq[p->fonti] = f;
359 /* Set font, save previous. */
361 term_fontpush(struct termp *p, enum termfont f)
364 p->fontl = p->fontq[p->fonti];
365 if (++p->fonti == p->fontsz) {
367 p->fontq = mandoc_reallocarray(p->fontq,
368 p->fontsz, sizeof(*p->fontq));
370 p->fontq[p->fonti] = f;
373 /* Flush to make the saved pointer current again. */
375 term_fontpopq(struct termp *p, int i)
383 /* Pop one font off the stack. */
385 term_fontpop(struct termp *p)
393 * Handle pwords, partial words, which may be either a single word or a
394 * phrase that cannot be broken down (such as a literal string). This
395 * handles word styling.
398 term_word(struct termp *p, const char *word)
400 const char nbrsp[2] = { ASCII_NBRSP, 0 };
401 const char *seq, *cp;
406 if ( ! (TERMP_NOSPACE & p->flags)) {
407 if ( ! (TERMP_KEEP & p->flags)) {
409 if (TERMP_SENTENCE & p->flags)
412 bufferc(p, ASCII_NBRSP);
414 if (TERMP_PREKEEP & p->flags)
415 p->flags |= TERMP_KEEP;
417 if ( ! (p->flags & TERMP_NONOSPACE))
418 p->flags &= ~TERMP_NOSPACE;
420 p->flags |= TERMP_NOSPACE;
422 p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);
425 while ('\0' != *word) {
427 if (TERMP_NBRWORD & p->flags) {
433 ssz = strcspn(word, "\\ ");
435 ssz = strcspn(word, "\\");
436 encode(p, word, ssz);
442 esc = mandoc_escape(&word, &seq, &sz);
443 if (ESCAPE_ERROR == esc)
448 uc = mchars_num2uc(seq + 1, sz - 1);
450 case ESCAPE_NUMBERED:
451 uc = mchars_num2char(seq, sz);
456 if (p->enc == TERMENC_ASCII) {
457 cp = mchars_spec2str(seq, sz, &ssz);
461 uc = mchars_spec2cp(seq, sz);
466 case ESCAPE_FONTBOLD:
467 term_fontrepl(p, TERMFONT_BOLD);
469 case ESCAPE_FONTITALIC:
470 term_fontrepl(p, TERMFONT_UNDER);
473 term_fontrepl(p, TERMFONT_BI);
476 case ESCAPE_FONTROMAN:
477 term_fontrepl(p, TERMFONT_NONE);
479 case ESCAPE_FONTPREV:
483 if (p->flags & TERMP_BACKAFTER)
484 p->flags &= ~TERMP_BACKAFTER;
485 else if (*word == '\0')
486 p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
488 case ESCAPE_SKIPCHAR:
489 p->flags |= TERMP_BACKAFTER;
491 case ESCAPE_OVERSTRIKE:
495 mandoc_escape(&seq, NULL, NULL);
500 if (p->flags & TERMP_BACKBEFORE)
501 p->flags |= TERMP_BACKAFTER;
503 p->flags |= TERMP_BACKBEFORE;
506 /* Trim trailing backspace/blank pair. */
507 if (p->col > 2 && p->buf[p->col - 1] == ' ')
515 * Common handling for Unicode and numbered
516 * character escape sequences.
519 if (p->enc == TERMENC_ASCII) {
520 cp = ascii_uc2str(uc);
521 encode(p, cp, strlen(cp));
523 if ((uc < 0x20 && uc != 0x09) ||
524 (uc > 0x7E && uc < 0xA0))
529 p->flags &= ~TERMP_NBRWORD;
533 adjbuf(struct termp *p, size_t sz)
538 while (sz >= p->maxcols)
541 p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int));
545 bufferc(struct termp *p, char c)
548 if (p->col + 1 >= p->maxcols)
549 adjbuf(p, p->col + 1);
551 p->buf[p->col++] = c;
556 * Do this for a single (probably unicode) value.
557 * Does not check for non-decorated glyphs.
560 encode1(struct termp *p, int c)
564 if (p->col + 7 >= p->maxcols)
565 adjbuf(p, p->col + 7);
567 f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
568 p->fontq[p->fonti] : TERMFONT_NONE;
570 if (p->flags & TERMP_BACKBEFORE) {
571 if (p->buf[p->col - 1] == ' ')
574 p->buf[p->col++] = 8;
575 p->flags &= ~TERMP_BACKBEFORE;
577 if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
578 p->buf[p->col++] = '_';
579 p->buf[p->col++] = 8;
581 if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
583 p->buf[p->col++] = '-';
585 p->buf[p->col++] = c;
586 p->buf[p->col++] = 8;
588 p->buf[p->col++] = c;
589 if (p->flags & TERMP_BACKAFTER) {
590 p->flags |= TERMP_BACKBEFORE;
591 p->flags &= ~TERMP_BACKAFTER;
596 encode(struct termp *p, const char *word, size_t sz)
600 if (p->col + 2 + (sz * 5) >= p->maxcols)
601 adjbuf(p, p->col + 2 + (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_setwidth(struct termp *p, const char *wstr)
633 if (a2roffsu(wstr, &su, SCALE_MAX))
634 width = term_hspan(p, &su);
638 (*p->setwidth)(p, iop, width);
642 term_len(const struct termp *p, size_t sz)
645 return (*p->width)(p, ' ') * sz;
649 cond_width(const struct termp *p, int c, int *skip)
656 return (*p->width)(p, c);
660 term_strlen(const struct termp *p, const char *cp)
664 const char *seq, *rhs;
666 static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
670 * Account for escaped sequences within string length
671 * calculations. This follows the logic in term_word() as we
672 * must calculate the width of produced strings.
677 while ('\0' != *cp) {
678 rsz = strcspn(cp, rej);
679 for (i = 0; i < rsz; i++)
680 sz += cond_width(p, *cp++, &skip);
685 esc = mandoc_escape(&cp, &seq, &ssz);
686 if (ESCAPE_ERROR == esc)
693 uc = mchars_num2uc(seq + 1, ssz - 1);
695 case ESCAPE_NUMBERED:
696 uc = mchars_num2char(seq, ssz);
701 if (p->enc == TERMENC_ASCII) {
702 rhs = mchars_spec2str(seq, ssz, &rsz);
706 uc = mchars_spec2cp(seq, ssz);
708 sz += cond_width(p, uc, &skip);
711 case ESCAPE_SKIPCHAR:
714 case ESCAPE_OVERSTRIKE:
719 mandoc_escape(&seq, NULL, NULL);
722 i = (*p->width)(p, *seq++);
733 * Common handling for Unicode and numbered
734 * character escape sequences.
738 if (p->enc == TERMENC_ASCII) {
739 rhs = ascii_uc2str(uc);
742 if ((uc < 0x20 && uc != 0x09) ||
743 (uc > 0x7E && uc < 0xA0))
745 sz += cond_width(p, uc, &skip);
756 * Common handling for all escape sequences
757 * printing more than one character.
760 for (i = 0; i < rsz; i++)
761 sz += (*p->width)(p, *rhs++);
764 sz += cond_width(p, ' ', &skip);
768 sz += cond_width(p, '-', &skip);
780 term_vspan(const struct termp *p, const struct roffsu *su)
787 r = su->scale / 40.0;
790 r = su->scale * 6.0 / 2.54;
793 r = su->scale * 65536.0 / 40.0;
799 r = su->scale * 0.006;
805 r = su->scale / 12.0;
817 ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
818 return ri < 66 ? ri : 1;
822 * Convert a scaling width to basic units, rounding down.
825 term_hspan(const struct termp *p, const struct roffsu *su)
828 return (*p->hspan)(p, su);