1 /* $Id: term.c,v 1.268 2017/06/08 12:54:58 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2017 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_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 *);
43 term_free(struct termp *p)
45 for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++)
53 term_begin(struct termp *p, term_margin head,
54 term_margin foot, const struct roff_meta *arg)
64 term_end(struct termp *p)
71 * Flush a chunk of text. By default, break the output line each time
72 * the right margin is reached, and continue output on the next line
73 * at the same offset as the chunk itself. By default, also break the
74 * output line at the end of the chunk.
75 * The following flags may be specified:
77 * - TERMP_NOBREAK: Do not break the output line at the right margin,
78 * but only at the max right margin. Also, do not break the output
79 * line at the end of the chunk, such that the next call can pad to
80 * the next column. However, if less than p->trailspace blanks,
81 * which can be 0, 1, or 2, remain to the right margin, the line
83 * - TERMP_BRTRSP: Consider trailing whitespace significant
84 * when deciding whether the chunk fits or not.
85 * - TERMP_BRIND: If the chunk does not fit and the output line has
86 * to be broken, start the next line at the right margin instead
87 * of at the offset. Used together with TERMP_NOBREAK for the tags
88 * in various kinds of tagged lists.
89 * - TERMP_HANG: Do not break the output line at the right margin,
90 * append the next chunk after it even if this one is too long.
91 * To be used together with TERMP_NOBREAK.
92 * - TERMP_NOPAD: Start writing at the current position,
93 * do not pad with blank characters up to the offset.
96 term_flushln(struct termp *p)
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->tcol->buf */
104 size_t jhy; /* last hyph before overflow w/r/t j */
105 size_t maxvis; /* output position of visible boundary */
106 int ntab; /* number of tabs to prepend */
108 vbl = (p->flags & TERMP_NOPAD) || p->tcol->offset < p->viscol ?
109 0 : p->tcol->offset - p->viscol;
110 if (p->minbl && vbl < p->minbl)
112 maxvis = p->tcol->rmargin > p->viscol + vbl ?
113 p->tcol->rmargin - p->viscol - vbl : 0;
114 bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
115 p->maxrmargin > p->viscol + vbl ?
116 p->maxrmargin - p->viscol - vbl : 0;
119 if (p->lasttcol == 0)
121 while (p->tcol->col < p->lastcol) {
124 * Handle literal tab characters: collapse all
125 * subsequent tabs into a single huge set of spaces.
129 while (p->tcol->col < p->lastcol &&
130 p->tcol->buf[p->tcol->col] == '\t') {
131 vend = term_tab_next(vis);
139 * Count up visible word characters. Control sequences
140 * (starting with the CSI) aren't counted. A space
141 * generates a non-printing word, which is valid (the
142 * space is printed according to regular spacing rules).
146 for (j = p->tcol->col; j < p->lastcol; j++) {
147 if (p->tcol->buf[j] == ' ' || p->tcol->buf[j] == '\t')
150 /* Back over the last printed character. */
151 if (p->tcol->buf[j] == '\b') {
153 vend -= (*p->width)(p, p->tcol->buf[j - 1]);
158 /* Break at the hyphen point if we overrun. */
159 if (vend > vis && vend < bp &&
160 (p->tcol->buf[j] == ASCII_HYPH||
161 p->tcol->buf[j] == ASCII_BREAK))
165 * Hyphenation now decided, put back a real
166 * hyphen such that we get the correct width.
168 if (p->tcol->buf[j] == ASCII_HYPH)
169 p->tcol->buf[j] = '-';
171 vend += (*p->width)(p, p->tcol->buf[j]);
175 * Find out whether we would exceed the right margin.
176 * If so, break to the next line.
179 if (vend > bp && jhy == 0 && vis > 0 &&
180 (p->flags & TERMP_BRNEVER) == 0) {
187 /* Use pending tabs on the new line. */
191 vbl = term_tab_next(vbl);
193 /* Re-establish indentation. */
195 if (p->flags & TERMP_BRIND)
196 vbl += p->tcol->rmargin;
198 vbl += p->tcol->offset;
199 maxvis = p->tcol->rmargin > vbl ?
200 p->tcol->rmargin - vbl : 0;
201 bp = !(p->flags & TERMP_NOBREAK) ? maxvis :
202 p->maxrmargin > vbl ? p->maxrmargin - vbl : 0;
206 * Write out the rest of the word.
209 for ( ; p->tcol->col < p->lastcol; p->tcol->col++) {
210 if (vend > bp && jhy > 0 && p->tcol->col > jhy)
212 if (p->tcol->buf[p->tcol->col] == '\t')
214 if (p->tcol->buf[p->tcol->col] == ' ') {
216 while (p->tcol->col < p->lastcol &&
217 p->tcol->buf[p->tcol->col] == ' ')
219 dv = (p->tcol->col - j) * (*p->width)(p, ' ');
224 if (p->tcol->buf[p->tcol->col] == ASCII_NBRSP) {
225 vbl += (*p->width)(p, ' ');
228 if (p->tcol->buf[p->tcol->col] == ASCII_BREAK)
232 * Now we definitely know there will be
233 * printable characters to output,
234 * so write preceding white space now.
237 (*p->advance)(p, vbl);
242 (*p->letter)(p, p->tcol->buf[p->tcol->col]);
243 if (p->tcol->buf[p->tcol->col] == '\b')
244 p->viscol -= (*p->width)(p,
245 p->tcol->buf[p->tcol->col - 1]);
247 p->viscol += (*p->width)(p,
248 p->tcol->buf[p->tcol->col]);
254 * If there was trailing white space, it was not printed;
255 * so reset the cursor position accordingly.
263 p->col = p->lastcol = 0;
264 p->minbl = p->trailspace;
265 p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
267 /* Trailing whitespace is significant in some columns. */
269 if (vis && vbl && (TERMP_BRTRSP & p->flags))
272 /* If the column was overrun, break the line. */
273 if ((p->flags & TERMP_NOBREAK) == 0 ||
274 ((p->flags & TERMP_HANG) == 0 &&
275 vis + p->trailspace * (*p->width)(p, ' ') > maxvis))
280 endline(struct termp *p)
282 if ((p->flags & (TERMP_NEWMC | TERMP_ENDMC)) == TERMP_ENDMC) {
284 p->flags &= ~TERMP_ENDMC;
287 if (p->viscol && p->maxrmargin >= p->viscol)
288 (*p->advance)(p, p->maxrmargin - p->viscol + 1);
289 p->flags |= TERMP_NOBUF | TERMP_NOSPACE;
291 p->flags &= ~(TERMP_NOBUF | TERMP_NEWMC);
299 * A newline only breaks an existing line; it won't assert vertical
300 * space. All data in the output buffer is flushed prior to the newline
304 term_newln(struct termp *p)
307 p->flags |= TERMP_NOSPACE;
308 if (p->lastcol || p->viscol)
313 * Asserts a vertical space (a full, empty line-break between lines).
314 * Note that if used twice, this will cause two blank spaces and so on.
315 * All data in the output buffer is flushed prior to the newline
319 term_vspace(struct termp *p)
331 /* Swap current and previous font; for \fP and .ft P */
333 term_fontlast(struct termp *p)
338 p->fontl = p->fontq[p->fonti];
339 p->fontq[p->fonti] = f;
342 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
344 term_fontrepl(struct termp *p, enum termfont f)
347 p->fontl = p->fontq[p->fonti];
348 p->fontq[p->fonti] = f;
351 /* Set font, save previous. */
353 term_fontpush(struct termp *p, enum termfont f)
356 p->fontl = p->fontq[p->fonti];
357 if (++p->fonti == p->fontsz) {
359 p->fontq = mandoc_reallocarray(p->fontq,
360 p->fontsz, sizeof(*p->fontq));
362 p->fontq[p->fonti] = f;
365 /* Flush to make the saved pointer current again. */
367 term_fontpopq(struct termp *p, int i)
375 /* Pop one font off the stack. */
377 term_fontpop(struct termp *p)
385 * Handle pwords, partial words, which may be either a single word or a
386 * phrase that cannot be broken down (such as a literal string). This
387 * handles word styling.
390 term_word(struct termp *p, const char *word)
393 const char nbrsp[2] = { ASCII_NBRSP, 0 };
394 const char *seq, *cp;
396 size_t csz, lsz, ssz;
399 if ((p->flags & TERMP_NOBUF) == 0) {
400 if ((p->flags & TERMP_NOSPACE) == 0) {
401 if ((p->flags & TERMP_KEEP) == 0) {
403 if (p->flags & TERMP_SENTENCE)
406 bufferc(p, ASCII_NBRSP);
408 if (p->flags & TERMP_PREKEEP)
409 p->flags |= TERMP_KEEP;
410 if (p->flags & TERMP_NONOSPACE)
411 p->flags |= TERMP_NOSPACE;
413 p->flags &= ~TERMP_NOSPACE;
414 p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);
418 while ('\0' != *word) {
420 if (TERMP_NBRWORD & p->flags) {
426 ssz = strcspn(word, "\\ ");
428 ssz = strcspn(word, "\\");
429 encode(p, word, ssz);
435 esc = mandoc_escape(&word, &seq, &sz);
436 if (ESCAPE_ERROR == esc)
441 uc = mchars_num2uc(seq + 1, sz - 1);
443 case ESCAPE_NUMBERED:
444 uc = mchars_num2char(seq, sz);
449 if (p->enc == TERMENC_ASCII) {
450 cp = mchars_spec2str(seq, sz, &ssz);
454 uc = mchars_spec2cp(seq, sz);
459 case ESCAPE_FONTBOLD:
460 term_fontrepl(p, TERMFONT_BOLD);
462 case ESCAPE_FONTITALIC:
463 term_fontrepl(p, TERMFONT_UNDER);
466 term_fontrepl(p, TERMFONT_BI);
469 case ESCAPE_FONTROMAN:
470 term_fontrepl(p, TERMFONT_NONE);
472 case ESCAPE_FONTPREV:
476 if (p->flags & TERMP_BACKAFTER)
477 p->flags &= ~TERMP_BACKAFTER;
478 else if (*word == '\0')
479 p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
482 if (a2roffsu(seq, &su, SCALE_EM) == NULL)
484 uc = term_hspan(p, &su) / 24;
487 bufferc(p, ASCII_NBRSP);
488 else if (p->col > (size_t)(-uc))
493 if (p->tcol->offset > (size_t)(-uc)) {
495 p->tcol->offset += uc;
497 p->ti -= p->tcol->offset;
503 if ((seq = a2roffsu(seq, &su, SCALE_EM)) == NULL)
505 uc = term_hspan(p, &su) / 24;
507 if (p->tcol->rmargin <= p->tcol->offset)
509 lsz = p->tcol->rmargin - p->tcol->offset;
514 else if (*seq == '\\') {
516 esc = mandoc_escape(&seq, &cp, &sz);
519 uc = mchars_num2uc(cp + 1, sz - 1);
521 case ESCAPE_NUMBERED:
522 uc = mchars_num2char(cp, sz);
525 uc = mchars_spec2cp(cp, sz);
533 if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
535 if (p->enc == TERMENC_ASCII) {
536 cp = ascii_uc2str(uc);
537 csz = term_strlen(p, cp);
540 csz = (*p->width)(p, uc);
542 if (p->enc == TERMENC_ASCII)
549 case ESCAPE_SKIPCHAR:
550 p->flags |= TERMP_BACKAFTER;
552 case ESCAPE_OVERSTRIKE:
556 mandoc_escape(&seq, NULL, NULL);
561 if (p->flags & TERMP_BACKBEFORE)
562 p->flags |= TERMP_BACKAFTER;
564 p->flags |= TERMP_BACKBEFORE;
567 /* Trim trailing backspace/blank pair. */
568 if (p->lastcol > 2 &&
569 (p->tcol->buf[p->lastcol - 1] == ' ' ||
570 p->tcol->buf[p->lastcol - 1] == '\t'))
572 if (p->col > p->lastcol)
580 * Common handling for Unicode and numbered
581 * character escape sequences.
584 if (p->enc == TERMENC_ASCII) {
585 cp = ascii_uc2str(uc);
586 encode(p, cp, strlen(cp));
588 if ((uc < 0x20 && uc != 0x09) ||
589 (uc > 0x7E && uc < 0xA0))
594 p->flags &= ~TERMP_NBRWORD;
598 adjbuf(struct termp_col *c, size_t sz)
602 while (c->maxcols <= sz)
604 c->buf = mandoc_reallocarray(c->buf, c->maxcols, sizeof(*c->buf));
608 bufferc(struct termp *p, char c)
610 if (p->flags & TERMP_NOBUF) {
614 if (p->col + 1 >= p->tcol->maxcols)
615 adjbuf(p->tcol, p->col + 1);
616 if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
617 p->tcol->buf[p->col] = c;
618 if (p->lastcol < ++p->col)
624 * Do this for a single (probably unicode) value.
625 * Does not check for non-decorated glyphs.
628 encode1(struct termp *p, int c)
632 if (p->flags & TERMP_NOBUF) {
637 if (p->col + 7 >= p->tcol->maxcols)
638 adjbuf(p->tcol, p->col + 7);
640 f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ?
641 p->fontq[p->fonti] : TERMFONT_NONE;
643 if (p->flags & TERMP_BACKBEFORE) {
644 if (p->tcol->buf[p->col - 1] == ' ' ||
645 p->tcol->buf[p->col - 1] == '\t')
648 p->tcol->buf[p->col++] = '\b';
649 p->flags &= ~TERMP_BACKBEFORE;
651 if (f == TERMFONT_UNDER || f == TERMFONT_BI) {
652 p->tcol->buf[p->col++] = '_';
653 p->tcol->buf[p->col++] = '\b';
655 if (f == TERMFONT_BOLD || f == TERMFONT_BI) {
657 p->tcol->buf[p->col++] = '-';
659 p->tcol->buf[p->col++] = c;
660 p->tcol->buf[p->col++] = '\b';
662 if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP))
663 p->tcol->buf[p->col] = c;
664 if (p->lastcol < ++p->col)
666 if (p->flags & TERMP_BACKAFTER) {
667 p->flags |= TERMP_BACKBEFORE;
668 p->flags &= ~TERMP_BACKAFTER;
673 encode(struct termp *p, const char *word, size_t sz)
677 if (p->flags & TERMP_NOBUF) {
678 for (i = 0; i < sz; i++)
679 (*p->letter)(p, word[i]);
683 if (p->col + 2 + (sz * 5) >= p->tcol->maxcols)
684 adjbuf(p->tcol, p->col + 2 + (sz * 5));
686 for (i = 0; i < sz; i++) {
687 if (ASCII_HYPH == word[i] ||
688 isgraph((unsigned char)word[i]))
691 if (p->lastcol <= p->col ||
692 (word[i] != ' ' && word[i] != ASCII_NBRSP))
693 p->tcol->buf[p->col] = word[i];
697 * Postpone the effect of \z while handling
698 * an overstrike sequence from ascii_uc2str().
701 if (word[i] == '\b' &&
702 (p->flags & TERMP_BACKBEFORE)) {
703 p->flags &= ~TERMP_BACKBEFORE;
704 p->flags |= TERMP_BACKAFTER;
708 if (p->lastcol < p->col)
713 term_setwidth(struct termp *p, const char *wstr)
733 if (a2roffsu(wstr, &su, SCALE_MAX) != NULL)
734 width = term_hspan(p, &su);
738 (*p->setwidth)(p, iop, width);
742 term_len(const struct termp *p, size_t sz)
745 return (*p->width)(p, ' ') * sz;
749 cond_width(const struct termp *p, int c, int *skip)
756 return (*p->width)(p, c);
760 term_strlen(const struct termp *p, const char *cp)
764 const char *seq, *rhs;
766 static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
770 * Account for escaped sequences within string length
771 * calculations. This follows the logic in term_word() as we
772 * must calculate the width of produced strings.
777 while ('\0' != *cp) {
778 rsz = strcspn(cp, rej);
779 for (i = 0; i < rsz; i++)
780 sz += cond_width(p, *cp++, &skip);
785 esc = mandoc_escape(&cp, &seq, &ssz);
786 if (ESCAPE_ERROR == esc)
793 uc = mchars_num2uc(seq + 1, ssz - 1);
795 case ESCAPE_NUMBERED:
796 uc = mchars_num2char(seq, ssz);
801 if (p->enc == TERMENC_ASCII) {
802 rhs = mchars_spec2str(seq, ssz, &rsz);
806 uc = mchars_spec2cp(seq, ssz);
808 sz += cond_width(p, uc, &skip);
811 case ESCAPE_SKIPCHAR:
814 case ESCAPE_OVERSTRIKE:
819 mandoc_escape(&seq, NULL, NULL);
822 i = (*p->width)(p, *seq++);
833 * Common handling for Unicode and numbered
834 * character escape sequences.
838 if (p->enc == TERMENC_ASCII) {
839 rhs = ascii_uc2str(uc);
842 if ((uc < 0x20 && uc != 0x09) ||
843 (uc > 0x7E && uc < 0xA0))
845 sz += cond_width(p, uc, &skip);
856 * Common handling for all escape sequences
857 * printing more than one character.
860 for (i = 0; i < rsz; i++)
861 sz += (*p->width)(p, *rhs++);
864 sz += cond_width(p, ' ', &skip);
868 sz += cond_width(p, '-', &skip);
880 term_vspan(const struct termp *p, const struct roffsu *su)
887 r = su->scale / 40.0;
890 r = su->scale * 6.0 / 2.54;
893 r = su->scale * 65536.0 / 40.0;
899 r = su->scale * 0.006;
905 r = su->scale / 12.0;
917 ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
918 return ri < 66 ? ri : 1;
922 * Convert a scaling width to basic units, rounding down.
925 term_hspan(const struct termp *p, const struct roffsu *su)
928 return (*p->hspan)(p, su);