1 /* $Id: term.c,v 1.245 2015/03/06 13:02:43 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 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.
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 void *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_BRIND: If the chunk does not fit and the output line has
82 * to be broken, start the next line at the right margin instead
83 * of at the offset. Used together with TERMP_NOBREAK for the tags
84 * in various kinds of tagged lists.
85 * - TERMP_DANGLE: Do not break the output line at the right margin,
86 * append the next chunk after it even if this one is too long.
87 * To be used together with TERMP_NOBREAK.
88 * - TERMP_HANG: Like TERMP_DANGLE, and also suppress padding before
89 * the next chunk if this column is not full.
92 term_flushln(struct termp *p)
94 size_t i; /* current input position in p->buf */
95 int ntab; /* number of tabs to prepend */
96 size_t vis; /* current visual position on output */
97 size_t vbl; /* number of blanks to prepend to output */
98 size_t vend; /* end of word visual position on output */
99 size_t bp; /* visual right border position */
100 size_t dv; /* temporary for visual pos calculations */
101 size_t j; /* temporary loop index for p->buf */
102 size_t jhy; /* last hyph before overflow w/r/t j */
103 size_t maxvis; /* output position of visible boundary */
106 * First, establish the maximum columns of "visible" content.
107 * This is usually the difference between the right-margin and
108 * an indentation, but can be, for tagged lists or columns, a
109 * small set of values.
111 * The following unsigned-signed subtractions look strange,
112 * but they are actually correct. If the int p->overstep
113 * is negative, it gets sign extended. Subtracting that
114 * very large size_t effectively adds a small number to dv.
116 dv = p->rmargin > p->offset ? p->rmargin - p->offset : 0;
117 maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0;
119 if (p->flags & TERMP_NOBREAK) {
120 dv = p->maxrmargin > p->offset ?
121 p->maxrmargin - p->offset : 0;
122 bp = (int)dv > p->overstep ?
123 dv - (size_t)p->overstep : 0;
128 * Calculate the required amount of padding.
130 vbl = p->offset + p->overstep > p->viscol ?
131 p->offset + p->overstep - p->viscol : 0;
138 * Handle literal tab characters: collapse all
139 * subsequent tabs into a single huge set of spaces.
142 while (i < p->col && '\t' == p->buf[i]) {
143 vend = (vis / p->tabwidth + 1) * p->tabwidth;
151 * Count up visible word characters. Control sequences
152 * (starting with the CSI) aren't counted. A space
153 * generates a non-printing word, which is valid (the
154 * space is printed according to regular spacing rules).
157 for (j = i, jhy = 0; j < p->col; j++) {
158 if (' ' == p->buf[j] || '\t' == p->buf[j])
161 /* Back over the the last printed character. */
162 if (8 == p->buf[j]) {
164 vend -= (*p->width)(p, p->buf[j - 1]);
169 /* Break at the hyphen point if we overrun. */
170 if (vend > vis && vend < bp &&
171 (ASCII_HYPH == p->buf[j] ||
172 ASCII_BREAK == p->buf[j]))
176 * Hyphenation now decided, put back a real
177 * hyphen such that we get the correct width.
179 if (ASCII_HYPH == p->buf[j])
182 vend += (*p->width)(p, p->buf[j]);
186 * Find out whether we would exceed the right margin.
187 * If so, break to the next line.
189 if (vend > bp && 0 == jhy && vis > 0) {
193 if (TERMP_BRIND & p->flags) {
200 /* use pending tabs on the new line */
203 vbl += ntab * p->tabwidth;
206 * Remove the p->overstep width.
207 * Again, if p->overstep is negative,
208 * sign extension does the right thing.
211 bp += (size_t)p->overstep;
215 /* Write out the [remaining] word. */
216 for ( ; i < p->col; i++) {
217 if (vend > bp && jhy > 0 && i > jhy)
219 if ('\t' == p->buf[i])
221 if (' ' == p->buf[i]) {
223 while (i < p->col && ' ' == p->buf[i])
225 dv = (i - j) * (*p->width)(p, ' ');
230 if (ASCII_NBRSP == p->buf[i]) {
231 vbl += (*p->width)(p, ' ');
234 if (ASCII_BREAK == p->buf[i])
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 (*p->letter)(p, p->buf[i]);
250 p->viscol -= (*p->width)(p, p->buf[i-1]);
252 p->viscol += (*p->width)(p, p->buf[i]);
258 * If there was trailing white space, it was not printed;
259 * so reset the cursor position accordingly.
269 if ( ! (TERMP_NOBREAK & p->flags)) {
275 if (TERMP_HANG & p->flags) {
276 p->overstep += (int)(p->offset + vis - p->rmargin +
277 p->trailspace * (*p->width)(p, ' '));
280 * If we have overstepped the margin, temporarily move
281 * it to the right and flag the rest of the line to be
283 * If there is a request to keep the columns together,
284 * allow negative overstep when the column is not full.
286 if (p->trailspace && p->overstep < 0)
290 } else if (TERMP_DANGLE & p->flags)
293 /* If the column was overrun, break the line. */
294 if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) {
301 * A newline only breaks an existing line; it won't assert vertical
302 * space. All data in the output buffer is flushed prior to the newline
306 term_newln(struct termp *p)
309 p->flags |= TERMP_NOSPACE;
310 if (p->col || p->viscol)
315 * Asserts a vertical space (a full, empty line-break between lines).
316 * Note that if used twice, this will cause two blank spaces and so on.
317 * All data in the output buffer is flushed prior to the newline
321 term_vspace(struct termp *p)
332 /* Swap current and previous font; for \fP and .ft P */
334 term_fontlast(struct termp *p)
339 p->fontl = p->fontq[p->fonti];
340 p->fontq[p->fonti] = f;
343 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
345 term_fontrepl(struct termp *p, enum termfont f)
348 p->fontl = p->fontq[p->fonti];
349 p->fontq[p->fonti] = f;
352 /* Set font, save previous. */
354 term_fontpush(struct termp *p, enum termfont f)
357 p->fontl = p->fontq[p->fonti];
358 if (++p->fonti == p->fontsz) {
360 p->fontq = mandoc_reallocarray(p->fontq,
361 p->fontsz, sizeof(enum termfont *));
363 p->fontq[p->fonti] = f;
366 /* Flush to make the saved pointer current again. */
368 term_fontpopq(struct termp *p, int i)
376 /* Pop one font off the stack. */
378 term_fontpop(struct termp *p)
386 * Handle pwords, partial words, which may be either a single word or a
387 * phrase that cannot be broken down (such as a literal string). This
388 * handles word styling.
391 term_word(struct termp *p, const char *word)
393 const char nbrsp[2] = { ASCII_NBRSP, 0 };
394 const char *seq, *cp;
399 if ( ! (TERMP_NOSPACE & p->flags)) {
400 if ( ! (TERMP_KEEP & p->flags)) {
402 if (TERMP_SENTENCE & p->flags)
405 bufferc(p, ASCII_NBRSP);
407 if (TERMP_PREKEEP & p->flags)
408 p->flags |= TERMP_KEEP;
410 if ( ! (p->flags & TERMP_NONOSPACE))
411 p->flags &= ~TERMP_NOSPACE;
413 p->flags |= TERMP_NOSPACE;
415 p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE);
418 while ('\0' != *word) {
420 if (TERMP_SKIPCHAR & p->flags) {
421 p->flags &= ~TERMP_SKIPCHAR;
425 if (TERMP_NBRWORD & p->flags) {
431 ssz = strcspn(word, "\\ ");
433 ssz = strcspn(word, "\\");
434 encode(p, word, ssz);
440 esc = mandoc_escape(&word, &seq, &sz);
441 if (ESCAPE_ERROR == esc)
446 uc = mchars_num2uc(seq + 1, sz - 1);
448 case ESCAPE_NUMBERED:
449 uc = mchars_num2char(seq, sz);
454 if (p->enc == TERMENC_ASCII) {
455 cp = mchars_spec2str(p->symtab,
460 uc = mchars_spec2cp(p->symtab, seq, sz);
465 case ESCAPE_FONTBOLD:
466 term_fontrepl(p, TERMFONT_BOLD);
468 case ESCAPE_FONTITALIC:
469 term_fontrepl(p, TERMFONT_UNDER);
472 term_fontrepl(p, TERMFONT_BI);
476 case ESCAPE_FONTROMAN:
477 term_fontrepl(p, TERMFONT_NONE);
479 case ESCAPE_FONTPREV:
483 if (TERMP_SKIPCHAR & p->flags)
484 p->flags &= ~TERMP_SKIPCHAR;
485 else if ('\0' == *word)
486 p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
488 case ESCAPE_SKIPCHAR:
489 p->flags |= TERMP_SKIPCHAR;
491 case ESCAPE_OVERSTRIKE:
495 mandoc_escape(&seq, NULL, NULL);
507 * Common handling for Unicode and numbered
508 * character escape sequences.
511 if (p->enc == TERMENC_ASCII) {
512 cp = ascii_uc2str(uc);
513 encode(p, cp, strlen(cp));
515 if ((uc < 0x20 && uc != 0x09) ||
516 (uc > 0x7E && uc < 0xA0))
521 p->flags &= ~TERMP_NBRWORD;
525 adjbuf(struct termp *p, size_t sz)
530 while (sz >= p->maxcols)
533 p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int));
537 bufferc(struct termp *p, char c)
540 if (p->col + 1 >= p->maxcols)
541 adjbuf(p, p->col + 1);
543 p->buf[p->col++] = c;
548 * Do this for a single (probably unicode) value.
549 * Does not check for non-decorated glyphs.
552 encode1(struct termp *p, int c)
556 if (TERMP_SKIPCHAR & p->flags) {
557 p->flags &= ~TERMP_SKIPCHAR;
561 if (p->col + 6 >= p->maxcols)
562 adjbuf(p, p->col + 6);
564 f = p->fontq[p->fonti];
566 if (TERMFONT_UNDER == f || TERMFONT_BI == f) {
567 p->buf[p->col++] = '_';
568 p->buf[p->col++] = 8;
570 if (TERMFONT_BOLD == f || TERMFONT_BI == f) {
572 p->buf[p->col++] = '-';
574 p->buf[p->col++] = c;
575 p->buf[p->col++] = 8;
577 p->buf[p->col++] = c;
581 encode(struct termp *p, const char *word, size_t sz)
585 if (TERMP_SKIPCHAR & p->flags) {
586 p->flags &= ~TERMP_SKIPCHAR;
591 * Encode and buffer a string of characters. If the current
592 * font mode is unset, buffer directly, else encode then buffer
593 * character by character.
596 if (p->fontq[p->fonti] == TERMFONT_NONE) {
597 if (p->col + sz >= p->maxcols)
598 adjbuf(p, p->col + sz);
599 for (i = 0; i < sz; i++)
600 p->buf[p->col++] = word[i];
604 /* Pre-buffer, assuming worst-case. */
606 if (p->col + 1 + (sz * 5) >= p->maxcols)
607 adjbuf(p, p->col + 1 + (sz * 5));
609 for (i = 0; i < sz; i++) {
610 if (ASCII_HYPH == word[i] ||
611 isgraph((unsigned char)word[i]))
614 p->buf[p->col++] = word[i];
619 term_setwidth(struct termp *p, const char *wstr)
640 if (a2roffsu(wstr, &su, SCALE_MAX))
641 width = term_hspan(p, &su);
645 (*p->setwidth)(p, iop, width);
649 term_len(const struct termp *p, size_t sz)
652 return((*p->width)(p, ' ') * sz);
656 cond_width(const struct termp *p, int c, int *skip)
663 return((*p->width)(p, c));
667 term_strlen(const struct termp *p, const char *cp)
671 const char *seq, *rhs;
673 static const char rej[] = { '\\', ASCII_NBRSP, ASCII_HYPH,
677 * Account for escaped sequences within string length
678 * calculations. This follows the logic in term_word() as we
679 * must calculate the width of produced strings.
684 while ('\0' != *cp) {
685 rsz = strcspn(cp, rej);
686 for (i = 0; i < rsz; i++)
687 sz += cond_width(p, *cp++, &skip);
692 esc = mandoc_escape(&cp, &seq, &ssz);
693 if (ESCAPE_ERROR == esc)
700 uc = mchars_num2uc(seq + 1, ssz - 1);
702 case ESCAPE_NUMBERED:
703 uc = mchars_num2char(seq, ssz);
708 if (p->enc == TERMENC_ASCII) {
709 rhs = mchars_spec2str(p->symtab,
714 uc = mchars_spec2cp(p->symtab,
717 sz += cond_width(p, uc, &skip);
720 case ESCAPE_SKIPCHAR:
723 case ESCAPE_OVERSTRIKE:
728 mandoc_escape(&seq, NULL, NULL);
731 i = (*p->width)(p, *seq++);
742 * Common handling for Unicode and numbered
743 * character escape sequences.
747 if (p->enc == TERMENC_ASCII) {
748 rhs = ascii_uc2str(uc);
751 if ((uc < 0x20 && uc != 0x09) ||
752 (uc > 0x7E && uc < 0xA0))
754 sz += cond_width(p, uc, &skip);
765 * Common handling for all escape sequences
766 * printing more than one character.
769 for (i = 0; i < rsz; i++)
770 sz += (*p->width)(p, *rhs++);
773 sz += cond_width(p, ' ', &skip);
777 sz += cond_width(p, '-', &skip);
791 term_vspan(const struct termp *p, const struct roffsu *su)
798 r = su->scale / 40.0;
801 r = su->scale * 6.0 / 2.54;
804 r = su->scale * 65536.0 / 40.0;
810 r = su->scale * 0.006;
816 r = su->scale / 12.0;
830 ri = r > 0.0 ? r + 0.4995 : r - 0.4995;
831 return(ri < 66 ? ri : 1);
835 term_hspan(const struct termp *p, const struct roffsu *su)
839 v = (*p->hspan)(p, su);
840 return(v > 0.0 ? v + 0.0005 : v - 0.0005);