]>
git.cameronkatri.com Git - mandoc.git/blob - term.c
1 /* $Id: term.c,v 1.248 2015/04/29 18:35:00 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_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.
268 p
->flags
&= ~(TERMP_BACKAFTER
| TERMP_BACKBEFORE
);
270 if ( ! (TERMP_NOBREAK
& p
->flags
)) {
276 if (TERMP_HANG
& p
->flags
) {
277 p
->overstep
+= (int)(p
->offset
+ vis
- p
->rmargin
+
278 p
->trailspace
* (*p
->width
)(p
, ' '));
281 * If we have overstepped the margin, temporarily move
282 * it to the right and flag the rest of the line to be
284 * If there is a request to keep the columns together,
285 * allow negative overstep when the column is not full.
287 if (p
->trailspace
&& p
->overstep
< 0)
291 } else if (TERMP_DANGLE
& p
->flags
)
294 /* If the column was overrun, break the line. */
295 if (maxvis
< vis
+ p
->trailspace
* (*p
->width
)(p
, ' ')) {
302 * A newline only breaks an existing line; it won't assert vertical
303 * space. All data in the output buffer is flushed prior to the newline
307 term_newln(struct termp
*p
)
310 p
->flags
|= TERMP_NOSPACE
;
311 if (p
->col
|| p
->viscol
)
316 * Asserts a vertical space (a full, empty line-break between lines).
317 * Note that if used twice, this will cause two blank spaces and so on.
318 * All data in the output buffer is flushed prior to the newline
322 term_vspace(struct termp
*p
)
333 /* Swap current and previous font; for \fP and .ft P */
335 term_fontlast(struct termp
*p
)
340 p
->fontl
= p
->fontq
[p
->fonti
];
341 p
->fontq
[p
->fonti
] = f
;
344 /* Set font, save current, discard previous; for \f, .ft, .B etc. */
346 term_fontrepl(struct termp
*p
, enum termfont f
)
349 p
->fontl
= p
->fontq
[p
->fonti
];
350 p
->fontq
[p
->fonti
] = f
;
353 /* Set font, save previous. */
355 term_fontpush(struct termp
*p
, enum termfont f
)
358 p
->fontl
= p
->fontq
[p
->fonti
];
359 if (++p
->fonti
== p
->fontsz
) {
361 p
->fontq
= mandoc_reallocarray(p
->fontq
,
362 p
->fontsz
, sizeof(enum termfont
*));
364 p
->fontq
[p
->fonti
] = f
;
367 /* Flush to make the saved pointer current again. */
369 term_fontpopq(struct termp
*p
, int i
)
377 /* Pop one font off the stack. */
379 term_fontpop(struct termp
*p
)
387 * Handle pwords, partial words, which may be either a single word or a
388 * phrase that cannot be broken down (such as a literal string). This
389 * handles word styling.
392 term_word(struct termp
*p
, const char *word
)
394 const char nbrsp
[2] = { ASCII_NBRSP
, 0 };
395 const char *seq
, *cp
;
400 if ( ! (TERMP_NOSPACE
& p
->flags
)) {
401 if ( ! (TERMP_KEEP
& p
->flags
)) {
403 if (TERMP_SENTENCE
& p
->flags
)
406 bufferc(p
, ASCII_NBRSP
);
408 if (TERMP_PREKEEP
& p
->flags
)
409 p
->flags
|= TERMP_KEEP
;
411 if ( ! (p
->flags
& TERMP_NONOSPACE
))
412 p
->flags
&= ~TERMP_NOSPACE
;
414 p
->flags
|= TERMP_NOSPACE
;
416 p
->flags
&= ~(TERMP_SENTENCE
| TERMP_NONEWLINE
);
419 while ('\0' != *word
) {
421 if (TERMP_NBRWORD
& p
->flags
) {
427 ssz
= strcspn(word
, "\\ ");
429 ssz
= strcspn(word
, "\\");
430 encode(p
, word
, ssz
);
436 esc
= mandoc_escape(&word
, &seq
, &sz
);
437 if (ESCAPE_ERROR
== esc
)
442 uc
= mchars_num2uc(seq
+ 1, sz
- 1);
444 case ESCAPE_NUMBERED
:
445 uc
= mchars_num2char(seq
, sz
);
450 if (p
->enc
== TERMENC_ASCII
) {
451 cp
= mchars_spec2str(p
->symtab
,
456 uc
= mchars_spec2cp(p
->symtab
, seq
, sz
);
461 case ESCAPE_FONTBOLD
:
462 term_fontrepl(p
, TERMFONT_BOLD
);
464 case ESCAPE_FONTITALIC
:
465 term_fontrepl(p
, TERMFONT_UNDER
);
468 term_fontrepl(p
, TERMFONT_BI
);
472 case ESCAPE_FONTROMAN
:
473 term_fontrepl(p
, TERMFONT_NONE
);
475 case ESCAPE_FONTPREV
:
479 if (p
->flags
& TERMP_BACKAFTER
)
480 p
->flags
&= ~TERMP_BACKAFTER
;
481 else if (*word
== '\0')
482 p
->flags
|= (TERMP_NOSPACE
| TERMP_NONEWLINE
);
484 case ESCAPE_SKIPCHAR
:
485 p
->flags
|= TERMP_BACKAFTER
;
487 case ESCAPE_OVERSTRIKE
:
491 mandoc_escape(&seq
, NULL
, NULL
);
496 if (p
->flags
& TERMP_BACKBEFORE
)
497 p
->flags
|= TERMP_BACKAFTER
;
499 p
->flags
|= TERMP_BACKBEFORE
;
508 * Common handling for Unicode and numbered
509 * character escape sequences.
512 if (p
->enc
== TERMENC_ASCII
) {
513 cp
= ascii_uc2str(uc
);
514 encode(p
, cp
, strlen(cp
));
516 if ((uc
< 0x20 && uc
!= 0x09) ||
517 (uc
> 0x7E && uc
< 0xA0))
522 p
->flags
&= ~TERMP_NBRWORD
;
526 adjbuf(struct termp
*p
, size_t sz
)
531 while (sz
>= p
->maxcols
)
534 p
->buf
= mandoc_reallocarray(p
->buf
, p
->maxcols
, sizeof(int));
538 bufferc(struct termp
*p
, char c
)
541 if (p
->col
+ 1 >= p
->maxcols
)
542 adjbuf(p
, p
->col
+ 1);
544 p
->buf
[p
->col
++] = c
;
549 * Do this for a single (probably unicode) value.
550 * Does not check for non-decorated glyphs.
553 encode1(struct termp
*p
, int c
)
557 if (p
->col
+ 7 >= p
->maxcols
)
558 adjbuf(p
, p
->col
+ 7);
560 f
= (c
== ASCII_HYPH
|| isgraph(c
)) ?
561 p
->fontq
[p
->fonti
] : TERMFONT_NONE
;
563 if (p
->flags
& TERMP_BACKBEFORE
) {
564 p
->buf
[p
->col
++] = 8;
565 p
->flags
&= ~TERMP_BACKBEFORE
;
567 if (TERMFONT_UNDER
== f
|| TERMFONT_BI
== f
) {
568 p
->buf
[p
->col
++] = '_';
569 p
->buf
[p
->col
++] = 8;
571 if (TERMFONT_BOLD
== f
|| TERMFONT_BI
== f
) {
573 p
->buf
[p
->col
++] = '-';
575 p
->buf
[p
->col
++] = c
;
576 p
->buf
[p
->col
++] = 8;
578 p
->buf
[p
->col
++] = c
;
579 if (p
->flags
& TERMP_BACKAFTER
) {
580 p
->flags
|= TERMP_BACKBEFORE
;
581 p
->flags
&= ~TERMP_BACKAFTER
;
586 encode(struct termp
*p
, const char *word
, size_t sz
)
590 if (p
->col
+ 2 + (sz
* 5) >= p
->maxcols
)
591 adjbuf(p
, p
->col
+ 2 + (sz
* 5));
593 for (i
= 0; i
< sz
; i
++) {
594 if (ASCII_HYPH
== word
[i
] ||
595 isgraph((unsigned char)word
[i
]))
598 p
->buf
[p
->col
++] = word
[i
];
603 term_setwidth(struct termp
*p
, const char *wstr
)
623 if (a2roffsu(wstr
, &su
, SCALE_MAX
))
624 width
= term_hspan(p
, &su
);
628 (*p
->setwidth
)(p
, iop
, width
);
632 term_len(const struct termp
*p
, size_t sz
)
635 return((*p
->width
)(p
, ' ') * sz
);
639 cond_width(const struct termp
*p
, int c
, int *skip
)
646 return((*p
->width
)(p
, c
));
650 term_strlen(const struct termp
*p
, const char *cp
)
654 const char *seq
, *rhs
;
656 static const char rej
[] = { '\\', ASCII_NBRSP
, ASCII_HYPH
,
660 * Account for escaped sequences within string length
661 * calculations. This follows the logic in term_word() as we
662 * must calculate the width of produced strings.
667 while ('\0' != *cp
) {
668 rsz
= strcspn(cp
, rej
);
669 for (i
= 0; i
< rsz
; i
++)
670 sz
+= cond_width(p
, *cp
++, &skip
);
675 esc
= mandoc_escape(&cp
, &seq
, &ssz
);
676 if (ESCAPE_ERROR
== esc
)
683 uc
= mchars_num2uc(seq
+ 1, ssz
- 1);
685 case ESCAPE_NUMBERED
:
686 uc
= mchars_num2char(seq
, ssz
);
691 if (p
->enc
== TERMENC_ASCII
) {
692 rhs
= mchars_spec2str(p
->symtab
,
697 uc
= mchars_spec2cp(p
->symtab
,
700 sz
+= cond_width(p
, uc
, &skip
);
703 case ESCAPE_SKIPCHAR
:
706 case ESCAPE_OVERSTRIKE
:
711 mandoc_escape(&seq
, NULL
, NULL
);
714 i
= (*p
->width
)(p
, *seq
++);
725 * Common handling for Unicode and numbered
726 * character escape sequences.
730 if (p
->enc
== TERMENC_ASCII
) {
731 rhs
= ascii_uc2str(uc
);
734 if ((uc
< 0x20 && uc
!= 0x09) ||
735 (uc
> 0x7E && uc
< 0xA0))
737 sz
+= cond_width(p
, uc
, &skip
);
748 * Common handling for all escape sequences
749 * printing more than one character.
752 for (i
= 0; i
< rsz
; i
++)
753 sz
+= (*p
->width
)(p
, *rhs
++);
756 sz
+= cond_width(p
, ' ', &skip
);
760 sz
+= cond_width(p
, '-', &skip
);
774 term_vspan(const struct termp
*p
, const struct roffsu
*su
)
781 r
= su
->scale
/ 40.0;
784 r
= su
->scale
* 6.0 / 2.54;
787 r
= su
->scale
* 65536.0 / 40.0;
793 r
= su
->scale
* 0.006;
799 r
= su
->scale
/ 12.0;
813 ri
= r
> 0.0 ? r
+ 0.4995 : r
- 0.4995;
814 return(ri
< 66 ? ri
: 1);
818 * Convert a scaling width to basic units, rounding down.
821 term_hspan(const struct termp
*p
, const struct roffsu
*su
)
824 return((*p
->hspan
)(p
, su
));