xref: /dflybsd-src/contrib/mdocml/out.c (revision 0eb2eccd5a86ef7dd7492d2651de55c3589f23d7)
1 /*	$Id: out.c,v 1.39 2011/03/17 08:49:34 kristaps Exp $ */
2 /*
3  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
5  *
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.
9  *
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.
17  */
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21 
22 #include <sys/types.h>
23 
24 #include <assert.h>
25 #include <ctype.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <time.h>
30 
31 #include "mandoc.h"
32 #include "out.h"
33 
34 static	void	tblcalc_data(struct rofftbl *, struct roffcol *,
35 			const struct tbl *, const struct tbl_dat *);
36 static	void	tblcalc_literal(struct rofftbl *, struct roffcol *,
37 			const struct tbl_dat *);
38 static	void	tblcalc_number(struct rofftbl *, struct roffcol *,
39 			const struct tbl *, const struct tbl_dat *);
40 
41 /*
42  * Convert a `scaling unit' to a consistent form, or fail.  Scaling
43  * units are documented in groff.7, mdoc.7, man.7.
44  */
45 int
46 a2roffsu(const char *src, struct roffsu *dst, enum roffscale def)
47 {
48 	char		 buf[BUFSIZ], hasd;
49 	int		 i;
50 	enum roffscale	 unit;
51 
52 	if ('\0' == *src)
53 		return(0);
54 
55 	i = hasd = 0;
56 
57 	switch (*src) {
58 	case ('+'):
59 		src++;
60 		break;
61 	case ('-'):
62 		buf[i++] = *src++;
63 		break;
64 	default:
65 		break;
66 	}
67 
68 	if ('\0' == *src)
69 		return(0);
70 
71 	while (i < BUFSIZ) {
72 		if ( ! isdigit((u_char)*src)) {
73 			if ('.' != *src)
74 				break;
75 			else if (hasd)
76 				break;
77 			else
78 				hasd = 1;
79 		}
80 		buf[i++] = *src++;
81 	}
82 
83 	if (BUFSIZ == i || (*src && *(src + 1)))
84 		return(0);
85 
86 	buf[i] = '\0';
87 
88 	switch (*src) {
89 	case ('c'):
90 		unit = SCALE_CM;
91 		break;
92 	case ('i'):
93 		unit = SCALE_IN;
94 		break;
95 	case ('P'):
96 		unit = SCALE_PC;
97 		break;
98 	case ('p'):
99 		unit = SCALE_PT;
100 		break;
101 	case ('f'):
102 		unit = SCALE_FS;
103 		break;
104 	case ('v'):
105 		unit = SCALE_VS;
106 		break;
107 	case ('m'):
108 		unit = SCALE_EM;
109 		break;
110 	case ('\0'):
111 		if (SCALE_MAX == def)
112 			return(0);
113 		unit = SCALE_BU;
114 		break;
115 	case ('u'):
116 		unit = SCALE_BU;
117 		break;
118 	case ('M'):
119 		unit = SCALE_MM;
120 		break;
121 	case ('n'):
122 		unit = SCALE_EN;
123 		break;
124 	default:
125 		return(0);
126 	}
127 
128 	/* FIXME: do this in the caller. */
129 	if ((dst->scale = atof(buf)) < 0)
130 		dst->scale = 0;
131 	dst->unit = unit;
132 	return(1);
133 }
134 
135 
136 /*
137  * Correctly writes the time in nroff form, which differs from standard
138  * form in that a space isn't printed in lieu of the extra %e field for
139  * single-digit dates.
140  */
141 void
142 time2a(time_t t, char *dst, size_t sz)
143 {
144 	struct tm	 tm;
145 	char		 buf[5];
146 	char		*p;
147 	size_t		 nsz;
148 
149 	assert(sz > 1);
150 	localtime_r(&t, &tm);
151 
152 	p = dst;
153 	nsz = 0;
154 
155 	dst[0] = '\0';
156 
157 	if (0 == (nsz = strftime(p, sz, "%B ", &tm)))
158 		return;
159 
160 	p += (int)nsz;
161 	sz -= nsz;
162 
163 	if (0 == strftime(buf, sizeof(buf), "%e, ", &tm))
164 		return;
165 
166 	nsz = strlcat(p, buf + (' ' == buf[0] ? 1 : 0), sz);
167 
168 	if (nsz >= sz)
169 		return;
170 
171 	p += (int)nsz;
172 	sz -= nsz;
173 
174 	(void)strftime(p, sz, "%Y", &tm);
175 }
176 
177 
178 int
179 a2roffdeco(enum roffdeco *d, const char **word, size_t *sz)
180 {
181 	int		 i, j, lim;
182 	char		 term, c;
183 	const char	*wp;
184 	enum roffdeco	 dd;
185 
186 	*d = DECO_NONE;
187 	lim = i = 0;
188 	term = '\0';
189 	wp = *word;
190 
191 	switch ((c = wp[i++])) {
192 	case ('('):
193 		*d = DECO_SPECIAL;
194 		lim = 2;
195 		break;
196 	case ('F'):
197 		/* FALLTHROUGH */
198 	case ('f'):
199 		*d = 'F' == c ? DECO_FFONT : DECO_FONT;
200 
201 		switch (wp[i++]) {
202 		case ('('):
203 			lim = 2;
204 			break;
205 		case ('['):
206 			term = ']';
207 			break;
208 		case ('3'):
209 			/* FALLTHROUGH */
210 		case ('B'):
211 			*d = DECO_BOLD;
212 			return(i);
213 		case ('2'):
214 			/* FALLTHROUGH */
215 		case ('I'):
216 			*d = DECO_ITALIC;
217 			return(i);
218 		case ('P'):
219 			*d = DECO_PREVIOUS;
220 			return(i);
221 		case ('1'):
222 			/* FALLTHROUGH */
223 		case ('R'):
224 			*d = DECO_ROMAN;
225 			return(i);
226 		default:
227 			i--;
228 			lim = 1;
229 			break;
230 		}
231 		break;
232 	case ('k'):
233 		/* FALLTHROUGH */
234 	case ('M'):
235 		/* FALLTHROUGH */
236 	case ('m'):
237 		/* FALLTHROUGH */
238 	case ('*'):
239 		if ('*' == c)
240 			*d = DECO_RESERVED;
241 
242 		switch (wp[i++]) {
243 		case ('('):
244 			lim = 2;
245 			break;
246 		case ('['):
247 			term = ']';
248 			break;
249 		default:
250 			i--;
251 			lim = 1;
252 			break;
253 		}
254 		break;
255 
256 	case ('N'):
257 
258 		/*
259 		 * Sequence of characters:  backslash,  'N' (i = 0),
260 		 * starting delimiter (i = 1), character number (i = 2).
261 		 */
262 
263 		*word = wp + 2;
264 		*sz = 0;
265 
266 		/*
267 		 * Cannot use a digit as a starting delimiter;
268 		 * but skip the digit anyway.
269 		 */
270 
271 		if (isdigit((int)wp[1]))
272 			return(2);
273 
274 		/*
275 		 * Any non-digit terminates the character number.
276 		 * That is, the terminating delimiter need not
277 		 * match the starting delimiter.
278 		 */
279 
280 		for (i = 2; isdigit((int)wp[i]); i++)
281 			(*sz)++;
282 
283 		/*
284 		 * This is only a numbered character
285 		 * if the character number has at least one digit.
286 		 */
287 
288 		if (*sz)
289 			*d = DECO_NUMBERED;
290 
291 		/*
292 		 * Skip the terminating delimiter, even if it does not
293 		 * match, and even if there is no character number.
294 		 */
295 
296 		return(++i);
297 
298 	case ('h'):
299 		/* FALLTHROUGH */
300 	case ('v'):
301 		/* FALLTHROUGH */
302 	case ('s'):
303 		j = 0;
304 		if ('+' == wp[i] || '-' == wp[i]) {
305 			i++;
306 			j = 1;
307 		}
308 
309 		switch (wp[i++]) {
310 		case ('('):
311 			lim = 2;
312 			break;
313 		case ('['):
314 			term = ']';
315 			break;
316 		case ('\''):
317 			term = '\'';
318 			break;
319 		case ('0'):
320 			j = 1;
321 			/* FALLTHROUGH */
322 		default:
323 			i--;
324 			lim = 1;
325 			break;
326 		}
327 
328 		if ('+' == wp[i] || '-' == wp[i]) {
329 			if (j)
330 				return(i);
331 			i++;
332 		}
333 
334 		/* Handle embedded numerical subexp or escape. */
335 
336 		if ('(' == wp[i]) {
337 			while (wp[i] && ')' != wp[i])
338 				if ('\\' == wp[i++]) {
339 					/* Handle embedded escape. */
340 					*word = &wp[i];
341 					i += a2roffdeco(&dd, word, sz);
342 				}
343 
344 			if (')' == wp[i++])
345 				break;
346 
347 			*d = DECO_NONE;
348 			return(i - 1);
349 		} else if ('\\' == wp[i]) {
350 			*word = &wp[++i];
351 			i += a2roffdeco(&dd, word, sz);
352 		}
353 
354 		break;
355 	case ('['):
356 		*d = DECO_SPECIAL;
357 		term = ']';
358 		break;
359 	case ('c'):
360 		*d = DECO_NOSPACE;
361 		return(i);
362 	case ('z'):
363 		*d = DECO_NONE;
364 		if ('\\' == wp[i]) {
365 			*word = &wp[++i];
366 			return(i + a2roffdeco(&dd, word, sz));
367 		} else
368 			lim = 1;
369 		break;
370 	case ('o'):
371 		/* FALLTHROUGH */
372 	case ('w'):
373 		if ('\'' == wp[i++]) {
374 			term = '\'';
375 			break;
376 		}
377 		/* FALLTHROUGH */
378 	default:
379 		*d = DECO_SSPECIAL;
380 		i--;
381 		lim = 1;
382 		break;
383 	}
384 
385 	assert(term || lim);
386 	*word = &wp[i];
387 
388 	if (term) {
389 		j = i;
390 		while (wp[i] && wp[i] != term)
391 			i++;
392 		if ('\0' == wp[i]) {
393 			*d = DECO_NONE;
394 			return(i);
395 		}
396 
397 		assert(i >= j);
398 		*sz = (size_t)(i - j);
399 
400 		return(i + 1);
401 	}
402 
403 	assert(lim > 0);
404 	*sz = (size_t)lim;
405 
406 	for (j = 0; wp[i] && j < lim; j++)
407 		i++;
408 	if (j < lim)
409 		*d = DECO_NONE;
410 
411 	return(i);
412 }
413 
414 /*
415  * Calculate the abstract widths and decimal positions of columns in a
416  * table.  This routine allocates the columns structures then runs over
417  * all rows and cells in the table.  The function pointers in "tbl" are
418  * used for the actual width calculations.
419  */
420 void
421 tblcalc(struct rofftbl *tbl, const struct tbl_span *sp)
422 {
423 	const struct tbl_dat	*dp;
424 	const struct tbl_head	*hp;
425 	struct roffcol		*col;
426 
427 	/*
428 	 * Allocate the master column specifiers.  These will hold the
429 	 * widths and decimal positions for all cells in the column.  It
430 	 * must be freed and nullified by the caller.
431 	 */
432 
433 	assert(NULL == tbl->cols);
434 	tbl->cols = mandoc_calloc
435 		((size_t)sp->tbl->cols, sizeof(struct roffcol));
436 
437 	hp = sp->head;
438 
439 	for ( ; sp; sp = sp->next) {
440 		if (TBL_SPAN_DATA != sp->pos)
441 			continue;
442 		/*
443 		 * Account for the data cells in the layout, matching it
444 		 * to data cells in the data section.
445 		 */
446 		for (dp = sp->first; dp; dp = dp->next) {
447 			assert(dp->layout);
448 			col = &tbl->cols[dp->layout->head->ident];
449 			tblcalc_data(tbl, col, sp->tbl, dp);
450 		}
451 	}
452 
453 	/*
454 	 * Calculate width of the spanners.  These get one space for a
455 	 * vertical line, two for a double-vertical line.
456 	 */
457 
458 	for ( ; hp; hp = hp->next) {
459 		col = &tbl->cols[hp->ident];
460 		switch (hp->pos) {
461 		case (TBL_HEAD_VERT):
462 			col->width = (*tbl->len)(1, tbl->arg);
463 			break;
464 		case (TBL_HEAD_DVERT):
465 			col->width = (*tbl->len)(2, tbl->arg);
466 			break;
467 		default:
468 			break;
469 		}
470 	}
471 }
472 
473 static void
474 tblcalc_data(struct rofftbl *tbl, struct roffcol *col,
475 		const struct tbl *tp, const struct tbl_dat *dp)
476 {
477 	size_t		 sz;
478 
479 	/* Branch down into data sub-types. */
480 
481 	switch (dp->layout->pos) {
482 	case (TBL_CELL_HORIZ):
483 		/* FALLTHROUGH */
484 	case (TBL_CELL_DHORIZ):
485 		sz = (*tbl->len)(1, tbl->arg);
486 		if (col->width < sz)
487 			col->width = sz;
488 		break;
489 	case (TBL_CELL_LONG):
490 		/* FALLTHROUGH */
491 	case (TBL_CELL_CENTRE):
492 		/* FALLTHROUGH */
493 	case (TBL_CELL_LEFT):
494 		/* FALLTHROUGH */
495 	case (TBL_CELL_RIGHT):
496 		tblcalc_literal(tbl, col, dp);
497 		break;
498 	case (TBL_CELL_NUMBER):
499 		tblcalc_number(tbl, col, tp, dp);
500 		break;
501 	case (TBL_CELL_DOWN):
502 		break;
503 	default:
504 		abort();
505 		/* NOTREACHED */
506 	}
507 }
508 
509 static void
510 tblcalc_literal(struct rofftbl *tbl, struct roffcol *col,
511 		const struct tbl_dat *dp)
512 {
513 	size_t		 sz, bufsz, spsz;
514 	const char	*str;
515 
516 	/*
517 	 * Calculate our width and use the spacing, with a minimum
518 	 * spacing dictated by position (centre, e.g,. gets a space on
519 	 * either side, while right/left get a single adjacent space).
520 	 */
521 
522 	bufsz = spsz = 0;
523 	str = dp->string ? dp->string : "";
524 	sz = (*tbl->slen)(str, tbl->arg);
525 
526 	/* FIXME: TBL_DATA_HORIZ et al.? */
527 
528 	assert(dp->layout);
529 	switch (dp->layout->pos) {
530 	case (TBL_CELL_LONG):
531 		/* FALLTHROUGH */
532 	case (TBL_CELL_CENTRE):
533 		bufsz = (*tbl->len)(1, tbl->arg);
534 		break;
535 	default:
536 		bufsz = (*tbl->len)(1, tbl->arg);
537 		break;
538 	}
539 
540 	if (dp->layout->spacing) {
541 		spsz = (*tbl->len)(dp->layout->spacing, tbl->arg);
542 		bufsz = bufsz > spsz ? bufsz : spsz;
543 	}
544 
545 	sz += bufsz;
546 	if (col->width < sz)
547 		col->width = sz;
548 }
549 
550 static void
551 tblcalc_number(struct rofftbl *tbl, struct roffcol *col,
552 		const struct tbl *tp, const struct tbl_dat *dp)
553 {
554 	int 		 i;
555 	size_t		 sz, psz, ssz, d;
556 	const char	*str;
557 	char		*cp;
558 	char		 buf[2];
559 
560 	/*
561 	 * First calculate number width and decimal place (last + 1 for
562 	 * no-decimal numbers).  If the stored decimal is subsequent
563 	 * ours, make our size longer by that difference
564 	 * (right-"shifting"); similarly, if ours is subsequent the
565 	 * stored, then extend the stored size by the difference.
566 	 * Finally, re-assign the stored values.
567 	 */
568 
569 	str = dp->string ? dp->string : "";
570 	sz = (*tbl->slen)(str, tbl->arg);
571 
572 	/* FIXME: TBL_DATA_HORIZ et al.? */
573 
574 	buf[0] = tp->decimal;
575 	buf[1] = '\0';
576 
577 	psz = (*tbl->slen)(buf, tbl->arg);
578 
579 	if (NULL != (cp = strrchr(str, tp->decimal))) {
580 		buf[1] = '\0';
581 		for (ssz = 0, i = 0; cp != &str[i]; i++) {
582 			buf[0] = str[i];
583 			ssz += (*tbl->slen)(buf, tbl->arg);
584 		}
585 		d = ssz + psz;
586 	} else
587 		d = sz + psz;
588 
589 	/* Padding. */
590 
591 	sz += (*tbl->len)(2, tbl->arg);
592 	d += (*tbl->len)(1, tbl->arg);
593 
594 	/* Adjust the settings for this column. */
595 
596 	if (col->decimal > d) {
597 		sz += col->decimal - d;
598 		d = col->decimal;
599 	} else
600 		col->width += d - col->decimal;
601 
602 	if (sz > col->width)
603 		col->width = sz;
604 	if (d > col->decimal)
605 		col->decimal = d;
606 
607 	/* Adjust for stipulated width. */
608 
609 	if (col->width < dp->layout->spacing)
610 		col->width = dp->layout->spacing;
611 }
612 
613 
614