xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision be38755c412cc72cb8d40f51ea70c9893196afff)
1 /*	$Id: man_html.c,v 1.42 2011/09/18 15:54:48 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 #include <sys/types.h>
18 
19 #include <assert.h>
20 #include <ctype.h>
21 #include <stdio.h>
22 #include <stdlib.h>
23 #include <string.h>
24 
25 #include "mandoc.h"
26 #include "out.h"
27 #include "html.h"
28 #include "man.h"
29 #include "main.h"
30 
31 /* TODO: preserve ident widths. */
32 /* FIXME: have PD set the default vspace width. */
33 
34 #define	INDENT		  5
35 
36 #define	MAN_ARGS	  const struct man_meta *m, \
37 			  const struct man_node *n, \
38 			  struct mhtml *mh, \
39 			  struct html *h
40 
41 struct	mhtml {
42 	int		  fl;
43 #define	MANH_LITERAL	 (1 << 0) /* literal context */
44 };
45 
46 struct	htmlman {
47 	int		(*pre)(MAN_ARGS);
48 	int		(*post)(MAN_ARGS);
49 };
50 
51 static	void		  print_bvspace(struct html *,
52 				const struct man_node *);
53 static	void		  print_man(MAN_ARGS);
54 static	void		  print_man_head(MAN_ARGS);
55 static	void		  print_man_nodelist(MAN_ARGS);
56 static	void		  print_man_node(MAN_ARGS);
57 
58 static	int		  a2width(const struct man_node *,
59 				struct roffsu *);
60 
61 static	int		  man_alt_pre(MAN_ARGS);
62 static	int		  man_br_pre(MAN_ARGS);
63 static	int		  man_ign_pre(MAN_ARGS);
64 static	int		  man_in_pre(MAN_ARGS);
65 static	int		  man_literal_pre(MAN_ARGS);
66 static	void		  man_root_post(MAN_ARGS);
67 static	void		  man_root_pre(MAN_ARGS);
68 static	int		  man_B_pre(MAN_ARGS);
69 static	int		  man_HP_pre(MAN_ARGS);
70 static	int		  man_I_pre(MAN_ARGS);
71 static	int		  man_IP_pre(MAN_ARGS);
72 static	int		  man_PP_pre(MAN_ARGS);
73 static	int		  man_RS_pre(MAN_ARGS);
74 static	int		  man_SH_pre(MAN_ARGS);
75 static	int		  man_SM_pre(MAN_ARGS);
76 static	int		  man_SS_pre(MAN_ARGS);
77 
78 static	const struct htmlman mans[MAN_MAX] = {
79 	{ man_br_pre, NULL }, /* br */
80 	{ NULL, NULL }, /* TH */
81 	{ man_SH_pre, NULL }, /* SH */
82 	{ man_SS_pre, NULL }, /* SS */
83 	{ man_IP_pre, NULL }, /* TP */
84 	{ man_PP_pre, NULL }, /* LP */
85 	{ man_PP_pre, NULL }, /* PP */
86 	{ man_PP_pre, NULL }, /* P */
87 	{ man_IP_pre, NULL }, /* IP */
88 	{ man_HP_pre, NULL }, /* HP */
89 	{ man_SM_pre, NULL }, /* SM */
90 	{ man_SM_pre, NULL }, /* SB */
91 	{ man_alt_pre, NULL }, /* BI */
92 	{ man_alt_pre, NULL }, /* IB */
93 	{ man_alt_pre, NULL }, /* BR */
94 	{ man_alt_pre, NULL }, /* RB */
95 	{ NULL, NULL }, /* R */
96 	{ man_B_pre, NULL }, /* B */
97 	{ man_I_pre, NULL }, /* I */
98 	{ man_alt_pre, NULL }, /* IR */
99 	{ man_alt_pre, NULL }, /* RI */
100 	{ man_ign_pre, NULL }, /* na */
101 	{ man_br_pre, NULL }, /* sp */
102 	{ man_literal_pre, NULL }, /* nf */
103 	{ man_literal_pre, NULL }, /* fi */
104 	{ NULL, NULL }, /* RE */
105 	{ man_RS_pre, NULL }, /* RS */
106 	{ man_ign_pre, NULL }, /* DT */
107 	{ man_ign_pre, NULL }, /* UC */
108 	{ man_ign_pre, NULL }, /* PD */
109 	{ man_ign_pre, NULL }, /* AT */
110 	{ man_in_pre, NULL }, /* in */
111 	{ man_ign_pre, NULL }, /* ft */
112 };
113 
114 /*
115  * Printing leading vertical space before a block.
116  * This is used for the paragraph macros.
117  * The rules are pretty simple, since there's very little nesting going
118  * on here.  Basically, if we're the first within another block (SS/SH),
119  * then don't emit vertical space.  If we are (RS), then do.  If not the
120  * first, print it.
121  */
122 static void
123 print_bvspace(struct html *h, const struct man_node *n)
124 {
125 
126 	if (n->body && n->body->child)
127 		if (MAN_TBL == n->body->child->type)
128 			return;
129 
130 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
131 		if (NULL == n->prev)
132 			return;
133 
134 	print_otag(h, TAG_P, 0, NULL);
135 }
136 
137 void
138 html_man(void *arg, const struct man *m)
139 {
140 	struct html	*h;
141 	struct tag	*t;
142 	struct mhtml	 mh;
143 
144 	h = (struct html *)arg;
145 
146 	print_gen_decls(h);
147 
148 	memset(&mh, 0, sizeof(struct mhtml));
149 
150 	t = print_otag(h, TAG_HTML, 0, NULL);
151 	print_man(man_meta(m), man_node(m), &mh, h);
152 	print_tagq(h, t);
153 
154 	printf("\n");
155 }
156 
157 static void
158 print_man(MAN_ARGS)
159 {
160 	struct tag	*t;
161 
162 	t = print_otag(h, TAG_HEAD, 0, NULL);
163 	print_man_head(m, n, mh, h);
164 	print_tagq(h, t);
165 
166 	t = print_otag(h, TAG_BODY, 0, NULL);
167 	print_man_nodelist(m, n, mh, h);
168 	print_tagq(h, t);
169 }
170 
171 
172 /* ARGSUSED */
173 static void
174 print_man_head(MAN_ARGS)
175 {
176 
177 	print_gen_head(h);
178 	bufcat_fmt(h, "%s(%s)", m->title, m->msec);
179 	print_otag(h, TAG_TITLE, 0, NULL);
180 	print_text(h, h->buf);
181 }
182 
183 
184 static void
185 print_man_nodelist(MAN_ARGS)
186 {
187 
188 	print_man_node(m, n, mh, h);
189 	if (n->next)
190 		print_man_nodelist(m, n->next, mh, h);
191 }
192 
193 
194 static void
195 print_man_node(MAN_ARGS)
196 {
197 	int		 child;
198 	struct tag	*t;
199 
200 	child = 1;
201 	t = h->tags.head;
202 
203 	switch (n->type) {
204 	case (MAN_ROOT):
205 		man_root_pre(m, n, mh, h);
206 		break;
207 	case (MAN_TEXT):
208 		/*
209 		 * If we have a blank line, output a vertical space.
210 		 * If we have a space as the first character, break
211 		 * before printing the line's data.
212 		 */
213 		if ('\0' == *n->string) {
214 			print_otag(h, TAG_P, 0, NULL);
215 			return;
216 		}
217 
218 		if (' ' == *n->string && MAN_LINE & n->flags)
219 			print_otag(h, TAG_BR, 0, NULL);
220 		else if (MANH_LITERAL & mh->fl && n->prev)
221 			print_otag(h, TAG_BR, 0, NULL);
222 
223 		print_text(h, n->string);
224 		return;
225 	case (MAN_EQN):
226 		print_eqn(h, n->eqn);
227 		break;
228 	case (MAN_TBL):
229 		/*
230 		 * This will take care of initialising all of the table
231 		 * state data for the first table, then tearing it down
232 		 * for the last one.
233 		 */
234 		print_tbl(h, n->span);
235 		return;
236 	default:
237 		/*
238 		 * Close out scope of font prior to opening a macro
239 		 * scope.
240 		 */
241 		if (HTMLFONT_NONE != h->metac) {
242 			h->metal = h->metac;
243 			h->metac = HTMLFONT_NONE;
244 		}
245 
246 		/*
247 		 * Close out the current table, if it's open, and unset
248 		 * the "meta" table state.  This will be reopened on the
249 		 * next table element.
250 		 */
251 		if (h->tblt) {
252 			print_tblclose(h);
253 			t = h->tags.head;
254 		}
255 		if (mans[n->tok].pre)
256 			child = (*mans[n->tok].pre)(m, n, mh, h);
257 		break;
258 	}
259 
260 	if (child && n->child)
261 		print_man_nodelist(m, n->child, mh, h);
262 
263 	/* This will automatically close out any font scope. */
264 	print_stagq(h, t);
265 
266 	switch (n->type) {
267 	case (MAN_ROOT):
268 		man_root_post(m, n, mh, h);
269 		break;
270 	case (MAN_EQN):
271 		break;
272 	default:
273 		if (mans[n->tok].post)
274 			(*mans[n->tok].post)(m, n, mh, h);
275 		break;
276 	}
277 }
278 
279 
280 static int
281 a2width(const struct man_node *n, struct roffsu *su)
282 {
283 
284 	if (MAN_TEXT != n->type)
285 		return(0);
286 	if (a2roffsu(n->string, su, SCALE_BU))
287 		return(1);
288 
289 	return(0);
290 }
291 
292 
293 /* ARGSUSED */
294 static void
295 man_root_pre(MAN_ARGS)
296 {
297 	struct htmlpair	 tag[3];
298 	struct tag	*t, *tt;
299 	char		 b[BUFSIZ], title[BUFSIZ];
300 
301 	b[0] = 0;
302 	if (m->vol)
303 		(void)strlcat(b, m->vol, BUFSIZ);
304 
305 	snprintf(title, BUFSIZ - 1, "%s(%s)", m->title, m->msec);
306 
307 	PAIR_SUMMARY_INIT(&tag[0], "Document Header");
308 	PAIR_CLASS_INIT(&tag[1], "head");
309 	if (NULL == h->style) {
310 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
311 		t = print_otag(h, TAG_TABLE, 3, tag);
312 		PAIR_INIT(&tag[0], ATTR_WIDTH, "30%");
313 		print_otag(h, TAG_COL, 1, tag);
314 		print_otag(h, TAG_COL, 1, tag);
315 		print_otag(h, TAG_COL, 1, tag);
316 	} else
317 		t = print_otag(h, TAG_TABLE, 2, tag);
318 
319 	print_otag(h, TAG_TBODY, 0, NULL);
320 
321 	tt = print_otag(h, TAG_TR, 0, NULL);
322 
323 	PAIR_CLASS_INIT(&tag[0], "head-ltitle");
324 	print_otag(h, TAG_TD, 1, tag);
325 
326 	print_text(h, title);
327 	print_stagq(h, tt);
328 
329 	PAIR_CLASS_INIT(&tag[0], "head-vol");
330 	if (NULL == h->style) {
331 		PAIR_INIT(&tag[1], ATTR_ALIGN, "center");
332 		print_otag(h, TAG_TD, 2, tag);
333 	} else
334 		print_otag(h, TAG_TD, 1, tag);
335 
336 	print_text(h, b);
337 	print_stagq(h, tt);
338 
339 	PAIR_CLASS_INIT(&tag[0], "head-rtitle");
340 	if (NULL == h->style) {
341 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
342 		print_otag(h, TAG_TD, 2, tag);
343 	} else
344 		print_otag(h, TAG_TD, 1, tag);
345 
346 	print_text(h, title);
347 	print_tagq(h, t);
348 }
349 
350 
351 /* ARGSUSED */
352 static void
353 man_root_post(MAN_ARGS)
354 {
355 	struct htmlpair	 tag[3];
356 	struct tag	*t, *tt;
357 
358 	PAIR_SUMMARY_INIT(&tag[0], "Document Footer");
359 	PAIR_CLASS_INIT(&tag[1], "foot");
360 	if (NULL == h->style) {
361 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
362 		t = print_otag(h, TAG_TABLE, 3, tag);
363 		PAIR_INIT(&tag[0], ATTR_WIDTH, "50%");
364 		print_otag(h, TAG_COL, 1, tag);
365 		print_otag(h, TAG_COL, 1, tag);
366 	} else
367 		t = print_otag(h, TAG_TABLE, 2, tag);
368 
369 	tt = print_otag(h, TAG_TR, 0, NULL);
370 
371 	PAIR_CLASS_INIT(&tag[0], "foot-date");
372 	print_otag(h, TAG_TD, 1, tag);
373 
374 	print_text(h, m->date);
375 	print_stagq(h, tt);
376 
377 	PAIR_CLASS_INIT(&tag[0], "foot-os");
378 	if (NULL == h->style) {
379 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
380 		print_otag(h, TAG_TD, 2, tag);
381 	} else
382 		print_otag(h, TAG_TD, 1, tag);
383 
384 	if (m->source)
385 		print_text(h, m->source);
386 	print_tagq(h, t);
387 }
388 
389 
390 /* ARGSUSED */
391 static int
392 man_br_pre(MAN_ARGS)
393 {
394 	struct roffsu	 su;
395 	struct htmlpair	 tag;
396 
397 	SCALE_VS_INIT(&su, 1);
398 
399 	if (MAN_sp == n->tok) {
400 		if (NULL != (n = n->child))
401 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
402 				SCALE_VS_INIT(&su, atoi(n->string));
403 	} else
404 		su.scale = 0;
405 
406 	bufinit(h);
407 	bufcat_su(h, "height", &su);
408 	PAIR_STYLE_INIT(&tag, h);
409 	print_otag(h, TAG_DIV, 1, &tag);
410 
411 	/* So the div isn't empty: */
412 	print_text(h, "\\~");
413 
414 	return(0);
415 }
416 
417 /* ARGSUSED */
418 static int
419 man_SH_pre(MAN_ARGS)
420 {
421 	struct htmlpair	 tag;
422 
423 	if (MAN_BLOCK == n->type) {
424 		mh->fl &= ~MANH_LITERAL;
425 		PAIR_CLASS_INIT(&tag, "section");
426 		print_otag(h, TAG_DIV, 1, &tag);
427 		return(1);
428 	} else if (MAN_BODY == n->type)
429 		return(1);
430 
431 	print_otag(h, TAG_H1, 0, NULL);
432 	return(1);
433 }
434 
435 /* ARGSUSED */
436 static int
437 man_alt_pre(MAN_ARGS)
438 {
439 	const struct man_node	*nn;
440 	int		 i, savelit;
441 	enum htmltag	 fp;
442 	struct tag	*t;
443 
444 	if ((savelit = mh->fl & MANH_LITERAL))
445 		print_otag(h, TAG_BR, 0, NULL);
446 
447 	mh->fl &= ~MANH_LITERAL;
448 
449 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
450 		t = NULL;
451 		switch (n->tok) {
452 		case (MAN_BI):
453 			fp = i % 2 ? TAG_I : TAG_B;
454 			break;
455 		case (MAN_IB):
456 			fp = i % 2 ? TAG_B : TAG_I;
457 			break;
458 		case (MAN_RI):
459 			fp = i % 2 ? TAG_I : TAG_MAX;
460 			break;
461 		case (MAN_IR):
462 			fp = i % 2 ? TAG_MAX : TAG_I;
463 			break;
464 		case (MAN_BR):
465 			fp = i % 2 ? TAG_MAX : TAG_B;
466 			break;
467 		case (MAN_RB):
468 			fp = i % 2 ? TAG_B : TAG_MAX;
469 			break;
470 		default:
471 			abort();
472 			/* NOTREACHED */
473 		}
474 
475 		if (i)
476 			h->flags |= HTML_NOSPACE;
477 
478 		if (TAG_MAX != fp)
479 			t = print_otag(h, fp, 0, NULL);
480 
481 		print_man_node(m, nn, mh, h);
482 
483 		if (t)
484 			print_tagq(h, t);
485 	}
486 
487 	if (savelit)
488 		mh->fl |= MANH_LITERAL;
489 
490 	return(0);
491 }
492 
493 /* ARGSUSED */
494 static int
495 man_SM_pre(MAN_ARGS)
496 {
497 
498 	print_otag(h, TAG_SMALL, 0, NULL);
499 	if (MAN_SB == n->tok)
500 		print_otag(h, TAG_B, 0, NULL);
501 	return(1);
502 }
503 
504 /* ARGSUSED */
505 static int
506 man_SS_pre(MAN_ARGS)
507 {
508 	struct htmlpair	 tag;
509 
510 	if (MAN_BLOCK == n->type) {
511 		mh->fl &= ~MANH_LITERAL;
512 		PAIR_CLASS_INIT(&tag, "subsection");
513 		print_otag(h, TAG_DIV, 1, &tag);
514 		return(1);
515 	} else if (MAN_BODY == n->type)
516 		return(1);
517 
518 	print_otag(h, TAG_H2, 0, NULL);
519 	return(1);
520 }
521 
522 /* ARGSUSED */
523 static int
524 man_PP_pre(MAN_ARGS)
525 {
526 
527 	if (MAN_HEAD == n->type)
528 		return(0);
529 	else if (MAN_BLOCK == n->type)
530 		print_bvspace(h, n);
531 
532 	return(1);
533 }
534 
535 /* ARGSUSED */
536 static int
537 man_IP_pre(MAN_ARGS)
538 {
539 	const struct man_node	*nn;
540 
541 	if (MAN_BODY == n->type) {
542 		print_otag(h, TAG_DD, 0, NULL);
543 		return(1);
544 	} else if (MAN_HEAD != n->type) {
545 		print_otag(h, TAG_DL, 0, NULL);
546 		return(1);
547 	}
548 
549 	/* FIXME: width specification. */
550 
551 	print_otag(h, TAG_DT, 0, NULL);
552 
553 	/* For IP, only print the first header element. */
554 
555 	if (MAN_IP == n->tok && n->child)
556 		print_man_node(m, n->child, mh, h);
557 
558 	/* For TP, only print next-line header elements. */
559 
560 	if (MAN_TP == n->tok)
561 		for (nn = n->child; nn; nn = nn->next)
562 			if (nn->line > n->line)
563 				print_man_node(m, nn, mh, h);
564 
565 	return(0);
566 }
567 
568 /* ARGSUSED */
569 static int
570 man_HP_pre(MAN_ARGS)
571 {
572 	struct htmlpair	 tag;
573 	struct roffsu	 su;
574 	const struct man_node *np;
575 
576 	if (MAN_HEAD == n->type)
577 		return(0);
578 	else if (MAN_BLOCK != n->type)
579 		return(1);
580 
581 	np = n->head->child;
582 
583 	if (NULL == np || ! a2width(np, &su))
584 		SCALE_HS_INIT(&su, INDENT);
585 
586 	bufinit(h);
587 
588 	print_bvspace(h, n);
589 	bufcat_su(h, "margin-left", &su);
590 	su.scale = -su.scale;
591 	bufcat_su(h, "text-indent", &su);
592 	PAIR_STYLE_INIT(&tag, h);
593 	print_otag(h, TAG_P, 1, &tag);
594 	return(1);
595 }
596 
597 /* ARGSUSED */
598 static int
599 man_B_pre(MAN_ARGS)
600 {
601 
602 	print_otag(h, TAG_B, 0, NULL);
603 	return(1);
604 }
605 
606 /* ARGSUSED */
607 static int
608 man_I_pre(MAN_ARGS)
609 {
610 
611 	print_otag(h, TAG_I, 0, NULL);
612 	return(1);
613 }
614 
615 /* ARGSUSED */
616 static int
617 man_literal_pre(MAN_ARGS)
618 {
619 
620 	if (MAN_nf != n->tok) {
621 		print_otag(h, TAG_BR, 0, NULL);
622 		mh->fl &= ~MANH_LITERAL;
623 	} else
624 		mh->fl |= MANH_LITERAL;
625 
626 	return(0);
627 }
628 
629 /* ARGSUSED */
630 static int
631 man_in_pre(MAN_ARGS)
632 {
633 
634 	print_otag(h, TAG_BR, 0, NULL);
635 	return(0);
636 }
637 
638 /* ARGSUSED */
639 static int
640 man_ign_pre(MAN_ARGS)
641 {
642 
643 	return(0);
644 }
645 
646 /* ARGSUSED */
647 static int
648 man_RS_pre(MAN_ARGS)
649 {
650 	struct htmlpair	 tag;
651 	struct roffsu	 su;
652 
653 	if (MAN_HEAD == n->type)
654 		return(0);
655 	else if (MAN_BODY == n->type)
656 		return(1);
657 
658 	SCALE_HS_INIT(&su, INDENT);
659 	if (n->head->child)
660 		a2width(n->head->child, &su);
661 
662 	bufinit(h);
663 	bufcat_su(h, "margin-left", &su);
664 	PAIR_STYLE_INIT(&tag, h);
665 	print_otag(h, TAG_DIV, 1, &tag);
666 	return(1);
667 }
668