xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision 83087275e8888ed7109e36d91357557af9d26270)
1 /*	$OpenBSD: man_html.c,v 1.61 2014/12/23 09:31:17 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013, 2014 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 #include <sys/types.h>
19 
20 #include <assert.h>
21 #include <ctype.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 
26 #include "mandoc_aux.h"
27 #include "man.h"
28 #include "out.h"
29 #include "html.h"
30 #include "main.h"
31 
32 /* TODO: preserve ident widths. */
33 /* FIXME: have PD set the default vspace width. */
34 
35 #define	INDENT		  5
36 
37 #define	MAN_ARGS	  const struct man_meta *man, \
38 			  const struct man_node *n, \
39 			  struct mhtml *mh, \
40 			  struct html *h
41 
42 struct	mhtml {
43 	int		  fl;
44 #define	MANH_LITERAL	 (1 << 0) /* literal context */
45 };
46 
47 struct	htmlman {
48 	int		(*pre)(MAN_ARGS);
49 	int		(*post)(MAN_ARGS);
50 };
51 
52 static	void		  print_bvspace(struct html *,
53 				const struct man_node *);
54 static	void		  print_man(MAN_ARGS);
55 static	void		  print_man_head(MAN_ARGS);
56 static	void		  print_man_nodelist(MAN_ARGS);
57 static	void		  print_man_node(MAN_ARGS);
58 static	int		  a2width(const struct man_node *,
59 				struct roffsu *);
60 static	int		  man_B_pre(MAN_ARGS);
61 static	int		  man_HP_pre(MAN_ARGS);
62 static	int		  man_IP_pre(MAN_ARGS);
63 static	int		  man_I_pre(MAN_ARGS);
64 static	int		  man_OP_pre(MAN_ARGS);
65 static	int		  man_PP_pre(MAN_ARGS);
66 static	int		  man_RS_pre(MAN_ARGS);
67 static	int		  man_SH_pre(MAN_ARGS);
68 static	int		  man_SM_pre(MAN_ARGS);
69 static	int		  man_SS_pre(MAN_ARGS);
70 static	int		  man_UR_pre(MAN_ARGS);
71 static	int		  man_alt_pre(MAN_ARGS);
72 static	int		  man_br_pre(MAN_ARGS);
73 static	int		  man_ign_pre(MAN_ARGS);
74 static	int		  man_in_pre(MAN_ARGS);
75 static	int		  man_literal_pre(MAN_ARGS);
76 static	void		  man_root_post(MAN_ARGS);
77 static	void		  man_root_pre(MAN_ARGS);
78 
79 static	const struct htmlman mans[MAN_MAX] = {
80 	{ man_br_pre, NULL }, /* br */
81 	{ NULL, NULL }, /* TH */
82 	{ man_SH_pre, NULL }, /* SH */
83 	{ man_SS_pre, NULL }, /* SS */
84 	{ man_IP_pre, NULL }, /* TP */
85 	{ man_PP_pre, NULL }, /* LP */
86 	{ man_PP_pre, NULL }, /* PP */
87 	{ man_PP_pre, NULL }, /* P */
88 	{ man_IP_pre, NULL }, /* IP */
89 	{ man_HP_pre, NULL }, /* HP */
90 	{ man_SM_pre, NULL }, /* SM */
91 	{ man_SM_pre, NULL }, /* SB */
92 	{ man_alt_pre, NULL }, /* BI */
93 	{ man_alt_pre, NULL }, /* IB */
94 	{ man_alt_pre, NULL }, /* BR */
95 	{ man_alt_pre, NULL }, /* RB */
96 	{ NULL, NULL }, /* R */
97 	{ man_B_pre, NULL }, /* B */
98 	{ man_I_pre, NULL }, /* I */
99 	{ man_alt_pre, NULL }, /* IR */
100 	{ man_alt_pre, NULL }, /* RI */
101 	{ man_ign_pre, NULL }, /* na */
102 	{ man_br_pre, NULL }, /* sp */
103 	{ man_literal_pre, NULL }, /* nf */
104 	{ man_literal_pre, NULL }, /* fi */
105 	{ NULL, NULL }, /* RE */
106 	{ man_RS_pre, NULL }, /* RS */
107 	{ man_ign_pre, NULL }, /* DT */
108 	{ man_ign_pre, NULL }, /* UC */
109 	{ man_ign_pre, NULL }, /* PD */
110 	{ man_ign_pre, NULL }, /* AT */
111 	{ man_in_pre, NULL }, /* in */
112 	{ man_ign_pre, NULL }, /* ft */
113 	{ man_OP_pre, NULL }, /* OP */
114 	{ man_literal_pre, NULL }, /* EX */
115 	{ man_literal_pre, NULL }, /* EE */
116 	{ man_UR_pre, NULL }, /* UR */
117 	{ NULL, NULL }, /* UE */
118 	{ man_ign_pre, NULL }, /* ll */
119 };
120 
121 
122 /*
123  * Printing leading vertical space before a block.
124  * This is used for the paragraph macros.
125  * The rules are pretty simple, since there's very little nesting going
126  * on here.  Basically, if we're the first within another block (SS/SH),
127  * then don't emit vertical space.  If we are (RS), then do.  If not the
128  * first, print it.
129  */
130 static void
131 print_bvspace(struct html *h, const struct man_node *n)
132 {
133 
134 	if (n->body && n->body->child)
135 		if (MAN_TBL == n->body->child->type)
136 			return;
137 
138 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
139 		if (NULL == n->prev)
140 			return;
141 
142 	print_paragraph(h);
143 }
144 
145 void
146 html_man(void *arg, const struct man *man)
147 {
148 	struct mhtml	 mh;
149 
150 	memset(&mh, 0, sizeof(struct mhtml));
151 	print_man(man_meta(man), man_node(man), &mh, (struct html *)arg);
152 	putchar('\n');
153 }
154 
155 static void
156 print_man(MAN_ARGS)
157 {
158 	struct tag	*t, *tt;
159 	struct htmlpair	 tag;
160 
161 	PAIR_CLASS_INIT(&tag, "mandoc");
162 
163 	if ( ! (HTML_FRAGMENT & h->oflags)) {
164 		print_gen_decls(h);
165 		t = print_otag(h, TAG_HTML, 0, NULL);
166 		tt = print_otag(h, TAG_HEAD, 0, NULL);
167 		print_man_head(man, n, mh, h);
168 		print_tagq(h, tt);
169 		print_otag(h, TAG_BODY, 0, NULL);
170 		print_otag(h, TAG_DIV, 1, &tag);
171 	} else
172 		t = print_otag(h, TAG_DIV, 1, &tag);
173 
174 	print_man_nodelist(man, n, mh, h);
175 	print_tagq(h, t);
176 }
177 
178 static void
179 print_man_head(MAN_ARGS)
180 {
181 
182 	print_gen_head(h);
183 	assert(man->title);
184 	assert(man->msec);
185 	bufcat_fmt(h, "%s(%s)", man->title, man->msec);
186 	print_otag(h, TAG_TITLE, 0, NULL);
187 	print_text(h, h->buf);
188 }
189 
190 static void
191 print_man_nodelist(MAN_ARGS)
192 {
193 
194 	print_man_node(man, n, mh, h);
195 	if (n->next)
196 		print_man_nodelist(man, n->next, mh, h);
197 }
198 
199 static void
200 print_man_node(MAN_ARGS)
201 {
202 	int		 child;
203 	struct tag	*t;
204 
205 	child = 1;
206 	t = h->tags.head;
207 
208 	switch (n->type) {
209 	case MAN_ROOT:
210 		man_root_pre(man, n, mh, h);
211 		break;
212 	case MAN_TEXT:
213 		if ('\0' == *n->string) {
214 			print_paragraph(h);
215 			return;
216 		}
217 		if (n->flags & MAN_LINE && (*n->string == ' ' ||
218 		    (n->prev != NULL && mh->fl & MANH_LITERAL &&
219 		     ! (h->flags & HTML_NONEWLINE))))
220 			print_otag(h, TAG_BR, 0, NULL);
221 		print_text(h, n->string);
222 		return;
223 	case MAN_EQN:
224 		print_eqn(h, n->eqn);
225 		break;
226 	case MAN_TBL:
227 		/*
228 		 * This will take care of initialising all of the table
229 		 * state data for the first table, then tearing it down
230 		 * for the last one.
231 		 */
232 		print_tbl(h, n->span);
233 		return;
234 	default:
235 		/*
236 		 * Close out scope of font prior to opening a macro
237 		 * scope.
238 		 */
239 		if (HTMLFONT_NONE != h->metac) {
240 			h->metal = h->metac;
241 			h->metac = HTMLFONT_NONE;
242 		}
243 
244 		/*
245 		 * Close out the current table, if it's open, and unset
246 		 * the "meta" table state.  This will be reopened on the
247 		 * next table element.
248 		 */
249 		if (h->tblt) {
250 			print_tblclose(h);
251 			t = h->tags.head;
252 		}
253 		if (mans[n->tok].pre)
254 			child = (*mans[n->tok].pre)(man, n, mh, h);
255 		break;
256 	}
257 
258 	if (child && n->child)
259 		print_man_nodelist(man, n->child, mh, h);
260 
261 	/* This will automatically close out any font scope. */
262 	print_stagq(h, t);
263 
264 	switch (n->type) {
265 	case MAN_ROOT:
266 		man_root_post(man, n, mh, h);
267 		break;
268 	case MAN_EQN:
269 		break;
270 	default:
271 		if (mans[n->tok].post)
272 			(*mans[n->tok].post)(man, n, mh, h);
273 		break;
274 	}
275 }
276 
277 static int
278 a2width(const struct man_node *n, struct roffsu *su)
279 {
280 
281 	if (MAN_TEXT != n->type)
282 		return(0);
283 	if (a2roffsu(n->string, su, SCALE_EN))
284 		return(1);
285 
286 	return(0);
287 }
288 
289 static void
290 man_root_pre(MAN_ARGS)
291 {
292 	struct htmlpair	 tag;
293 	struct tag	*t, *tt;
294 	char		*title;
295 
296 	assert(man->title);
297 	assert(man->msec);
298 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
299 
300 	PAIR_CLASS_INIT(&tag, "head");
301 	t = print_otag(h, TAG_TABLE, 1, &tag);
302 
303 	print_otag(h, TAG_TBODY, 0, NULL);
304 
305 	tt = print_otag(h, TAG_TR, 0, NULL);
306 
307 	PAIR_CLASS_INIT(&tag, "head-ltitle");
308 	print_otag(h, TAG_TD, 1, &tag);
309 	print_text(h, title);
310 	print_stagq(h, tt);
311 
312 	PAIR_CLASS_INIT(&tag, "head-vol");
313 	print_otag(h, TAG_TD, 1, &tag);
314 	if (NULL != man->vol)
315 		print_text(h, man->vol);
316 	print_stagq(h, tt);
317 
318 	PAIR_CLASS_INIT(&tag, "head-rtitle");
319 	print_otag(h, TAG_TD, 1, &tag);
320 	print_text(h, title);
321 	print_tagq(h, t);
322 	free(title);
323 }
324 
325 static void
326 man_root_post(MAN_ARGS)
327 {
328 	struct htmlpair	 tag;
329 	struct tag	*t, *tt;
330 
331 	PAIR_CLASS_INIT(&tag, "foot");
332 	t = print_otag(h, TAG_TABLE, 1, &tag);
333 
334 	tt = print_otag(h, TAG_TR, 0, NULL);
335 
336 	PAIR_CLASS_INIT(&tag, "foot-date");
337 	print_otag(h, TAG_TD, 1, &tag);
338 
339 	assert(man->date);
340 	print_text(h, man->date);
341 	print_stagq(h, tt);
342 
343 	PAIR_CLASS_INIT(&tag, "foot-os");
344 	print_otag(h, TAG_TD, 1, &tag);
345 
346 	if (man->source)
347 		print_text(h, man->source);
348 	print_tagq(h, t);
349 }
350 
351 
352 static int
353 man_br_pre(MAN_ARGS)
354 {
355 	struct roffsu	 su;
356 	struct htmlpair	 tag;
357 
358 	SCALE_VS_INIT(&su, 1);
359 
360 	if (MAN_sp == n->tok) {
361 		if (NULL != (n = n->child))
362 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
363 				su.scale = 1.0;
364 	} else
365 		su.scale = 0.0;
366 
367 	bufinit(h);
368 	bufcat_su(h, "height", &su);
369 	PAIR_STYLE_INIT(&tag, h);
370 	print_otag(h, TAG_DIV, 1, &tag);
371 
372 	/* So the div isn't empty: */
373 	print_text(h, "\\~");
374 
375 	return(0);
376 }
377 
378 static int
379 man_SH_pre(MAN_ARGS)
380 {
381 	struct htmlpair	 tag;
382 
383 	if (MAN_BLOCK == n->type) {
384 		mh->fl &= ~MANH_LITERAL;
385 		PAIR_CLASS_INIT(&tag, "section");
386 		print_otag(h, TAG_DIV, 1, &tag);
387 		return(1);
388 	} else if (MAN_BODY == n->type)
389 		return(1);
390 
391 	print_otag(h, TAG_H1, 0, NULL);
392 	return(1);
393 }
394 
395 static int
396 man_alt_pre(MAN_ARGS)
397 {
398 	const struct man_node	*nn;
399 	int		 i, savelit;
400 	enum htmltag	 fp;
401 	struct tag	*t;
402 
403 	if ((savelit = mh->fl & MANH_LITERAL))
404 		print_otag(h, TAG_BR, 0, NULL);
405 
406 	mh->fl &= ~MANH_LITERAL;
407 
408 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
409 		t = NULL;
410 		switch (n->tok) {
411 		case MAN_BI:
412 			fp = i % 2 ? TAG_I : TAG_B;
413 			break;
414 		case MAN_IB:
415 			fp = i % 2 ? TAG_B : TAG_I;
416 			break;
417 		case MAN_RI:
418 			fp = i % 2 ? TAG_I : TAG_MAX;
419 			break;
420 		case MAN_IR:
421 			fp = i % 2 ? TAG_MAX : TAG_I;
422 			break;
423 		case MAN_BR:
424 			fp = i % 2 ? TAG_MAX : TAG_B;
425 			break;
426 		case MAN_RB:
427 			fp = i % 2 ? TAG_B : TAG_MAX;
428 			break;
429 		default:
430 			abort();
431 			/* NOTREACHED */
432 		}
433 
434 		if (i)
435 			h->flags |= HTML_NOSPACE;
436 
437 		if (TAG_MAX != fp)
438 			t = print_otag(h, fp, 0, NULL);
439 
440 		print_man_node(man, nn, mh, h);
441 
442 		if (t)
443 			print_tagq(h, t);
444 	}
445 
446 	if (savelit)
447 		mh->fl |= MANH_LITERAL;
448 
449 	return(0);
450 }
451 
452 static int
453 man_SM_pre(MAN_ARGS)
454 {
455 
456 	print_otag(h, TAG_SMALL, 0, NULL);
457 	if (MAN_SB == n->tok)
458 		print_otag(h, TAG_B, 0, NULL);
459 	return(1);
460 }
461 
462 static int
463 man_SS_pre(MAN_ARGS)
464 {
465 	struct htmlpair	 tag;
466 
467 	if (MAN_BLOCK == n->type) {
468 		mh->fl &= ~MANH_LITERAL;
469 		PAIR_CLASS_INIT(&tag, "subsection");
470 		print_otag(h, TAG_DIV, 1, &tag);
471 		return(1);
472 	} else if (MAN_BODY == n->type)
473 		return(1);
474 
475 	print_otag(h, TAG_H2, 0, NULL);
476 	return(1);
477 }
478 
479 static int
480 man_PP_pre(MAN_ARGS)
481 {
482 
483 	if (MAN_HEAD == n->type)
484 		return(0);
485 	else if (MAN_BLOCK == n->type)
486 		print_bvspace(h, n);
487 
488 	return(1);
489 }
490 
491 static int
492 man_IP_pre(MAN_ARGS)
493 {
494 	const struct man_node	*nn;
495 
496 	if (MAN_BODY == n->type) {
497 		print_otag(h, TAG_DD, 0, NULL);
498 		return(1);
499 	} else if (MAN_HEAD != n->type) {
500 		print_otag(h, TAG_DL, 0, NULL);
501 		return(1);
502 	}
503 
504 	/* FIXME: width specification. */
505 
506 	print_otag(h, TAG_DT, 0, NULL);
507 
508 	/* For IP, only print the first header element. */
509 
510 	if (MAN_IP == n->tok && n->child)
511 		print_man_node(man, n->child, mh, h);
512 
513 	/* For TP, only print next-line header elements. */
514 
515 	if (MAN_TP == n->tok) {
516 		nn = n->child;
517 		while (NULL != nn && 0 == (MAN_LINE & nn->flags))
518 			nn = nn->next;
519 		while (NULL != nn) {
520 			print_man_node(man, nn, mh, h);
521 			nn = nn->next;
522 		}
523 	}
524 
525 	return(0);
526 }
527 
528 static int
529 man_HP_pre(MAN_ARGS)
530 {
531 	struct htmlpair	 tag[2];
532 	struct roffsu	 su;
533 	const struct man_node *np;
534 
535 	if (MAN_HEAD == n->type)
536 		return(0);
537 	else if (MAN_BLOCK != n->type)
538 		return(1);
539 
540 	np = n->head->child;
541 
542 	if (NULL == np || ! a2width(np, &su))
543 		SCALE_HS_INIT(&su, INDENT);
544 
545 	bufinit(h);
546 
547 	print_bvspace(h, n);
548 	bufcat_su(h, "margin-left", &su);
549 	su.scale = -su.scale;
550 	bufcat_su(h, "text-indent", &su);
551 	PAIR_STYLE_INIT(&tag[0], h);
552 	PAIR_CLASS_INIT(&tag[1], "spacer");
553 	print_otag(h, TAG_DIV, 2, tag);
554 	return(1);
555 }
556 
557 static int
558 man_OP_pre(MAN_ARGS)
559 {
560 	struct tag	*tt;
561 	struct htmlpair	 tag;
562 
563 	print_text(h, "[");
564 	h->flags |= HTML_NOSPACE;
565 	PAIR_CLASS_INIT(&tag, "opt");
566 	tt = print_otag(h, TAG_SPAN, 1, &tag);
567 
568 	if (NULL != (n = n->child)) {
569 		print_otag(h, TAG_B, 0, NULL);
570 		print_text(h, n->string);
571 	}
572 
573 	print_stagq(h, tt);
574 
575 	if (NULL != n && NULL != n->next) {
576 		print_otag(h, TAG_I, 0, NULL);
577 		print_text(h, n->next->string);
578 	}
579 
580 	print_stagq(h, tt);
581 	h->flags |= HTML_NOSPACE;
582 	print_text(h, "]");
583 	return(0);
584 }
585 
586 static int
587 man_B_pre(MAN_ARGS)
588 {
589 
590 	print_otag(h, TAG_B, 0, NULL);
591 	return(1);
592 }
593 
594 static int
595 man_I_pre(MAN_ARGS)
596 {
597 
598 	print_otag(h, TAG_I, 0, NULL);
599 	return(1);
600 }
601 
602 static int
603 man_literal_pre(MAN_ARGS)
604 {
605 
606 	if (MAN_fi == n->tok || MAN_EE == n->tok) {
607 		print_otag(h, TAG_BR, 0, NULL);
608 		mh->fl &= ~MANH_LITERAL;
609 	} else
610 		mh->fl |= MANH_LITERAL;
611 
612 	return(0);
613 }
614 
615 static int
616 man_in_pre(MAN_ARGS)
617 {
618 
619 	print_otag(h, TAG_BR, 0, NULL);
620 	return(0);
621 }
622 
623 static int
624 man_ign_pre(MAN_ARGS)
625 {
626 
627 	return(0);
628 }
629 
630 static int
631 man_RS_pre(MAN_ARGS)
632 {
633 	struct htmlpair	 tag;
634 	struct roffsu	 su;
635 
636 	if (MAN_HEAD == n->type)
637 		return(0);
638 	else if (MAN_BODY == n->type)
639 		return(1);
640 
641 	SCALE_HS_INIT(&su, INDENT);
642 	if (n->head->child)
643 		a2width(n->head->child, &su);
644 
645 	bufinit(h);
646 	bufcat_su(h, "margin-left", &su);
647 	PAIR_STYLE_INIT(&tag, h);
648 	print_otag(h, TAG_DIV, 1, &tag);
649 	return(1);
650 }
651 
652 static int
653 man_UR_pre(MAN_ARGS)
654 {
655 	struct htmlpair		 tag[2];
656 
657 	n = n->child;
658 	assert(MAN_HEAD == n->type);
659 	if (n->nchild) {
660 		assert(MAN_TEXT == n->child->type);
661 		PAIR_CLASS_INIT(&tag[0], "link-ext");
662 		PAIR_HREF_INIT(&tag[1], n->child->string);
663 		print_otag(h, TAG_A, 2, tag);
664 	}
665 
666 	assert(MAN_BODY == n->next->type);
667 	if (n->next->nchild)
668 		n = n->next;
669 
670 	print_man_nodelist(man, n->child, mh, h);
671 
672 	return(0);
673 }
674