xref: /dragonfly/contrib/mdocml/man_html.c (revision 0db87cb7)
1 /*	$Id: man_html.c,v 1.96 2014/08/01 19:25:52 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012 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 #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 
30 #include "mandoc.h"
31 #include "mandoc_aux.h"
32 #include "out.h"
33 #include "html.h"
34 #include "man.h"
35 #include "main.h"
36 
37 /* TODO: preserve ident widths. */
38 /* FIXME: have PD set the default vspace width. */
39 
40 #define	INDENT		  5
41 
42 #define	MAN_ARGS	  const struct man_meta *man, \
43 			  const struct man_node *n, \
44 			  struct mhtml *mh, \
45 			  struct html *h
46 
47 struct	mhtml {
48 	int		  fl;
49 #define	MANH_LITERAL	 (1 << 0) /* literal context */
50 };
51 
52 struct	htmlman {
53 	int		(*pre)(MAN_ARGS);
54 	int		(*post)(MAN_ARGS);
55 };
56 
57 static	void		  print_bvspace(struct html *,
58 				const struct man_node *);
59 static	void		  print_man(MAN_ARGS);
60 static	void		  print_man_head(MAN_ARGS);
61 static	void		  print_man_nodelist(MAN_ARGS);
62 static	void		  print_man_node(MAN_ARGS);
63 static	int		  a2width(const struct man_node *,
64 				struct roffsu *);
65 static	int		  man_B_pre(MAN_ARGS);
66 static	int		  man_HP_pre(MAN_ARGS);
67 static	int		  man_IP_pre(MAN_ARGS);
68 static	int		  man_I_pre(MAN_ARGS);
69 static	int		  man_OP_pre(MAN_ARGS);
70 static	int		  man_PP_pre(MAN_ARGS);
71 static	int		  man_RS_pre(MAN_ARGS);
72 static	int		  man_SH_pre(MAN_ARGS);
73 static	int		  man_SM_pre(MAN_ARGS);
74 static	int		  man_SS_pre(MAN_ARGS);
75 static	int		  man_UR_pre(MAN_ARGS);
76 static	int		  man_alt_pre(MAN_ARGS);
77 static	int		  man_br_pre(MAN_ARGS);
78 static	int		  man_ign_pre(MAN_ARGS);
79 static	int		  man_in_pre(MAN_ARGS);
80 static	int		  man_literal_pre(MAN_ARGS);
81 static	void		  man_root_post(MAN_ARGS);
82 static	void		  man_root_pre(MAN_ARGS);
83 
84 static	const struct htmlman mans[MAN_MAX] = {
85 	{ man_br_pre, NULL }, /* br */
86 	{ NULL, NULL }, /* TH */
87 	{ man_SH_pre, NULL }, /* SH */
88 	{ man_SS_pre, NULL }, /* SS */
89 	{ man_IP_pre, NULL }, /* TP */
90 	{ man_PP_pre, NULL }, /* LP */
91 	{ man_PP_pre, NULL }, /* PP */
92 	{ man_PP_pre, NULL }, /* P */
93 	{ man_IP_pre, NULL }, /* IP */
94 	{ man_HP_pre, NULL }, /* HP */
95 	{ man_SM_pre, NULL }, /* SM */
96 	{ man_SM_pre, NULL }, /* SB */
97 	{ man_alt_pre, NULL }, /* BI */
98 	{ man_alt_pre, NULL }, /* IB */
99 	{ man_alt_pre, NULL }, /* BR */
100 	{ man_alt_pre, NULL }, /* RB */
101 	{ NULL, NULL }, /* R */
102 	{ man_B_pre, NULL }, /* B */
103 	{ man_I_pre, NULL }, /* I */
104 	{ man_alt_pre, NULL }, /* IR */
105 	{ man_alt_pre, NULL }, /* RI */
106 	{ man_ign_pre, NULL }, /* na */
107 	{ man_br_pre, NULL }, /* sp */
108 	{ man_literal_pre, NULL }, /* nf */
109 	{ man_literal_pre, NULL }, /* fi */
110 	{ NULL, NULL }, /* RE */
111 	{ man_RS_pre, NULL }, /* RS */
112 	{ man_ign_pre, NULL }, /* DT */
113 	{ man_ign_pre, NULL }, /* UC */
114 	{ man_ign_pre, NULL }, /* PD */
115 	{ man_ign_pre, NULL }, /* AT */
116 	{ man_in_pre, NULL }, /* in */
117 	{ man_ign_pre, NULL }, /* ft */
118 	{ man_OP_pre, NULL }, /* OP */
119 	{ man_literal_pre, NULL }, /* EX */
120 	{ man_literal_pre, NULL }, /* EE */
121 	{ man_UR_pre, NULL }, /* UR */
122 	{ NULL, NULL }, /* UE */
123 	{ man_ign_pre, NULL }, /* ll */
124 };
125 
126 
127 /*
128  * Printing leading vertical space before a block.
129  * This is used for the paragraph macros.
130  * The rules are pretty simple, since there's very little nesting going
131  * on here.  Basically, if we're the first within another block (SS/SH),
132  * then don't emit vertical space.  If we are (RS), then do.  If not the
133  * first, print it.
134  */
135 static void
136 print_bvspace(struct html *h, const struct man_node *n)
137 {
138 
139 	if (n->body && n->body->child)
140 		if (MAN_TBL == n->body->child->type)
141 			return;
142 
143 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
144 		if (NULL == n->prev)
145 			return;
146 
147 	print_otag(h, TAG_P, 0, NULL);
148 }
149 
150 void
151 html_man(void *arg, const struct man *man)
152 {
153 	struct mhtml	 mh;
154 
155 	memset(&mh, 0, sizeof(struct mhtml));
156 	print_man(man_meta(man), man_node(man), &mh, (struct html *)arg);
157 	putchar('\n');
158 }
159 
160 static void
161 print_man(MAN_ARGS)
162 {
163 	struct tag	*t, *tt;
164 	struct htmlpair	 tag;
165 
166 	PAIR_CLASS_INIT(&tag, "mandoc");
167 
168 	if ( ! (HTML_FRAGMENT & h->oflags)) {
169 		print_gen_decls(h);
170 		t = print_otag(h, TAG_HTML, 0, NULL);
171 		tt = print_otag(h, TAG_HEAD, 0, NULL);
172 		print_man_head(man, n, mh, h);
173 		print_tagq(h, tt);
174 		print_otag(h, TAG_BODY, 0, NULL);
175 		print_otag(h, TAG_DIV, 1, &tag);
176 	} else
177 		t = print_otag(h, TAG_DIV, 1, &tag);
178 
179 	print_man_nodelist(man, n, mh, h);
180 	print_tagq(h, t);
181 }
182 
183 static void
184 print_man_head(MAN_ARGS)
185 {
186 
187 	print_gen_head(h);
188 	assert(man->title);
189 	assert(man->msec);
190 	bufcat_fmt(h, "%s(%s)", man->title, man->msec);
191 	print_otag(h, TAG_TITLE, 0, NULL);
192 	print_text(h, h->buf);
193 }
194 
195 static void
196 print_man_nodelist(MAN_ARGS)
197 {
198 
199 	print_man_node(man, n, mh, h);
200 	if (n->next)
201 		print_man_nodelist(man, n->next, mh, h);
202 }
203 
204 static void
205 print_man_node(MAN_ARGS)
206 {
207 	int		 child;
208 	struct tag	*t;
209 
210 	child = 1;
211 	t = h->tags.head;
212 
213 	switch (n->type) {
214 	case MAN_ROOT:
215 		man_root_pre(man, n, mh, h);
216 		break;
217 	case MAN_TEXT:
218 		/*
219 		 * If we have a blank line, output a vertical space.
220 		 * If we have a space as the first character, break
221 		 * before printing the line's data.
222 		 */
223 		if ('\0' == *n->string) {
224 			print_otag(h, TAG_P, 0, NULL);
225 			return;
226 		}
227 
228 		if (' ' == *n->string && MAN_LINE & n->flags)
229 			print_otag(h, TAG_BR, 0, NULL);
230 		else if (MANH_LITERAL & mh->fl && n->prev)
231 			print_otag(h, TAG_BR, 0, NULL);
232 
233 		print_text(h, n->string);
234 		return;
235 	case MAN_EQN:
236 		print_eqn(h, n->eqn);
237 		break;
238 	case MAN_TBL:
239 		/*
240 		 * This will take care of initialising all of the table
241 		 * state data for the first table, then tearing it down
242 		 * for the last one.
243 		 */
244 		print_tbl(h, n->span);
245 		return;
246 	default:
247 		/*
248 		 * Close out scope of font prior to opening a macro
249 		 * scope.
250 		 */
251 		if (HTMLFONT_NONE != h->metac) {
252 			h->metal = h->metac;
253 			h->metac = HTMLFONT_NONE;
254 		}
255 
256 		/*
257 		 * Close out the current table, if it's open, and unset
258 		 * the "meta" table state.  This will be reopened on the
259 		 * next table element.
260 		 */
261 		if (h->tblt) {
262 			print_tblclose(h);
263 			t = h->tags.head;
264 		}
265 		if (mans[n->tok].pre)
266 			child = (*mans[n->tok].pre)(man, n, mh, h);
267 		break;
268 	}
269 
270 	if (child && n->child)
271 		print_man_nodelist(man, n->child, mh, h);
272 
273 	/* This will automatically close out any font scope. */
274 	print_stagq(h, t);
275 
276 	switch (n->type) {
277 	case MAN_ROOT:
278 		man_root_post(man, n, mh, h);
279 		break;
280 	case MAN_EQN:
281 		break;
282 	default:
283 		if (mans[n->tok].post)
284 			(*mans[n->tok].post)(man, n, mh, h);
285 		break;
286 	}
287 }
288 
289 static int
290 a2width(const struct man_node *n, struct roffsu *su)
291 {
292 
293 	if (MAN_TEXT != n->type)
294 		return(0);
295 	if (a2roffsu(n->string, su, SCALE_BU))
296 		return(1);
297 
298 	return(0);
299 }
300 
301 static void
302 man_root_pre(MAN_ARGS)
303 {
304 	struct htmlpair	 tag[3];
305 	struct tag	*t, *tt;
306 	char		*title;
307 
308 	assert(man->title);
309 	assert(man->msec);
310 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
311 
312 	PAIR_SUMMARY_INIT(&tag[0], "Document Header");
313 	PAIR_CLASS_INIT(&tag[1], "head");
314 	PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
315 	t = print_otag(h, TAG_TABLE, 3, tag);
316 	PAIR_INIT(&tag[0], ATTR_WIDTH, "30%");
317 	print_otag(h, TAG_COL, 1, tag);
318 	print_otag(h, TAG_COL, 1, tag);
319 	print_otag(h, TAG_COL, 1, tag);
320 
321 	print_otag(h, TAG_TBODY, 0, NULL);
322 
323 	tt = print_otag(h, TAG_TR, 0, NULL);
324 
325 	PAIR_CLASS_INIT(&tag[0], "head-ltitle");
326 	print_otag(h, TAG_TD, 1, tag);
327 	print_text(h, title);
328 	print_stagq(h, tt);
329 
330 	PAIR_CLASS_INIT(&tag[0], "head-vol");
331 	PAIR_INIT(&tag[1], ATTR_ALIGN, "center");
332 	print_otag(h, TAG_TD, 2, tag);
333 	if (NULL != man->vol)
334 		print_text(h, man->vol);
335 	print_stagq(h, tt);
336 
337 	PAIR_CLASS_INIT(&tag[0], "head-rtitle");
338 	PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
339 	print_otag(h, TAG_TD, 2, tag);
340 	print_text(h, title);
341 	print_tagq(h, t);
342 	free(title);
343 }
344 
345 static void
346 man_root_post(MAN_ARGS)
347 {
348 	struct htmlpair	 tag[3];
349 	struct tag	*t, *tt;
350 
351 	PAIR_SUMMARY_INIT(&tag[0], "Document Footer");
352 	PAIR_CLASS_INIT(&tag[1], "foot");
353 	PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
354 	t = print_otag(h, TAG_TABLE, 3, tag);
355 	PAIR_INIT(&tag[0], ATTR_WIDTH, "50%");
356 	print_otag(h, TAG_COL, 1, tag);
357 	print_otag(h, TAG_COL, 1, tag);
358 
359 	tt = print_otag(h, TAG_TR, 0, NULL);
360 
361 	PAIR_CLASS_INIT(&tag[0], "foot-date");
362 	print_otag(h, TAG_TD, 1, tag);
363 
364 	assert(man->date);
365 	print_text(h, man->date);
366 	print_stagq(h, tt);
367 
368 	PAIR_CLASS_INIT(&tag[0], "foot-os");
369 	PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
370 	print_otag(h, TAG_TD, 2, tag);
371 
372 	if (man->source)
373 		print_text(h, man->source);
374 	print_tagq(h, t);
375 }
376 
377 
378 static int
379 man_br_pre(MAN_ARGS)
380 {
381 	struct roffsu	 su;
382 	struct htmlpair	 tag;
383 
384 	SCALE_VS_INIT(&su, 1);
385 
386 	if (MAN_sp == n->tok) {
387 		if (NULL != (n = n->child))
388 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
389 				SCALE_VS_INIT(&su, atoi(n->string));
390 	} else
391 		su.scale = 0.0;
392 
393 	bufinit(h);
394 	bufcat_su(h, "height", &su);
395 	PAIR_STYLE_INIT(&tag, h);
396 	print_otag(h, TAG_DIV, 1, &tag);
397 
398 	/* So the div isn't empty: */
399 	print_text(h, "\\~");
400 
401 	return(0);
402 }
403 
404 static int
405 man_SH_pre(MAN_ARGS)
406 {
407 	struct htmlpair	 tag;
408 
409 	if (MAN_BLOCK == n->type) {
410 		mh->fl &= ~MANH_LITERAL;
411 		PAIR_CLASS_INIT(&tag, "section");
412 		print_otag(h, TAG_DIV, 1, &tag);
413 		return(1);
414 	} else if (MAN_BODY == n->type)
415 		return(1);
416 
417 	print_otag(h, TAG_H1, 0, NULL);
418 	return(1);
419 }
420 
421 static int
422 man_alt_pre(MAN_ARGS)
423 {
424 	const struct man_node	*nn;
425 	int		 i, savelit;
426 	enum htmltag	 fp;
427 	struct tag	*t;
428 
429 	if ((savelit = mh->fl & MANH_LITERAL))
430 		print_otag(h, TAG_BR, 0, NULL);
431 
432 	mh->fl &= ~MANH_LITERAL;
433 
434 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
435 		t = NULL;
436 		switch (n->tok) {
437 		case MAN_BI:
438 			fp = i % 2 ? TAG_I : TAG_B;
439 			break;
440 		case MAN_IB:
441 			fp = i % 2 ? TAG_B : TAG_I;
442 			break;
443 		case MAN_RI:
444 			fp = i % 2 ? TAG_I : TAG_MAX;
445 			break;
446 		case MAN_IR:
447 			fp = i % 2 ? TAG_MAX : TAG_I;
448 			break;
449 		case MAN_BR:
450 			fp = i % 2 ? TAG_MAX : TAG_B;
451 			break;
452 		case MAN_RB:
453 			fp = i % 2 ? TAG_B : TAG_MAX;
454 			break;
455 		default:
456 			abort();
457 			/* NOTREACHED */
458 		}
459 
460 		if (i)
461 			h->flags |= HTML_NOSPACE;
462 
463 		if (TAG_MAX != fp)
464 			t = print_otag(h, fp, 0, NULL);
465 
466 		print_man_node(man, nn, mh, h);
467 
468 		if (t)
469 			print_tagq(h, t);
470 	}
471 
472 	if (savelit)
473 		mh->fl |= MANH_LITERAL;
474 
475 	return(0);
476 }
477 
478 static int
479 man_SM_pre(MAN_ARGS)
480 {
481 
482 	print_otag(h, TAG_SMALL, 0, NULL);
483 	if (MAN_SB == n->tok)
484 		print_otag(h, TAG_B, 0, NULL);
485 	return(1);
486 }
487 
488 static int
489 man_SS_pre(MAN_ARGS)
490 {
491 	struct htmlpair	 tag;
492 
493 	if (MAN_BLOCK == n->type) {
494 		mh->fl &= ~MANH_LITERAL;
495 		PAIR_CLASS_INIT(&tag, "subsection");
496 		print_otag(h, TAG_DIV, 1, &tag);
497 		return(1);
498 	} else if (MAN_BODY == n->type)
499 		return(1);
500 
501 	print_otag(h, TAG_H2, 0, NULL);
502 	return(1);
503 }
504 
505 static int
506 man_PP_pre(MAN_ARGS)
507 {
508 
509 	if (MAN_HEAD == n->type)
510 		return(0);
511 	else if (MAN_BLOCK == n->type)
512 		print_bvspace(h, n);
513 
514 	return(1);
515 }
516 
517 static int
518 man_IP_pre(MAN_ARGS)
519 {
520 	const struct man_node	*nn;
521 
522 	if (MAN_BODY == n->type) {
523 		print_otag(h, TAG_DD, 0, NULL);
524 		return(1);
525 	} else if (MAN_HEAD != n->type) {
526 		print_otag(h, TAG_DL, 0, NULL);
527 		return(1);
528 	}
529 
530 	/* FIXME: width specification. */
531 
532 	print_otag(h, TAG_DT, 0, NULL);
533 
534 	/* For IP, only print the first header element. */
535 
536 	if (MAN_IP == n->tok && n->child)
537 		print_man_node(man, n->child, mh, h);
538 
539 	/* For TP, only print next-line header elements. */
540 
541 	if (MAN_TP == n->tok) {
542 		nn = n->child;
543 		while (NULL != nn && 0 == (MAN_LINE & nn->flags))
544 			nn = nn->next;
545 		while (NULL != nn) {
546 			print_man_node(man, nn, mh, h);
547 			nn = nn->next;
548 		}
549 	}
550 
551 	return(0);
552 }
553 
554 static int
555 man_HP_pre(MAN_ARGS)
556 {
557 	struct htmlpair	 tag;
558 	struct roffsu	 su;
559 	const struct man_node *np;
560 
561 	if (MAN_HEAD == n->type)
562 		return(0);
563 	else if (MAN_BLOCK != n->type)
564 		return(1);
565 
566 	np = n->head->child;
567 
568 	if (NULL == np || ! a2width(np, &su))
569 		SCALE_HS_INIT(&su, INDENT);
570 
571 	bufinit(h);
572 
573 	print_bvspace(h, n);
574 	bufcat_su(h, "margin-left", &su);
575 	su.scale = -su.scale;
576 	bufcat_su(h, "text-indent", &su);
577 	PAIR_STYLE_INIT(&tag, h);
578 	print_otag(h, TAG_P, 1, &tag);
579 	return(1);
580 }
581 
582 static int
583 man_OP_pre(MAN_ARGS)
584 {
585 	struct tag	*tt;
586 	struct htmlpair	 tag;
587 
588 	print_text(h, "[");
589 	h->flags |= HTML_NOSPACE;
590 	PAIR_CLASS_INIT(&tag, "opt");
591 	tt = print_otag(h, TAG_SPAN, 1, &tag);
592 
593 	if (NULL != (n = n->child)) {
594 		print_otag(h, TAG_B, 0, NULL);
595 		print_text(h, n->string);
596 	}
597 
598 	print_stagq(h, tt);
599 
600 	if (NULL != n && NULL != n->next) {
601 		print_otag(h, TAG_I, 0, NULL);
602 		print_text(h, n->next->string);
603 	}
604 
605 	print_stagq(h, tt);
606 	h->flags |= HTML_NOSPACE;
607 	print_text(h, "]");
608 	return(0);
609 }
610 
611 static int
612 man_B_pre(MAN_ARGS)
613 {
614 
615 	print_otag(h, TAG_B, 0, NULL);
616 	return(1);
617 }
618 
619 static int
620 man_I_pre(MAN_ARGS)
621 {
622 
623 	print_otag(h, TAG_I, 0, NULL);
624 	return(1);
625 }
626 
627 static int
628 man_literal_pre(MAN_ARGS)
629 {
630 
631 	if (MAN_fi == n->tok || MAN_EE == n->tok) {
632 		print_otag(h, TAG_BR, 0, NULL);
633 		mh->fl &= ~MANH_LITERAL;
634 	} else
635 		mh->fl |= MANH_LITERAL;
636 
637 	return(0);
638 }
639 
640 static int
641 man_in_pre(MAN_ARGS)
642 {
643 
644 	print_otag(h, TAG_BR, 0, NULL);
645 	return(0);
646 }
647 
648 static int
649 man_ign_pre(MAN_ARGS)
650 {
651 
652 	return(0);
653 }
654 
655 static int
656 man_RS_pre(MAN_ARGS)
657 {
658 	struct htmlpair	 tag;
659 	struct roffsu	 su;
660 
661 	if (MAN_HEAD == n->type)
662 		return(0);
663 	else if (MAN_BODY == n->type)
664 		return(1);
665 
666 	SCALE_HS_INIT(&su, INDENT);
667 	if (n->head->child)
668 		a2width(n->head->child, &su);
669 
670 	bufinit(h);
671 	bufcat_su(h, "margin-left", &su);
672 	PAIR_STYLE_INIT(&tag, h);
673 	print_otag(h, TAG_DIV, 1, &tag);
674 	return(1);
675 }
676 
677 static int
678 man_UR_pre(MAN_ARGS)
679 {
680 	struct htmlpair		 tag[2];
681 
682 	n = n->child;
683 	assert(MAN_HEAD == n->type);
684 	if (n->nchild) {
685 		assert(MAN_TEXT == n->child->type);
686 		PAIR_CLASS_INIT(&tag[0], "link-ext");
687 		PAIR_HREF_INIT(&tag[1], n->child->string);
688 		print_otag(h, TAG_A, 2, tag);
689 	}
690 
691 	assert(MAN_BODY == n->next->type);
692 	if (n->next->nchild)
693 		n = n->next;
694 
695 	print_man_nodelist(man, n->child, mh, h);
696 
697 	return(0);
698 }
699