xref: /illumos-gate/usr/src/cmd/mandoc/man_html.c (revision 2bbdd445a21f9d61f4a0ca0faf05d5ceb2bd91f3)
1 /*	$Id: man_html.c,v 1.90 2013/10/17 20:54:58 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013 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 "out.h"
32 #include "html.h"
33 #include "man.h"
34 #include "main.h"
35 
36 /* TODO: preserve ident widths. */
37 /* FIXME: have PD set the default vspace width. */
38 
39 #define	INDENT		  5
40 
41 #define	MAN_ARGS	  const struct man_meta *man, \
42 			  const struct man_node *n, \
43 			  struct mhtml *mh, \
44 			  struct html *h
45 
46 struct	mhtml {
47 	int		  fl;
48 #define	MANH_LITERAL	 (1 << 0) /* literal context */
49 };
50 
51 struct	htmlman {
52 	int		(*pre)(MAN_ARGS);
53 	int		(*post)(MAN_ARGS);
54 };
55 
56 static	void		  print_bvspace(struct html *,
57 				const struct man_node *);
58 static	void		  print_man(MAN_ARGS);
59 static	void		  print_man_head(MAN_ARGS);
60 static	void		  print_man_nodelist(MAN_ARGS);
61 static	void		  print_man_node(MAN_ARGS);
62 static	int		  a2width(const struct man_node *,
63 				struct roffsu *);
64 static	int		  man_B_pre(MAN_ARGS);
65 static	int		  man_HP_pre(MAN_ARGS);
66 static	int		  man_IP_pre(MAN_ARGS);
67 static	int		  man_I_pre(MAN_ARGS);
68 static	int		  man_OP_pre(MAN_ARGS);
69 static	int		  man_PP_pre(MAN_ARGS);
70 static	int		  man_RS_pre(MAN_ARGS);
71 static	int		  man_SH_pre(MAN_ARGS);
72 static	int		  man_SM_pre(MAN_ARGS);
73 static	int		  man_SS_pre(MAN_ARGS);
74 static	int		  man_UR_pre(MAN_ARGS);
75 static	int		  man_alt_pre(MAN_ARGS);
76 static	int		  man_br_pre(MAN_ARGS);
77 static	int		  man_ign_pre(MAN_ARGS);
78 static	int		  man_in_pre(MAN_ARGS);
79 static	int		  man_literal_pre(MAN_ARGS);
80 static	void		  man_root_post(MAN_ARGS);
81 static	void		  man_root_pre(MAN_ARGS);
82 
83 static	const struct htmlman mans[MAN_MAX] = {
84 	{ man_br_pre, NULL }, /* br */
85 	{ NULL, NULL }, /* TH */
86 	{ man_SH_pre, NULL }, /* SH */
87 	{ man_SS_pre, NULL }, /* SS */
88 	{ man_IP_pre, NULL }, /* TP */
89 	{ man_PP_pre, NULL }, /* LP */
90 	{ man_PP_pre, NULL }, /* PP */
91 	{ man_PP_pre, NULL }, /* P */
92 	{ man_IP_pre, NULL }, /* IP */
93 	{ man_HP_pre, NULL }, /* HP */
94 	{ man_SM_pre, NULL }, /* SM */
95 	{ man_SM_pre, NULL }, /* SB */
96 	{ man_alt_pre, NULL }, /* BI */
97 	{ man_alt_pre, NULL }, /* IB */
98 	{ man_alt_pre, NULL }, /* BR */
99 	{ man_alt_pre, NULL }, /* RB */
100 	{ NULL, NULL }, /* R */
101 	{ man_B_pre, NULL }, /* B */
102 	{ man_I_pre, NULL }, /* I */
103 	{ man_alt_pre, NULL }, /* IR */
104 	{ man_alt_pre, NULL }, /* RI */
105 	{ man_ign_pre, NULL }, /* na */
106 	{ man_br_pre, NULL }, /* sp */
107 	{ man_literal_pre, NULL }, /* nf */
108 	{ man_literal_pre, NULL }, /* fi */
109 	{ NULL, NULL }, /* RE */
110 	{ man_RS_pre, NULL }, /* RS */
111 	{ man_ign_pre, NULL }, /* DT */
112 	{ man_ign_pre, NULL }, /* UC */
113 	{ man_ign_pre, NULL }, /* PD */
114 	{ man_ign_pre, NULL }, /* AT */
115 	{ man_in_pre, NULL }, /* in */
116 	{ man_ign_pre, NULL }, /* ft */
117 	{ man_OP_pre, NULL }, /* OP */
118 	{ man_literal_pre, NULL }, /* EX */
119 	{ man_literal_pre, NULL }, /* EE */
120 	{ man_UR_pre, NULL }, /* UR */
121 	{ NULL, NULL }, /* UE */
122 };
123 
124 /*
125  * Printing leading vertical space before a block.
126  * This is used for the paragraph macros.
127  * The rules are pretty simple, since there's very little nesting going
128  * on here.  Basically, if we're the first within another block (SS/SH),
129  * then don't emit vertical space.  If we are (RS), then do.  If not the
130  * first, print it.
131  */
132 static void
133 print_bvspace(struct html *h, const struct man_node *n)
134 {
135 
136 	if (n->body && n->body->child)
137 		if (MAN_TBL == n->body->child->type)
138 			return;
139 
140 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
141 		if (NULL == n->prev)
142 			return;
143 
144 	print_otag(h, TAG_P, 0, NULL);
145 }
146 
147 void
148 html_man(void *arg, const struct man *man)
149 {
150 	struct mhtml	 mh;
151 
152 	memset(&mh, 0, sizeof(struct mhtml));
153 	print_man(man_meta(man), man_node(man), &mh, (struct html *)arg);
154 	putchar('\n');
155 }
156 
157 static void
158 print_man(MAN_ARGS)
159 {
160 	struct tag	*t, *tt;
161 	struct htmlpair	 tag;
162 
163 	PAIR_CLASS_INIT(&tag, "mandoc");
164 
165 	if ( ! (HTML_FRAGMENT & h->oflags)) {
166 		print_gen_decls(h);
167 		t = print_otag(h, TAG_HTML, 0, NULL);
168 		tt = print_otag(h, TAG_HEAD, 0, NULL);
169 		print_man_head(man, n, mh, h);
170 		print_tagq(h, tt);
171 		print_otag(h, TAG_BODY, 0, NULL);
172 		print_otag(h, TAG_DIV, 1, &tag);
173 	} else
174 		t = print_otag(h, TAG_DIV, 1, &tag);
175 
176 	print_man_nodelist(man, n, mh, h);
177 	print_tagq(h, t);
178 }
179 
180 
181 /* ARGSUSED */
182 static void
183 print_man_head(MAN_ARGS)
184 {
185 
186 	print_gen_head(h);
187 	assert(man->title);
188 	assert(man->msec);
189 	bufcat_fmt(h, "%s(%s)", man->title, man->msec);
190 	print_otag(h, TAG_TITLE, 0, NULL);
191 	print_text(h, h->buf);
192 }
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 
205 static void
206 print_man_node(MAN_ARGS)
207 {
208 	int		 child;
209 	struct tag	*t;
210 
211 	child = 1;
212 	t = h->tags.head;
213 
214 	switch (n->type) {
215 	case (MAN_ROOT):
216 		man_root_pre(man, n, mh, h);
217 		break;
218 	case (MAN_TEXT):
219 		/*
220 		 * If we have a blank line, output a vertical space.
221 		 * If we have a space as the first character, break
222 		 * before printing the line's data.
223 		 */
224 		if ('\0' == *n->string) {
225 			print_otag(h, TAG_P, 0, NULL);
226 			return;
227 		}
228 
229 		if (' ' == *n->string && MAN_LINE & n->flags)
230 			print_otag(h, TAG_BR, 0, NULL);
231 		else if (MANH_LITERAL & mh->fl && n->prev)
232 			print_otag(h, TAG_BR, 0, NULL);
233 
234 		print_text(h, n->string);
235 		return;
236 	case (MAN_EQN):
237 		print_eqn(h, n->eqn);
238 		break;
239 	case (MAN_TBL):
240 		/*
241 		 * This will take care of initialising all of the table
242 		 * state data for the first table, then tearing it down
243 		 * for the last one.
244 		 */
245 		print_tbl(h, n->span);
246 		return;
247 	default:
248 		/*
249 		 * Close out scope of font prior to opening a macro
250 		 * scope.
251 		 */
252 		if (HTMLFONT_NONE != h->metac) {
253 			h->metal = h->metac;
254 			h->metac = HTMLFONT_NONE;
255 		}
256 
257 		/*
258 		 * Close out the current table, if it's open, and unset
259 		 * the "meta" table state.  This will be reopened on the
260 		 * next table element.
261 		 */
262 		if (h->tblt) {
263 			print_tblclose(h);
264 			t = h->tags.head;
265 		}
266 		if (mans[n->tok].pre)
267 			child = (*mans[n->tok].pre)(man, n, mh, h);
268 		break;
269 	}
270 
271 	if (child && n->child)
272 		print_man_nodelist(man, n->child, mh, h);
273 
274 	/* This will automatically close out any font scope. */
275 	print_stagq(h, t);
276 
277 	switch (n->type) {
278 	case (MAN_ROOT):
279 		man_root_post(man, n, mh, h);
280 		break;
281 	case (MAN_EQN):
282 		break;
283 	default:
284 		if (mans[n->tok].post)
285 			(*mans[n->tok].post)(man, n, mh, h);
286 		break;
287 	}
288 }
289 
290 
291 static int
292 a2width(const struct man_node *n, struct roffsu *su)
293 {
294 
295 	if (MAN_TEXT != n->type)
296 		return(0);
297 	if (a2roffsu(n->string, su, SCALE_BU))
298 		return(1);
299 
300 	return(0);
301 }
302 
303 
304 /* ARGSUSED */
305 static void
306 man_root_pre(MAN_ARGS)
307 {
308 	struct htmlpair	 tag[3];
309 	struct tag	*t, *tt;
310 	char		 b[BUFSIZ], title[BUFSIZ];
311 
312 	b[0] = 0;
313 	if (man->vol)
314 		(void)strlcat(b, man->vol, BUFSIZ);
315 
316 	assert(man->title);
317 	assert(man->msec);
318 	snprintf(title, BUFSIZ - 1, "%s(%s)", man->title, man->msec);
319 
320 	PAIR_SUMMARY_INIT(&tag[0], "Document Header");
321 	PAIR_CLASS_INIT(&tag[1], "head");
322 	PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
323 	t = print_otag(h, TAG_TABLE, 3, tag);
324 	PAIR_INIT(&tag[0], ATTR_WIDTH, "30%");
325 	print_otag(h, TAG_COL, 1, tag);
326 	print_otag(h, TAG_COL, 1, tag);
327 	print_otag(h, TAG_COL, 1, tag);
328 
329 	print_otag(h, TAG_TBODY, 0, NULL);
330 
331 	tt = print_otag(h, TAG_TR, 0, NULL);
332 
333 	PAIR_CLASS_INIT(&tag[0], "head-ltitle");
334 	print_otag(h, TAG_TD, 1, tag);
335 	print_text(h, title);
336 	print_stagq(h, tt);
337 
338 	PAIR_CLASS_INIT(&tag[0], "head-vol");
339 	PAIR_INIT(&tag[1], ATTR_ALIGN, "center");
340 	print_otag(h, TAG_TD, 2, tag);
341 	print_text(h, b);
342 	print_stagq(h, tt);
343 
344 	PAIR_CLASS_INIT(&tag[0], "head-rtitle");
345 	PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
346 	print_otag(h, TAG_TD, 2, tag);
347 	print_text(h, title);
348 	print_tagq(h, t);
349 }
350 
351 
352 /* ARGSUSED */
353 static void
354 man_root_post(MAN_ARGS)
355 {
356 	struct htmlpair	 tag[3];
357 	struct tag	*t, *tt;
358 
359 	PAIR_SUMMARY_INIT(&tag[0], "Document Footer");
360 	PAIR_CLASS_INIT(&tag[1], "foot");
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 
367 	tt = print_otag(h, TAG_TR, 0, NULL);
368 
369 	PAIR_CLASS_INIT(&tag[0], "foot-date");
370 	print_otag(h, TAG_TD, 1, tag);
371 
372 	assert(man->date);
373 	print_text(h, man->date);
374 	print_stagq(h, tt);
375 
376 	PAIR_CLASS_INIT(&tag[0], "foot-os");
377 	PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
378 	print_otag(h, TAG_TD, 2, tag);
379 
380 	if (man->source)
381 		print_text(h, man->source);
382 	print_tagq(h, t);
383 }
384 
385 
386 /* ARGSUSED */
387 static int
388 man_br_pre(MAN_ARGS)
389 {
390 	struct roffsu	 su;
391 	struct htmlpair	 tag;
392 
393 	SCALE_VS_INIT(&su, 1);
394 
395 	if (MAN_sp == n->tok) {
396 		if (NULL != (n = n->child))
397 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
398 				SCALE_VS_INIT(&su, atoi(n->string));
399 	} else
400 		su.scale = 0;
401 
402 	bufinit(h);
403 	bufcat_su(h, "height", &su);
404 	PAIR_STYLE_INIT(&tag, h);
405 	print_otag(h, TAG_DIV, 1, &tag);
406 
407 	/* So the div isn't empty: */
408 	print_text(h, "\\~");
409 
410 	return(0);
411 }
412 
413 /* ARGSUSED */
414 static int
415 man_SH_pre(MAN_ARGS)
416 {
417 	struct htmlpair	 tag;
418 
419 	if (MAN_BLOCK == n->type) {
420 		mh->fl &= ~MANH_LITERAL;
421 		PAIR_CLASS_INIT(&tag, "section");
422 		print_otag(h, TAG_DIV, 1, &tag);
423 		return(1);
424 	} else if (MAN_BODY == n->type)
425 		return(1);
426 
427 	print_otag(h, TAG_H1, 0, NULL);
428 	return(1);
429 }
430 
431 /* ARGSUSED */
432 static int
433 man_alt_pre(MAN_ARGS)
434 {
435 	const struct man_node	*nn;
436 	int		 i, savelit;
437 	enum htmltag	 fp;
438 	struct tag	*t;
439 
440 	if ((savelit = mh->fl & MANH_LITERAL))
441 		print_otag(h, TAG_BR, 0, NULL);
442 
443 	mh->fl &= ~MANH_LITERAL;
444 
445 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
446 		t = NULL;
447 		switch (n->tok) {
448 		case (MAN_BI):
449 			fp = i % 2 ? TAG_I : TAG_B;
450 			break;
451 		case (MAN_IB):
452 			fp = i % 2 ? TAG_B : TAG_I;
453 			break;
454 		case (MAN_RI):
455 			fp = i % 2 ? TAG_I : TAG_MAX;
456 			break;
457 		case (MAN_IR):
458 			fp = i % 2 ? TAG_MAX : TAG_I;
459 			break;
460 		case (MAN_BR):
461 			fp = i % 2 ? TAG_MAX : TAG_B;
462 			break;
463 		case (MAN_RB):
464 			fp = i % 2 ? TAG_B : TAG_MAX;
465 			break;
466 		default:
467 			abort();
468 			/* NOTREACHED */
469 		}
470 
471 		if (i)
472 			h->flags |= HTML_NOSPACE;
473 
474 		if (TAG_MAX != fp)
475 			t = print_otag(h, fp, 0, NULL);
476 
477 		print_man_node(man, nn, mh, h);
478 
479 		if (t)
480 			print_tagq(h, t);
481 	}
482 
483 	if (savelit)
484 		mh->fl |= MANH_LITERAL;
485 
486 	return(0);
487 }
488 
489 /* ARGSUSED */
490 static int
491 man_SM_pre(MAN_ARGS)
492 {
493 
494 	print_otag(h, TAG_SMALL, 0, NULL);
495 	if (MAN_SB == n->tok)
496 		print_otag(h, TAG_B, 0, NULL);
497 	return(1);
498 }
499 
500 /* ARGSUSED */
501 static int
502 man_SS_pre(MAN_ARGS)
503 {
504 	struct htmlpair	 tag;
505 
506 	if (MAN_BLOCK == n->type) {
507 		mh->fl &= ~MANH_LITERAL;
508 		PAIR_CLASS_INIT(&tag, "subsection");
509 		print_otag(h, TAG_DIV, 1, &tag);
510 		return(1);
511 	} else if (MAN_BODY == n->type)
512 		return(1);
513 
514 	print_otag(h, TAG_H2, 0, NULL);
515 	return(1);
516 }
517 
518 /* ARGSUSED */
519 static int
520 man_PP_pre(MAN_ARGS)
521 {
522 
523 	if (MAN_HEAD == n->type)
524 		return(0);
525 	else if (MAN_BLOCK == n->type)
526 		print_bvspace(h, n);
527 
528 	return(1);
529 }
530 
531 /* ARGSUSED */
532 static int
533 man_IP_pre(MAN_ARGS)
534 {
535 	const struct man_node	*nn;
536 
537 	if (MAN_BODY == n->type) {
538 		print_otag(h, TAG_DD, 0, NULL);
539 		return(1);
540 	} else if (MAN_HEAD != n->type) {
541 		print_otag(h, TAG_DL, 0, NULL);
542 		return(1);
543 	}
544 
545 	/* FIXME: width specification. */
546 
547 	print_otag(h, TAG_DT, 0, NULL);
548 
549 	/* For IP, only print the first header element. */
550 
551 	if (MAN_IP == n->tok && n->child)
552 		print_man_node(man, n->child, mh, h);
553 
554 	/* For TP, only print next-line header elements. */
555 
556 	if (MAN_TP == n->tok)
557 		for (nn = n->child; nn; nn = nn->next)
558 			if (nn->line > n->line)
559 				print_man_node(man, nn, mh, h);
560 
561 	return(0);
562 }
563 
564 /* ARGSUSED */
565 static int
566 man_HP_pre(MAN_ARGS)
567 {
568 	struct htmlpair	 tag;
569 	struct roffsu	 su;
570 	const struct man_node *np;
571 
572 	if (MAN_HEAD == n->type)
573 		return(0);
574 	else if (MAN_BLOCK != n->type)
575 		return(1);
576 
577 	np = n->head->child;
578 
579 	if (NULL == np || ! a2width(np, &su))
580 		SCALE_HS_INIT(&su, INDENT);
581 
582 	bufinit(h);
583 
584 	print_bvspace(h, n);
585 	bufcat_su(h, "margin-left", &su);
586 	su.scale = -su.scale;
587 	bufcat_su(h, "text-indent", &su);
588 	PAIR_STYLE_INIT(&tag, h);
589 	print_otag(h, TAG_P, 1, &tag);
590 	return(1);
591 }
592 
593 /* ARGSUSED */
594 static int
595 man_OP_pre(MAN_ARGS)
596 {
597 	struct tag	*tt;
598 	struct htmlpair	 tag;
599 
600 	print_text(h, "[");
601 	h->flags |= HTML_NOSPACE;
602 	PAIR_CLASS_INIT(&tag, "opt");
603 	tt = print_otag(h, TAG_SPAN, 1, &tag);
604 
605 	if (NULL != (n = n->child)) {
606 		print_otag(h, TAG_B, 0, NULL);
607 		print_text(h, n->string);
608 	}
609 
610 	print_stagq(h, tt);
611 
612 	if (NULL != n && NULL != n->next) {
613 		print_otag(h, TAG_I, 0, NULL);
614 		print_text(h, n->next->string);
615 	}
616 
617 	print_stagq(h, tt);
618 	h->flags |= HTML_NOSPACE;
619 	print_text(h, "]");
620 	return(0);
621 }
622 
623 
624 /* ARGSUSED */
625 static int
626 man_B_pre(MAN_ARGS)
627 {
628 
629 	print_otag(h, TAG_B, 0, NULL);
630 	return(1);
631 }
632 
633 /* ARGSUSED */
634 static int
635 man_I_pre(MAN_ARGS)
636 {
637 
638 	print_otag(h, TAG_I, 0, NULL);
639 	return(1);
640 }
641 
642 /* ARGSUSED */
643 static int
644 man_literal_pre(MAN_ARGS)
645 {
646 
647 	if (MAN_fi == n->tok || MAN_EE == n->tok) {
648 		print_otag(h, TAG_BR, 0, NULL);
649 		mh->fl &= ~MANH_LITERAL;
650 	} else
651 		mh->fl |= MANH_LITERAL;
652 
653 	return(0);
654 }
655 
656 /* ARGSUSED */
657 static int
658 man_in_pre(MAN_ARGS)
659 {
660 
661 	print_otag(h, TAG_BR, 0, NULL);
662 	return(0);
663 }
664 
665 /* ARGSUSED */
666 static int
667 man_ign_pre(MAN_ARGS)
668 {
669 
670 	return(0);
671 }
672 
673 /* ARGSUSED */
674 static int
675 man_RS_pre(MAN_ARGS)
676 {
677 	struct htmlpair	 tag;
678 	struct roffsu	 su;
679 
680 	if (MAN_HEAD == n->type)
681 		return(0);
682 	else if (MAN_BODY == n->type)
683 		return(1);
684 
685 	SCALE_HS_INIT(&su, INDENT);
686 	if (n->head->child)
687 		a2width(n->head->child, &su);
688 
689 	bufinit(h);
690 	bufcat_su(h, "margin-left", &su);
691 	PAIR_STYLE_INIT(&tag, h);
692 	print_otag(h, TAG_DIV, 1, &tag);
693 	return(1);
694 }
695 
696 /* ARGSUSED */
697 static int
698 man_UR_pre(MAN_ARGS)
699 {
700 	struct htmlpair		 tag[2];
701 
702 	n = n->child;
703 	assert(MAN_HEAD == n->type);
704 	if (n->nchild) {
705 		assert(MAN_TEXT == n->child->type);
706 		PAIR_CLASS_INIT(&tag[0], "link-ext");
707 		PAIR_HREF_INIT(&tag[1], n->child->string);
708 		print_otag(h, TAG_A, 2, tag);
709 	}
710 
711 	assert(MAN_BODY == n->next->type);
712 	if (n->next->nchild)
713 		n = n->next;
714 
715 	print_man_nodelist(man, n->child, mh, h);
716 
717 	return(0);
718 }
719