xref: /netbsd-src/external/bsd/mdocml/dist/man_html.c (revision e39ef1d61eee3ccba837ee281f1e098c864487aa)
1 /*	$Vendor-Id: man_html.c,v 1.82 2011/10/05 21:35:17 kristaps 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 #ifdef HAVE_CONFIG_H
18 #include "config.h"
19 #endif
20 
21 #include <sys/types.h>
22 
23 #include <assert.h>
24 #include <ctype.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 
29 #include "mandoc.h"
30 #include "out.h"
31 #include "html.h"
32 #include "man.h"
33 #include "main.h"
34 
35 /* TODO: preserve ident widths. */
36 /* FIXME: have PD set the default vspace width. */
37 
38 #define	INDENT		  5
39 
40 #define	MAN_ARGS	  const struct man_meta *m, \
41 			  const struct man_node *n, \
42 			  struct mhtml *mh, \
43 			  struct html *h
44 
45 struct	mhtml {
46 	int		  fl;
47 #define	MANH_LITERAL	 (1 << 0) /* literal context */
48 };
49 
50 struct	htmlman {
51 	int		(*pre)(MAN_ARGS);
52 	int		(*post)(MAN_ARGS);
53 };
54 
55 static	void		  print_bvspace(struct html *,
56 				const struct man_node *);
57 static	void		  print_man(MAN_ARGS);
58 static	void		  print_man_head(MAN_ARGS);
59 static	void		  print_man_nodelist(MAN_ARGS);
60 static	void		  print_man_node(MAN_ARGS);
61 
62 static	int		  a2width(const struct man_node *,
63 				struct roffsu *);
64 
65 static	int		  man_alt_pre(MAN_ARGS);
66 static	int		  man_br_pre(MAN_ARGS);
67 static	int		  man_ign_pre(MAN_ARGS);
68 static	int		  man_in_pre(MAN_ARGS);
69 static	int		  man_literal_pre(MAN_ARGS);
70 static	void		  man_root_post(MAN_ARGS);
71 static	void		  man_root_pre(MAN_ARGS);
72 static	int		  man_B_pre(MAN_ARGS);
73 static	int		  man_HP_pre(MAN_ARGS);
74 static	int		  man_I_pre(MAN_ARGS);
75 static	int		  man_IP_pre(MAN_ARGS);
76 static	int		  man_PP_pre(MAN_ARGS);
77 static	int		  man_RS_pre(MAN_ARGS);
78 static	int		  man_SH_pre(MAN_ARGS);
79 static	int		  man_SM_pre(MAN_ARGS);
80 static	int		  man_SS_pre(MAN_ARGS);
81 
82 static	const struct htmlman mans[MAN_MAX] = {
83 	{ man_br_pre, NULL }, /* br */
84 	{ NULL, NULL }, /* TH */
85 	{ man_SH_pre, NULL }, /* SH */
86 	{ man_SS_pre, NULL }, /* SS */
87 	{ man_IP_pre, NULL }, /* TP */
88 	{ man_PP_pre, NULL }, /* LP */
89 	{ man_PP_pre, NULL }, /* PP */
90 	{ man_PP_pre, NULL }, /* P */
91 	{ man_IP_pre, NULL }, /* IP */
92 	{ man_HP_pre, NULL }, /* HP */
93 	{ man_SM_pre, NULL }, /* SM */
94 	{ man_SM_pre, NULL }, /* SB */
95 	{ man_alt_pre, NULL }, /* BI */
96 	{ man_alt_pre, NULL }, /* IB */
97 	{ man_alt_pre, NULL }, /* BR */
98 	{ man_alt_pre, NULL }, /* RB */
99 	{ NULL, NULL }, /* R */
100 	{ man_B_pre, NULL }, /* B */
101 	{ man_I_pre, NULL }, /* I */
102 	{ man_alt_pre, NULL }, /* IR */
103 	{ man_alt_pre, NULL }, /* RI */
104 	{ man_ign_pre, NULL }, /* na */
105 	{ man_br_pre, NULL }, /* sp */
106 	{ man_literal_pre, NULL }, /* nf */
107 	{ man_literal_pre, NULL }, /* fi */
108 	{ NULL, NULL }, /* RE */
109 	{ man_RS_pre, NULL }, /* RS */
110 	{ man_ign_pre, NULL }, /* DT */
111 	{ man_ign_pre, NULL }, /* UC */
112 	{ man_ign_pre, NULL }, /* PD */
113 	{ man_ign_pre, NULL }, /* AT */
114 	{ man_in_pre, NULL }, /* in */
115 	{ man_ign_pre, NULL }, /* ft */
116 };
117 
118 /*
119  * Printing leading vertical space before a block.
120  * This is used for the paragraph macros.
121  * The rules are pretty simple, since there's very little nesting going
122  * on here.  Basically, if we're the first within another block (SS/SH),
123  * then don't emit vertical space.  If we are (RS), then do.  If not the
124  * first, print it.
125  */
126 static void
127 print_bvspace(struct html *h, const struct man_node *n)
128 {
129 
130 	if (n->body && n->body->child)
131 		if (MAN_TBL == n->body->child->type)
132 			return;
133 
134 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
135 		if (NULL == n->prev)
136 			return;
137 
138 	print_otag(h, TAG_P, 0, NULL);
139 }
140 
141 void
142 html_man(void *arg, const struct man *m)
143 {
144 	struct mhtml	 mh;
145 
146 	memset(&mh, 0, sizeof(struct mhtml));
147 	print_man(man_meta(m), man_node(m), &mh, (struct html *)arg);
148 	putchar('\n');
149 }
150 
151 static void
152 print_man(MAN_ARGS)
153 {
154 	struct tag	*t, *tt;
155 	struct htmlpair	 tag;
156 
157 	PAIR_CLASS_INIT(&tag, "mandoc");
158 
159 	if ( ! (HTML_FRAGMENT & h->oflags)) {
160 		print_gen_decls(h);
161 		t = print_otag(h, TAG_HTML, 0, NULL);
162 		tt = print_otag(h, TAG_HEAD, 0, NULL);
163 		print_man_head(m, n, mh, h);
164 		print_tagq(h, tt);
165 		print_otag(h, TAG_BODY, 0, NULL);
166 		print_otag(h, TAG_DIV, 1, &tag);
167 	} else
168 		t = print_otag(h, TAG_DIV, 1, &tag);
169 
170 	print_man_nodelist(m, n, mh, h);
171 	print_tagq(h, t);
172 }
173 
174 
175 /* ARGSUSED */
176 static void
177 print_man_head(MAN_ARGS)
178 {
179 
180 	print_gen_head(h);
181 	bufcat_fmt(h, "%s(%s)", m->title, m->msec);
182 	print_otag(h, TAG_TITLE, 0, NULL);
183 	print_text(h, h->buf);
184 }
185 
186 
187 static void
188 print_man_nodelist(MAN_ARGS)
189 {
190 
191 	print_man_node(m, n, mh, h);
192 	if (n->next)
193 		print_man_nodelist(m, n->next, mh, h);
194 }
195 
196 
197 static void
198 print_man_node(MAN_ARGS)
199 {
200 	int		 child;
201 	struct tag	*t;
202 
203 	child = 1;
204 	t = h->tags.head;
205 
206 	switch (n->type) {
207 	case (MAN_ROOT):
208 		man_root_pre(m, n, mh, h);
209 		break;
210 	case (MAN_TEXT):
211 		/*
212 		 * If we have a blank line, output a vertical space.
213 		 * If we have a space as the first character, break
214 		 * before printing the line's data.
215 		 */
216 		if ('\0' == *n->string) {
217 			print_otag(h, TAG_P, 0, NULL);
218 			return;
219 		}
220 
221 		if (' ' == *n->string && MAN_LINE & n->flags)
222 			print_otag(h, TAG_BR, 0, NULL);
223 		else if (MANH_LITERAL & mh->fl && n->prev)
224 			print_otag(h, TAG_BR, 0, NULL);
225 
226 		print_text(h, n->string);
227 		return;
228 	case (MAN_EQN):
229 		print_eqn(h, n->eqn);
230 		break;
231 	case (MAN_TBL):
232 		/*
233 		 * This will take care of initialising all of the table
234 		 * state data for the first table, then tearing it down
235 		 * for the last one.
236 		 */
237 		print_tbl(h, n->span);
238 		return;
239 	default:
240 		/*
241 		 * Close out scope of font prior to opening a macro
242 		 * scope.
243 		 */
244 		if (HTMLFONT_NONE != h->metac) {
245 			h->metal = h->metac;
246 			h->metac = HTMLFONT_NONE;
247 		}
248 
249 		/*
250 		 * Close out the current table, if it's open, and unset
251 		 * the "meta" table state.  This will be reopened on the
252 		 * next table element.
253 		 */
254 		if (h->tblt) {
255 			print_tblclose(h);
256 			t = h->tags.head;
257 		}
258 		if (mans[n->tok].pre)
259 			child = (*mans[n->tok].pre)(m, n, mh, h);
260 		break;
261 	}
262 
263 	if (child && n->child)
264 		print_man_nodelist(m, n->child, mh, h);
265 
266 	/* This will automatically close out any font scope. */
267 	print_stagq(h, t);
268 
269 	switch (n->type) {
270 	case (MAN_ROOT):
271 		man_root_post(m, n, mh, h);
272 		break;
273 	case (MAN_EQN):
274 		break;
275 	default:
276 		if (mans[n->tok].post)
277 			(*mans[n->tok].post)(m, n, mh, h);
278 		break;
279 	}
280 }
281 
282 
283 static int
284 a2width(const struct man_node *n, struct roffsu *su)
285 {
286 
287 	if (MAN_TEXT != n->type)
288 		return(0);
289 	if (a2roffsu(n->string, su, SCALE_BU))
290 		return(1);
291 
292 	return(0);
293 }
294 
295 
296 /* ARGSUSED */
297 static void
298 man_root_pre(MAN_ARGS)
299 {
300 	struct htmlpair	 tag[3];
301 	struct tag	*t, *tt;
302 	char		 b[BUFSIZ], title[BUFSIZ];
303 
304 	b[0] = 0;
305 	if (m->vol)
306 		(void)strlcat(b, m->vol, BUFSIZ);
307 
308 	snprintf(title, BUFSIZ - 1, "%s(%s)", m->title, m->msec);
309 
310 	PAIR_SUMMARY_INIT(&tag[0], "Document Header");
311 	PAIR_CLASS_INIT(&tag[1], "head");
312 	if (NULL == h->style) {
313 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
314 		t = print_otag(h, TAG_TABLE, 3, tag);
315 		PAIR_INIT(&tag[0], ATTR_WIDTH, "30%");
316 		print_otag(h, TAG_COL, 1, tag);
317 		print_otag(h, TAG_COL, 1, tag);
318 		print_otag(h, TAG_COL, 1, tag);
319 	} else
320 		t = print_otag(h, TAG_TABLE, 2, tag);
321 
322 	print_otag(h, TAG_TBODY, 0, NULL);
323 
324 	tt = print_otag(h, TAG_TR, 0, NULL);
325 
326 	PAIR_CLASS_INIT(&tag[0], "head-ltitle");
327 	print_otag(h, TAG_TD, 1, tag);
328 
329 	print_text(h, title);
330 	print_stagq(h, tt);
331 
332 	PAIR_CLASS_INIT(&tag[0], "head-vol");
333 	if (NULL == h->style) {
334 		PAIR_INIT(&tag[1], ATTR_ALIGN, "center");
335 		print_otag(h, TAG_TD, 2, tag);
336 	} else
337 		print_otag(h, TAG_TD, 1, tag);
338 
339 	print_text(h, b);
340 	print_stagq(h, tt);
341 
342 	PAIR_CLASS_INIT(&tag[0], "head-rtitle");
343 	if (NULL == h->style) {
344 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
345 		print_otag(h, TAG_TD, 2, tag);
346 	} else
347 		print_otag(h, TAG_TD, 1, tag);
348 
349 	print_text(h, title);
350 	print_tagq(h, t);
351 }
352 
353 
354 /* ARGSUSED */
355 static void
356 man_root_post(MAN_ARGS)
357 {
358 	struct htmlpair	 tag[3];
359 	struct tag	*t, *tt;
360 
361 	PAIR_SUMMARY_INIT(&tag[0], "Document Footer");
362 	PAIR_CLASS_INIT(&tag[1], "foot");
363 	if (NULL == h->style) {
364 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
365 		t = print_otag(h, TAG_TABLE, 3, tag);
366 		PAIR_INIT(&tag[0], ATTR_WIDTH, "50%");
367 		print_otag(h, TAG_COL, 1, tag);
368 		print_otag(h, TAG_COL, 1, tag);
369 	} else
370 		t = print_otag(h, TAG_TABLE, 2, tag);
371 
372 	tt = print_otag(h, TAG_TR, 0, NULL);
373 
374 	PAIR_CLASS_INIT(&tag[0], "foot-date");
375 	print_otag(h, TAG_TD, 1, tag);
376 
377 	print_text(h, m->date);
378 	print_stagq(h, tt);
379 
380 	PAIR_CLASS_INIT(&tag[0], "foot-os");
381 	if (NULL == h->style) {
382 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
383 		print_otag(h, TAG_TD, 2, tag);
384 	} else
385 		print_otag(h, TAG_TD, 1, tag);
386 
387 	if (m->source)
388 		print_text(h, m->source);
389 	print_tagq(h, t);
390 }
391 
392 
393 /* ARGSUSED */
394 static int
395 man_br_pre(MAN_ARGS)
396 {
397 	struct roffsu	 su;
398 	struct htmlpair	 tag;
399 
400 	SCALE_VS_INIT(&su, 1);
401 
402 	if (MAN_sp == n->tok) {
403 		if (NULL != (n = n->child))
404 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
405 				SCALE_VS_INIT(&su, atoi(n->string));
406 	} else
407 		su.scale = 0;
408 
409 	bufinit(h);
410 	bufcat_su(h, "height", &su);
411 	PAIR_STYLE_INIT(&tag, h);
412 	print_otag(h, TAG_DIV, 1, &tag);
413 
414 	/* So the div isn't empty: */
415 	print_text(h, "\\~");
416 
417 	return(0);
418 }
419 
420 /* ARGSUSED */
421 static int
422 man_SH_pre(MAN_ARGS)
423 {
424 	struct htmlpair	 tag;
425 
426 	if (MAN_BLOCK == n->type) {
427 		mh->fl &= ~MANH_LITERAL;
428 		PAIR_CLASS_INIT(&tag, "section");
429 		print_otag(h, TAG_DIV, 1, &tag);
430 		return(1);
431 	} else if (MAN_BODY == n->type)
432 		return(1);
433 
434 	print_otag(h, TAG_H1, 0, NULL);
435 	return(1);
436 }
437 
438 /* ARGSUSED */
439 static int
440 man_alt_pre(MAN_ARGS)
441 {
442 	const struct man_node	*nn;
443 	int		 i, savelit;
444 	enum htmltag	 fp;
445 	struct tag	*t;
446 
447 	if ((savelit = mh->fl & MANH_LITERAL))
448 		print_otag(h, TAG_BR, 0, NULL);
449 
450 	mh->fl &= ~MANH_LITERAL;
451 
452 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
453 		t = NULL;
454 		switch (n->tok) {
455 		case (MAN_BI):
456 			fp = i % 2 ? TAG_I : TAG_B;
457 			break;
458 		case (MAN_IB):
459 			fp = i % 2 ? TAG_B : TAG_I;
460 			break;
461 		case (MAN_RI):
462 			fp = i % 2 ? TAG_I : TAG_MAX;
463 			break;
464 		case (MAN_IR):
465 			fp = i % 2 ? TAG_MAX : TAG_I;
466 			break;
467 		case (MAN_BR):
468 			fp = i % 2 ? TAG_MAX : TAG_B;
469 			break;
470 		case (MAN_RB):
471 			fp = i % 2 ? TAG_B : TAG_MAX;
472 			break;
473 		default:
474 			abort();
475 			/* NOTREACHED */
476 		}
477 
478 		if (i)
479 			h->flags |= HTML_NOSPACE;
480 
481 		if (TAG_MAX != fp)
482 			t = print_otag(h, fp, 0, NULL);
483 
484 		print_man_node(m, nn, mh, h);
485 
486 		if (t)
487 			print_tagq(h, t);
488 	}
489 
490 	if (savelit)
491 		mh->fl |= MANH_LITERAL;
492 
493 	return(0);
494 }
495 
496 /* ARGSUSED */
497 static int
498 man_SM_pre(MAN_ARGS)
499 {
500 
501 	print_otag(h, TAG_SMALL, 0, NULL);
502 	if (MAN_SB == n->tok)
503 		print_otag(h, TAG_B, 0, NULL);
504 	return(1);
505 }
506 
507 /* ARGSUSED */
508 static int
509 man_SS_pre(MAN_ARGS)
510 {
511 	struct htmlpair	 tag;
512 
513 	if (MAN_BLOCK == n->type) {
514 		mh->fl &= ~MANH_LITERAL;
515 		PAIR_CLASS_INIT(&tag, "subsection");
516 		print_otag(h, TAG_DIV, 1, &tag);
517 		return(1);
518 	} else if (MAN_BODY == n->type)
519 		return(1);
520 
521 	print_otag(h, TAG_H2, 0, NULL);
522 	return(1);
523 }
524 
525 /* ARGSUSED */
526 static int
527 man_PP_pre(MAN_ARGS)
528 {
529 
530 	if (MAN_HEAD == n->type)
531 		return(0);
532 	else if (MAN_BLOCK == n->type)
533 		print_bvspace(h, n);
534 
535 	return(1);
536 }
537 
538 /* ARGSUSED */
539 static int
540 man_IP_pre(MAN_ARGS)
541 {
542 	const struct man_node	*nn;
543 
544 	if (MAN_BODY == n->type) {
545 		print_otag(h, TAG_DD, 0, NULL);
546 		return(1);
547 	} else if (MAN_HEAD != n->type) {
548 		print_otag(h, TAG_DL, 0, NULL);
549 		return(1);
550 	}
551 
552 	/* FIXME: width specification. */
553 
554 	print_otag(h, TAG_DT, 0, NULL);
555 
556 	/* For IP, only print the first header element. */
557 
558 	if (MAN_IP == n->tok && n->child)
559 		print_man_node(m, n->child, mh, h);
560 
561 	/* For TP, only print next-line header elements. */
562 
563 	if (MAN_TP == n->tok)
564 		for (nn = n->child; nn; nn = nn->next)
565 			if (nn->line > n->line)
566 				print_man_node(m, nn, mh, h);
567 
568 	return(0);
569 }
570 
571 /* ARGSUSED */
572 static int
573 man_HP_pre(MAN_ARGS)
574 {
575 	struct htmlpair	 tag;
576 	struct roffsu	 su;
577 	const struct man_node *np;
578 
579 	if (MAN_HEAD == n->type)
580 		return(0);
581 	else if (MAN_BLOCK != n->type)
582 		return(1);
583 
584 	np = n->head->child;
585 
586 	if (NULL == np || ! a2width(np, &su))
587 		SCALE_HS_INIT(&su, INDENT);
588 
589 	bufinit(h);
590 
591 	print_bvspace(h, n);
592 	bufcat_su(h, "margin-left", &su);
593 	su.scale = -su.scale;
594 	bufcat_su(h, "text-indent", &su);
595 	PAIR_STYLE_INIT(&tag, h);
596 	print_otag(h, TAG_P, 1, &tag);
597 	return(1);
598 }
599 
600 /* ARGSUSED */
601 static int
602 man_B_pre(MAN_ARGS)
603 {
604 
605 	print_otag(h, TAG_B, 0, NULL);
606 	return(1);
607 }
608 
609 /* ARGSUSED */
610 static int
611 man_I_pre(MAN_ARGS)
612 {
613 
614 	print_otag(h, TAG_I, 0, NULL);
615 	return(1);
616 }
617 
618 /* ARGSUSED */
619 static int
620 man_literal_pre(MAN_ARGS)
621 {
622 
623 	if (MAN_nf != n->tok) {
624 		print_otag(h, TAG_BR, 0, NULL);
625 		mh->fl &= ~MANH_LITERAL;
626 	} else
627 		mh->fl |= MANH_LITERAL;
628 
629 	return(0);
630 }
631 
632 /* ARGSUSED */
633 static int
634 man_in_pre(MAN_ARGS)
635 {
636 
637 	print_otag(h, TAG_BR, 0, NULL);
638 	return(0);
639 }
640 
641 /* ARGSUSED */
642 static int
643 man_ign_pre(MAN_ARGS)
644 {
645 
646 	return(0);
647 }
648 
649 /* ARGSUSED */
650 static int
651 man_RS_pre(MAN_ARGS)
652 {
653 	struct htmlpair	 tag;
654 	struct roffsu	 su;
655 
656 	if (MAN_HEAD == n->type)
657 		return(0);
658 	else if (MAN_BODY == n->type)
659 		return(1);
660 
661 	SCALE_HS_INIT(&su, INDENT);
662 	if (n->head->child)
663 		a2width(n->head->child, &su);
664 
665 	bufinit(h);
666 	bufcat_su(h, "margin-left", &su);
667 	PAIR_STYLE_INIT(&tag, h);
668 	print_otag(h, TAG_DIV, 1, &tag);
669 	return(1);
670 }
671