xref: /minix3/external/bsd/mdocml/dist/man_html.c (revision f14fb602092e015ff630df58e17c2a9cd57d29b3)
1 /*	$Vendor-Id: man_html.c,v 1.62 2011/01/07 13:20:58 kristaps Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010 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 #define	HALFINDENT	  3
40 
41 #define	MAN_ARGS	  const struct man_meta *m, \
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_man(MAN_ARGS);
57 static	void		  print_man_head(MAN_ARGS);
58 static	void		  print_man_nodelist(MAN_ARGS);
59 static	void		  print_man_node(MAN_ARGS);
60 
61 static	int		  a2width(const struct man_node *,
62 				struct roffsu *);
63 
64 static	int		  man_alt_pre(MAN_ARGS);
65 static	int		  man_br_pre(MAN_ARGS);
66 static	int		  man_ign_pre(MAN_ARGS);
67 static	int		  man_in_pre(MAN_ARGS);
68 static	int		  man_literal_pre(MAN_ARGS);
69 static	void		  man_root_post(MAN_ARGS);
70 static	int		  man_root_pre(MAN_ARGS);
71 static	int		  man_B_pre(MAN_ARGS);
72 static	int		  man_HP_pre(MAN_ARGS);
73 static	int		  man_I_pre(MAN_ARGS);
74 static	int		  man_IP_pre(MAN_ARGS);
75 static	int		  man_PP_pre(MAN_ARGS);
76 static	int		  man_RS_pre(MAN_ARGS);
77 static	int		  man_SH_pre(MAN_ARGS);
78 static	int		  man_SM_pre(MAN_ARGS);
79 static	int		  man_SS_pre(MAN_ARGS);
80 
81 static	const struct htmlman mans[MAN_MAX] = {
82 	{ man_br_pre, NULL }, /* br */
83 	{ NULL, NULL }, /* TH */
84 	{ man_SH_pre, NULL }, /* SH */
85 	{ man_SS_pre, NULL }, /* SS */
86 	{ man_IP_pre, NULL }, /* TP */
87 	{ man_PP_pre, NULL }, /* LP */
88 	{ man_PP_pre, NULL }, /* PP */
89 	{ man_PP_pre, NULL }, /* P */
90 	{ man_IP_pre, NULL }, /* IP */
91 	{ man_HP_pre, NULL }, /* HP */
92 	{ man_SM_pre, NULL }, /* SM */
93 	{ man_SM_pre, NULL }, /* SB */
94 	{ man_alt_pre, NULL }, /* BI */
95 	{ man_alt_pre, NULL }, /* IB */
96 	{ man_alt_pre, NULL }, /* BR */
97 	{ man_alt_pre, NULL }, /* RB */
98 	{ NULL, NULL }, /* R */
99 	{ man_B_pre, NULL }, /* B */
100 	{ man_I_pre, NULL }, /* I */
101 	{ man_alt_pre, NULL }, /* IR */
102 	{ man_alt_pre, NULL }, /* RI */
103 	{ NULL, NULL }, /* na */
104 	{ man_br_pre, NULL }, /* sp */
105 	{ man_literal_pre, NULL }, /* nf */
106 	{ man_literal_pre, NULL }, /* fi */
107 	{ NULL, NULL }, /* RE */
108 	{ man_RS_pre, NULL }, /* RS */
109 	{ man_ign_pre, NULL }, /* DT */
110 	{ man_ign_pre, NULL }, /* UC */
111 	{ man_ign_pre, NULL }, /* PD */
112 	{ man_ign_pre, NULL }, /* AT */
113 	{ man_in_pre, NULL }, /* in */
114 	{ man_ign_pre, NULL }, /* ft */
115 };
116 
117 
118 void
119 html_man(void *arg, const struct man *m)
120 {
121 	struct html	*h;
122 	struct tag	*t;
123 	struct mhtml	 mh;
124 
125 	h = (struct html *)arg;
126 
127 	print_gen_decls(h);
128 
129 	memset(&mh, 0, sizeof(struct mhtml));
130 
131 	t = print_otag(h, TAG_HTML, 0, NULL);
132 	print_man(man_meta(m), man_node(m), &mh, h);
133 	print_tagq(h, t);
134 
135 	printf("\n");
136 }
137 
138 
139 static void
140 print_man(MAN_ARGS)
141 {
142 	struct tag	*t;
143 
144 	t = print_otag(h, TAG_HEAD, 0, NULL);
145 	print_man_head(m, n, mh, h);
146 	print_tagq(h, t);
147 
148 	t = print_otag(h, TAG_BODY, 0, NULL);
149 	print_man_nodelist(m, n, mh, h);
150 	print_tagq(h, t);
151 }
152 
153 
154 /* ARGSUSED */
155 static void
156 print_man_head(MAN_ARGS)
157 {
158 
159 	print_gen_head(h);
160 	bufinit(h);
161 	buffmt(h, "%s(%s)", m->title, m->msec);
162 
163 	print_otag(h, TAG_TITLE, 0, NULL);
164 	print_text(h, h->buf);
165 }
166 
167 
168 static void
169 print_man_nodelist(MAN_ARGS)
170 {
171 
172 	print_man_node(m, n, mh, h);
173 	if (n->next)
174 		print_man_nodelist(m, n->next, mh, h);
175 }
176 
177 
178 static void
179 print_man_node(MAN_ARGS)
180 {
181 	int		 child;
182 	struct tag	*t;
183 
184 	child = 1;
185 	t = h->tags.head;
186 
187 	bufinit(h);
188 
189 	/*
190 	 * FIXME: embedded elements within next-line scopes (e.g., `br'
191 	 * within an empty `B') will cause formatting to be forgotten
192 	 * due to scope closing out.
193 	 */
194 
195 	switch (n->type) {
196 	case (MAN_ROOT):
197 		child = man_root_pre(m, n, mh, h);
198 		break;
199 	case (MAN_TEXT):
200 		print_text(h, n->string);
201 		if (MANH_LITERAL & mh->fl)
202 			print_otag(h, TAG_BR, 0, NULL);
203 		return;
204 	case (MAN_TBL):
205 		print_tbl(h, n->span);
206 		break;
207 	default:
208 		/*
209 		 * Close out scope of font prior to opening a macro
210 		 * scope.  Assert that the metafont is on the top of the
211 		 * stack (it's never nested).
212 		 */
213 		if (HTMLFONT_NONE != h->metac) {
214 			h->metal = h->metac;
215 			h->metac = HTMLFONT_NONE;
216 		}
217 		if (mans[n->tok].pre)
218 			child = (*mans[n->tok].pre)(m, n, mh, h);
219 		break;
220 	}
221 
222 	if (child && n->child)
223 		print_man_nodelist(m, n->child, mh, h);
224 
225 	/* This will automatically close out any font scope. */
226 	print_stagq(h, t);
227 
228 	bufinit(h);
229 
230 	switch (n->type) {
231 	case (MAN_ROOT):
232 		man_root_post(m, n, mh, h);
233 		break;
234 	case (MAN_TBL):
235 		break;
236 	default:
237 		if (mans[n->tok].post)
238 			(*mans[n->tok].post)(m, n, mh, h);
239 		break;
240 	}
241 }
242 
243 
244 static int
245 a2width(const struct man_node *n, struct roffsu *su)
246 {
247 
248 	if (MAN_TEXT != n->type)
249 		return(0);
250 	if (a2roffsu(n->string, su, SCALE_BU))
251 		return(1);
252 
253 	return(0);
254 }
255 
256 
257 /* ARGSUSED */
258 static int
259 man_root_pre(MAN_ARGS)
260 {
261 	struct htmlpair	 tag[3];
262 	struct tag	*t, *tt;
263 	char		 b[BUFSIZ], title[BUFSIZ];
264 
265 	b[0] = 0;
266 	if (m->vol)
267 		(void)strlcat(b, m->vol, BUFSIZ);
268 
269 	snprintf(title, BUFSIZ - 1, "%s(%s)", m->title, m->msec);
270 
271 	PAIR_SUMMARY_INIT(&tag[0], "Document Header");
272 	PAIR_CLASS_INIT(&tag[1], "head");
273 	if (NULL == h->style) {
274 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
275 		t = print_otag(h, TAG_TABLE, 3, tag);
276 		PAIR_INIT(&tag[0], ATTR_WIDTH, "30%");
277 		print_otag(h, TAG_COL, 1, tag);
278 		print_otag(h, TAG_COL, 1, tag);
279 		print_otag(h, TAG_COL, 1, tag);
280 	} else
281 		t = print_otag(h, TAG_TABLE, 2, tag);
282 
283 	print_otag(h, TAG_TBODY, 0, NULL);
284 
285 	tt = print_otag(h, TAG_TR, 0, NULL);
286 
287 	PAIR_CLASS_INIT(&tag[0], "head-ltitle");
288 	print_otag(h, TAG_TD, 1, tag);
289 
290 	print_text(h, title);
291 	print_stagq(h, tt);
292 
293 	PAIR_CLASS_INIT(&tag[0], "head-vol");
294 	if (NULL == h->style) {
295 		PAIR_INIT(&tag[1], ATTR_ALIGN, "center");
296 		print_otag(h, TAG_TD, 2, tag);
297 	} else
298 		print_otag(h, TAG_TD, 1, tag);
299 
300 	print_text(h, b);
301 	print_stagq(h, tt);
302 
303 	PAIR_CLASS_INIT(&tag[0], "head-rtitle");
304 	if (NULL == h->style) {
305 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
306 		print_otag(h, TAG_TD, 2, tag);
307 	} else
308 		print_otag(h, TAG_TD, 1, tag);
309 
310 	print_text(h, title);
311 	print_tagq(h, t);
312 	return(1);
313 }
314 
315 
316 /* ARGSUSED */
317 static void
318 man_root_post(MAN_ARGS)
319 {
320 	struct htmlpair	 tag[3];
321 	struct tag	*t, *tt;
322 	char		 b[DATESIZ];
323 
324 	if (m->rawdate)
325 		strlcpy(b, m->rawdate, DATESIZ);
326 	else
327 		time2a(m->date, b, DATESIZ);
328 
329 	PAIR_SUMMARY_INIT(&tag[0], "Document Footer");
330 	PAIR_CLASS_INIT(&tag[1], "foot");
331 	if (NULL == h->style) {
332 		PAIR_INIT(&tag[2], ATTR_WIDTH, "100%");
333 		t = print_otag(h, TAG_TABLE, 3, tag);
334 		PAIR_INIT(&tag[0], ATTR_WIDTH, "50%");
335 		print_otag(h, TAG_COL, 1, tag);
336 		print_otag(h, TAG_COL, 1, tag);
337 	} else
338 		t = print_otag(h, TAG_TABLE, 2, tag);
339 
340 	tt = print_otag(h, TAG_TR, 0, NULL);
341 
342 	PAIR_CLASS_INIT(&tag[0], "foot-date");
343 	print_otag(h, TAG_TD, 1, tag);
344 
345 	print_text(h, b);
346 	print_stagq(h, tt);
347 
348 	PAIR_CLASS_INIT(&tag[0], "foot-os");
349 	if (NULL == h->style) {
350 		PAIR_INIT(&tag[1], ATTR_ALIGN, "right");
351 		print_otag(h, TAG_TD, 2, tag);
352 	} else
353 		print_otag(h, TAG_TD, 1, tag);
354 
355 	if (m->source)
356 		print_text(h, m->source);
357 	print_tagq(h, t);
358 }
359 
360 
361 
362 /* ARGSUSED */
363 static int
364 man_br_pre(MAN_ARGS)
365 {
366 	struct roffsu	 su;
367 	struct htmlpair	 tag;
368 
369 	SCALE_VS_INIT(&su, 1);
370 
371 	if (MAN_sp == n->tok) {
372 		if (n->child)
373 			a2roffsu(n->child->string, &su, SCALE_VS);
374 	} else
375 		su.scale = 0;
376 
377 	bufcat_su(h, "height", &su);
378 	PAIR_STYLE_INIT(&tag, h);
379 	print_otag(h, TAG_DIV, 1, &tag);
380 
381 	/* So the div isn't empty: */
382 	print_text(h, "\\~");
383 
384 	return(0);
385 }
386 
387 
388 /* ARGSUSED */
389 static int
390 man_SH_pre(MAN_ARGS)
391 {
392 	struct htmlpair	 tag;
393 
394 	if (MAN_BLOCK == n->type) {
395 		PAIR_CLASS_INIT(&tag, "section");
396 		print_otag(h, TAG_DIV, 1, &tag);
397 		return(1);
398 	} else if (MAN_BODY == n->type)
399 		return(1);
400 
401 	print_otag(h, TAG_H1, 0, NULL);
402 	return(1);
403 }
404 
405 
406 /* ARGSUSED */
407 static int
408 man_alt_pre(MAN_ARGS)
409 {
410 	const struct man_node	*nn;
411 	int		 i;
412 	enum htmltag	 fp;
413 	struct tag	*t;
414 
415 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
416 		t = NULL;
417 		switch (n->tok) {
418 		case (MAN_BI):
419 			fp = i % 2 ? TAG_I : TAG_B;
420 			break;
421 		case (MAN_IB):
422 			fp = i % 2 ? TAG_B : TAG_I;
423 			break;
424 		case (MAN_RI):
425 			fp = i % 2 ? TAG_I : TAG_MAX;
426 			break;
427 		case (MAN_IR):
428 			fp = i % 2 ? TAG_MAX : TAG_I;
429 			break;
430 		case (MAN_BR):
431 			fp = i % 2 ? TAG_MAX : TAG_B;
432 			break;
433 		case (MAN_RB):
434 			fp = i % 2 ? TAG_B : TAG_MAX;
435 			break;
436 		default:
437 			abort();
438 			/* NOTREACHED */
439 		}
440 
441 		if (i)
442 			h->flags |= HTML_NOSPACE;
443 
444 		if (TAG_MAX != fp)
445 			t = print_otag(h, fp, 0, NULL);
446 
447 		print_man_node(m, nn, mh, h);
448 
449 		if (t)
450 			print_tagq(h, t);
451 	}
452 
453 	return(0);
454 }
455 
456 
457 /* ARGSUSED */
458 static int
459 man_SM_pre(MAN_ARGS)
460 {
461 
462 	print_otag(h, TAG_SMALL, 0, NULL);
463 	if (MAN_SB == n->tok)
464 		print_otag(h, TAG_B, 0, NULL);
465 	return(1);
466 }
467 
468 
469 /* ARGSUSED */
470 static int
471 man_SS_pre(MAN_ARGS)
472 {
473 	struct htmlpair	 tag;
474 
475 	if (MAN_BLOCK == n->type) {
476 		PAIR_CLASS_INIT(&tag, "subsection");
477 		print_otag(h, TAG_DIV, 1, &tag);
478 		return(1);
479 	} else if (MAN_BODY == n->type)
480 		return(1);
481 
482 	print_otag(h, TAG_H2, 0, NULL);
483 	return(1);
484 }
485 
486 
487 /* ARGSUSED */
488 static int
489 man_PP_pre(MAN_ARGS)
490 {
491 
492 	if (MAN_HEAD == n->type)
493 		return(0);
494 	else if (MAN_BODY == n->type && n->prev)
495 		print_otag(h, TAG_P, 0, NULL);
496 
497 	return(1);
498 }
499 
500 
501 /* ARGSUSED */
502 static int
503 man_IP_pre(MAN_ARGS)
504 {
505 	struct roffsu		 su;
506 	struct htmlpair	 	 tag;
507 	const struct man_node	*nn;
508 
509 	/*
510 	 * This scattering of 1-BU margins and pads is to make sure that
511 	 * when text overruns its box, the subsequent text isn't flush
512 	 * up against it.  However, the rest of the right-hand box must
513 	 * also be adjusted in consideration of this 1-BU space.
514 	 */
515 
516 	if (MAN_BODY == n->type) {
517 		print_otag(h, TAG_TD, 0, NULL);
518 		return(1);
519 	}
520 
521 	nn = MAN_BLOCK == n->type ?
522 		n->head->child : n->parent->head->child;
523 
524 	SCALE_HS_INIT(&su, INDENT);
525 
526 	/* Width is the second token. */
527 
528 	if (MAN_IP == n->tok && NULL != nn)
529 		if (NULL != (nn = nn->next))
530 			a2width(nn, &su);
531 
532 	/* Width is the first token. */
533 
534 	if (MAN_TP == n->tok && NULL != nn) {
535 		/* Skip past non-text children. */
536 		while (nn && MAN_TEXT != nn->type)
537 			nn = nn->next;
538 		if (nn)
539 			a2width(nn, &su);
540 	}
541 
542 	if (MAN_BLOCK == n->type) {
543 		print_otag(h, TAG_P, 0, NULL);
544 		print_otag(h, TAG_TABLE, 0, NULL);
545 		bufcat_su(h, "width", &su);
546 		PAIR_STYLE_INIT(&tag, h);
547 		print_otag(h, TAG_COL, 1, &tag);
548 		print_otag(h, TAG_COL, 0, NULL);
549 		print_otag(h, TAG_TBODY, 0, NULL);
550 		print_otag(h, TAG_TR, 0, NULL);
551 		return(1);
552 	}
553 
554 	print_otag(h, TAG_TD, 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 
572 /* ARGSUSED */
573 static int
574 man_HP_pre(MAN_ARGS)
575 {
576 	struct htmlpair	 tag;
577 	struct roffsu	 su;
578 	const struct man_node *np;
579 
580 	np = MAN_BLOCK == n->type ?
581 		n->head->child :
582 		n->parent->head->child;
583 
584 	if (NULL == np || ! a2width(np, &su))
585 		SCALE_HS_INIT(&su, INDENT);
586 
587 	if (MAN_HEAD == n->type) {
588 		print_otag(h, TAG_TD, 0, NULL);
589 		return(0);
590 	} else if (MAN_BLOCK == n->type) {
591 		print_otag(h, TAG_P, 0, NULL);
592 		print_otag(h, TAG_TABLE, 0, NULL);
593 		bufcat_su(h, "width", &su);
594 		PAIR_STYLE_INIT(&tag, h);
595 		print_otag(h, TAG_COL, 1, &tag);
596 		print_otag(h, TAG_COL, 0, NULL);
597 		print_otag(h, TAG_TBODY, 0, NULL);
598 		print_otag(h, TAG_TR, 0, NULL);
599 		return(1);
600 	}
601 
602 	su.scale = -su.scale;
603 	bufcat_su(h, "text-indent", &su);
604 	PAIR_STYLE_INIT(&tag, h);
605 	print_otag(h, TAG_TD, 1, &tag);
606 	return(1);
607 }
608 
609 
610 /* ARGSUSED */
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 
620 /* ARGSUSED */
621 static int
622 man_I_pre(MAN_ARGS)
623 {
624 
625 	print_otag(h, TAG_I, 0, NULL);
626 	return(1);
627 }
628 
629 
630 /* ARGSUSED */
631 static int
632 man_literal_pre(MAN_ARGS)
633 {
634 
635 	if (MAN_nf == n->tok) {
636 		print_otag(h, TAG_BR, 0, NULL);
637 		mh->fl |= MANH_LITERAL;
638 	} else
639 		mh->fl &= ~MANH_LITERAL;
640 
641 	return(1);
642 }
643 
644 
645 /* ARGSUSED */
646 static int
647 man_in_pre(MAN_ARGS)
648 {
649 
650 	print_otag(h, TAG_BR, 0, NULL);
651 	return(0);
652 }
653 
654 
655 /* ARGSUSED */
656 static int
657 man_ign_pre(MAN_ARGS)
658 {
659 
660 	return(0);
661 }
662 
663 
664 /* ARGSUSED */
665 static int
666 man_RS_pre(MAN_ARGS)
667 {
668 	struct htmlpair	 tag;
669 	struct roffsu	 su;
670 
671 	if (MAN_HEAD == n->type)
672 		return(0);
673 	else if (MAN_BODY == n->type)
674 		return(1);
675 
676 	SCALE_HS_INIT(&su, INDENT);
677 	if (n->head->child)
678 		a2width(n->head->child, &su);
679 
680 	bufcat_su(h, "margin-left", &su);
681 	PAIR_STYLE_INIT(&tag, h);
682 	print_otag(h, TAG_DIV, 1, &tag);
683 	return(1);
684 }
685