xref: /illumos-gate/usr/src/cmd/mandoc/man_html.c (revision c3d26abc9ee97b4f60233556aadeb57e0bd30bb9)
1 /*	$Id: man_html.c,v 1.112 2015/03/03 21:11:34 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013, 2014, 2015 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 "config.h"
19 
20 #include <sys/types.h>
21 
22 #include <assert.h>
23 #include <ctype.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 
28 #include "mandoc_aux.h"
29 #include "man.h"
30 #include "out.h"
31 #include "html.h"
32 #include "main.h"
33 
34 /* TODO: preserve ident widths. */
35 /* FIXME: have PD set the default vspace width. */
36 
37 #define	INDENT		  5
38 
39 #define	MAN_ARGS	  const struct man_meta *man, \
40 			  const struct man_node *n, \
41 			  struct mhtml *mh, \
42 			  struct html *h
43 
44 struct	mhtml {
45 	int		  fl;
46 #define	MANH_LITERAL	 (1 << 0) /* literal context */
47 };
48 
49 struct	htmlman {
50 	int		(*pre)(MAN_ARGS);
51 	int		(*post)(MAN_ARGS);
52 };
53 
54 static	void		  print_bvspace(struct html *,
55 				const struct man_node *);
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 static	int		  a2width(const struct man_node *,
61 				struct roffsu *);
62 static	int		  man_B_pre(MAN_ARGS);
63 static	int		  man_HP_pre(MAN_ARGS);
64 static	int		  man_IP_pre(MAN_ARGS);
65 static	int		  man_I_pre(MAN_ARGS);
66 static	int		  man_OP_pre(MAN_ARGS);
67 static	int		  man_PP_pre(MAN_ARGS);
68 static	int		  man_RS_pre(MAN_ARGS);
69 static	int		  man_SH_pre(MAN_ARGS);
70 static	int		  man_SM_pre(MAN_ARGS);
71 static	int		  man_SS_pre(MAN_ARGS);
72 static	int		  man_UR_pre(MAN_ARGS);
73 static	int		  man_alt_pre(MAN_ARGS);
74 static	int		  man_br_pre(MAN_ARGS);
75 static	int		  man_ign_pre(MAN_ARGS);
76 static	int		  man_in_pre(MAN_ARGS);
77 static	int		  man_literal_pre(MAN_ARGS);
78 static	void		  man_root_post(MAN_ARGS);
79 static	void		  man_root_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 	{ man_br_pre, NULL }, /* sp */
104 	{ man_literal_pre, NULL }, /* nf */
105 	{ man_literal_pre, NULL }, /* fi */
106 	{ NULL, NULL }, /* RE */
107 	{ man_RS_pre, NULL }, /* RS */
108 	{ man_ign_pre, NULL }, /* DT */
109 	{ man_ign_pre, NULL }, /* UC */
110 	{ man_ign_pre, NULL }, /* PD */
111 	{ man_ign_pre, NULL }, /* AT */
112 	{ man_in_pre, NULL }, /* in */
113 	{ man_ign_pre, NULL }, /* ft */
114 	{ man_OP_pre, NULL }, /* OP */
115 	{ man_literal_pre, NULL }, /* EX */
116 	{ man_literal_pre, NULL }, /* EE */
117 	{ man_UR_pre, NULL }, /* UR */
118 	{ NULL, NULL }, /* UE */
119 	{ man_ign_pre, NULL }, /* ll */
120 };
121 
122 
123 /*
124  * Printing leading vertical space before a block.
125  * This is used for the paragraph macros.
126  * The rules are pretty simple, since there's very little nesting going
127  * on here.  Basically, if we're the first within another block (SS/SH),
128  * then don't emit vertical space.  If we are (RS), then do.  If not the
129  * first, print it.
130  */
131 static void
132 print_bvspace(struct html *h, const struct man_node *n)
133 {
134 
135 	if (n->body && n->body->child)
136 		if (MAN_TBL == n->body->child->type)
137 			return;
138 
139 	if (MAN_ROOT == n->parent->type || MAN_RS != n->parent->tok)
140 		if (NULL == n->prev)
141 			return;
142 
143 	print_paragraph(h);
144 }
145 
146 void
147 html_man(void *arg, const struct man *man)
148 {
149 	struct mhtml	 mh;
150 
151 	memset(&mh, 0, sizeof(struct mhtml));
152 	print_man(man_meta(man), man_node(man), &mh, (struct html *)arg);
153 	putchar('\n');
154 }
155 
156 static void
157 print_man(MAN_ARGS)
158 {
159 	struct tag	*t, *tt;
160 	struct htmlpair	 tag;
161 
162 	PAIR_CLASS_INIT(&tag, "mandoc");
163 
164 	if ( ! (HTML_FRAGMENT & h->oflags)) {
165 		print_gen_decls(h);
166 		t = print_otag(h, TAG_HTML, 0, NULL);
167 		tt = print_otag(h, TAG_HEAD, 0, NULL);
168 		print_man_head(man, n, mh, h);
169 		print_tagq(h, tt);
170 		print_otag(h, TAG_BODY, 0, NULL);
171 		print_otag(h, TAG_DIV, 1, &tag);
172 	} else
173 		t = print_otag(h, TAG_DIV, 1, &tag);
174 
175 	print_man_nodelist(man, n, mh, h);
176 	print_tagq(h, t);
177 }
178 
179 static void
180 print_man_head(MAN_ARGS)
181 {
182 
183 	print_gen_head(h);
184 	assert(man->title);
185 	assert(man->msec);
186 	bufcat_fmt(h, "%s(%s)", man->title, man->msec);
187 	print_otag(h, TAG_TITLE, 0, NULL);
188 	print_text(h, h->buf);
189 }
190 
191 static void
192 print_man_nodelist(MAN_ARGS)
193 {
194 
195 	while (n != NULL) {
196 		print_man_node(man, n, mh, h);
197 		n = n->next;
198 	}
199 }
200 
201 static void
202 print_man_node(MAN_ARGS)
203 {
204 	int		 child;
205 	struct tag	*t;
206 
207 	child = 1;
208 	t = h->tags.head;
209 
210 	switch (n->type) {
211 	case MAN_ROOT:
212 		man_root_pre(man, n, mh, h);
213 		break;
214 	case MAN_TEXT:
215 		if ('\0' == *n->string) {
216 			print_paragraph(h);
217 			return;
218 		}
219 		if (n->flags & MAN_LINE && (*n->string == ' ' ||
220 		    (n->prev != NULL && mh->fl & MANH_LITERAL &&
221 		     ! (h->flags & HTML_NONEWLINE))))
222 			print_otag(h, TAG_BR, 0, NULL);
223 		print_text(h, n->string);
224 		return;
225 	case MAN_EQN:
226 		if (n->flags & MAN_LINE)
227 			putchar('\n');
228 		print_eqn(h, n->eqn);
229 		break;
230 	case MAN_TBL:
231 		/*
232 		 * This will take care of initialising all of the table
233 		 * state data for the first table, then tearing it down
234 		 * for the last one.
235 		 */
236 		print_tbl(h, n->span);
237 		return;
238 	default:
239 		/*
240 		 * Close out scope of font prior to opening a macro
241 		 * scope.
242 		 */
243 		if (HTMLFONT_NONE != h->metac) {
244 			h->metal = h->metac;
245 			h->metac = HTMLFONT_NONE;
246 		}
247 
248 		/*
249 		 * Close out the current table, if it's open, and unset
250 		 * the "meta" table state.  This will be reopened on the
251 		 * next table element.
252 		 */
253 		if (h->tblt) {
254 			print_tblclose(h);
255 			t = h->tags.head;
256 		}
257 		if (mans[n->tok].pre)
258 			child = (*mans[n->tok].pre)(man, n, mh, h);
259 		break;
260 	}
261 
262 	if (child && n->child)
263 		print_man_nodelist(man, n->child, mh, h);
264 
265 	/* This will automatically close out any font scope. */
266 	print_stagq(h, t);
267 
268 	switch (n->type) {
269 	case MAN_ROOT:
270 		man_root_post(man, n, mh, h);
271 		break;
272 	case MAN_EQN:
273 		break;
274 	default:
275 		if (mans[n->tok].post)
276 			(*mans[n->tok].post)(man, n, mh, h);
277 		break;
278 	}
279 }
280 
281 static int
282 a2width(const struct man_node *n, struct roffsu *su)
283 {
284 
285 	if (MAN_TEXT != n->type)
286 		return(0);
287 	if (a2roffsu(n->string, su, SCALE_EN))
288 		return(1);
289 
290 	return(0);
291 }
292 
293 static void
294 man_root_pre(MAN_ARGS)
295 {
296 	struct htmlpair	 tag;
297 	struct tag	*t, *tt;
298 	char		*title;
299 
300 	assert(man->title);
301 	assert(man->msec);
302 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
303 
304 	PAIR_CLASS_INIT(&tag, "head");
305 	t = print_otag(h, TAG_TABLE, 1, &tag);
306 
307 	print_otag(h, TAG_TBODY, 0, NULL);
308 
309 	tt = print_otag(h, TAG_TR, 0, NULL);
310 
311 	PAIR_CLASS_INIT(&tag, "head-ltitle");
312 	print_otag(h, TAG_TD, 1, &tag);
313 	print_text(h, title);
314 	print_stagq(h, tt);
315 
316 	PAIR_CLASS_INIT(&tag, "head-vol");
317 	print_otag(h, TAG_TD, 1, &tag);
318 	if (NULL != man->vol)
319 		print_text(h, man->vol);
320 	print_stagq(h, tt);
321 
322 	PAIR_CLASS_INIT(&tag, "head-rtitle");
323 	print_otag(h, TAG_TD, 1, &tag);
324 	print_text(h, title);
325 	print_tagq(h, t);
326 	free(title);
327 }
328 
329 static void
330 man_root_post(MAN_ARGS)
331 {
332 	struct htmlpair	 tag;
333 	struct tag	*t, *tt;
334 
335 	PAIR_CLASS_INIT(&tag, "foot");
336 	t = print_otag(h, TAG_TABLE, 1, &tag);
337 
338 	tt = print_otag(h, TAG_TR, 0, NULL);
339 
340 	PAIR_CLASS_INIT(&tag, "foot-date");
341 	print_otag(h, TAG_TD, 1, &tag);
342 
343 	assert(man->date);
344 	print_text(h, man->date);
345 	print_stagq(h, tt);
346 
347 	PAIR_CLASS_INIT(&tag, "foot-os");
348 	print_otag(h, TAG_TD, 1, &tag);
349 
350 	if (man->source)
351 		print_text(h, man->source);
352 	print_tagq(h, t);
353 }
354 
355 
356 static int
357 man_br_pre(MAN_ARGS)
358 {
359 	struct roffsu	 su;
360 	struct htmlpair	 tag;
361 
362 	SCALE_VS_INIT(&su, 1);
363 
364 	if (MAN_sp == n->tok) {
365 		if (NULL != (n = n->child))
366 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
367 				su.scale = 1.0;
368 	} else
369 		su.scale = 0.0;
370 
371 	bufinit(h);
372 	bufcat_su(h, "height", &su);
373 	PAIR_STYLE_INIT(&tag, h);
374 	print_otag(h, TAG_DIV, 1, &tag);
375 
376 	/* So the div isn't empty: */
377 	print_text(h, "\\~");
378 
379 	return(0);
380 }
381 
382 static int
383 man_SH_pre(MAN_ARGS)
384 {
385 	struct htmlpair	 tag;
386 
387 	if (MAN_BLOCK == n->type) {
388 		mh->fl &= ~MANH_LITERAL;
389 		PAIR_CLASS_INIT(&tag, "section");
390 		print_otag(h, TAG_DIV, 1, &tag);
391 		return(1);
392 	} else if (MAN_BODY == n->type)
393 		return(1);
394 
395 	print_otag(h, TAG_H1, 0, NULL);
396 	return(1);
397 }
398 
399 static int
400 man_alt_pre(MAN_ARGS)
401 {
402 	const struct man_node	*nn;
403 	int		 i, savelit;
404 	enum htmltag	 fp;
405 	struct tag	*t;
406 
407 	if ((savelit = mh->fl & MANH_LITERAL))
408 		print_otag(h, TAG_BR, 0, NULL);
409 
410 	mh->fl &= ~MANH_LITERAL;
411 
412 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
413 		t = NULL;
414 		switch (n->tok) {
415 		case MAN_BI:
416 			fp = i % 2 ? TAG_I : TAG_B;
417 			break;
418 		case MAN_IB:
419 			fp = i % 2 ? TAG_B : TAG_I;
420 			break;
421 		case MAN_RI:
422 			fp = i % 2 ? TAG_I : TAG_MAX;
423 			break;
424 		case MAN_IR:
425 			fp = i % 2 ? TAG_MAX : TAG_I;
426 			break;
427 		case MAN_BR:
428 			fp = i % 2 ? TAG_MAX : TAG_B;
429 			break;
430 		case MAN_RB:
431 			fp = i % 2 ? TAG_B : TAG_MAX;
432 			break;
433 		default:
434 			abort();
435 			/* NOTREACHED */
436 		}
437 
438 		if (i)
439 			h->flags |= HTML_NOSPACE;
440 
441 		if (TAG_MAX != fp)
442 			t = print_otag(h, fp, 0, NULL);
443 
444 		print_man_node(man, nn, mh, h);
445 
446 		if (t)
447 			print_tagq(h, t);
448 	}
449 
450 	if (savelit)
451 		mh->fl |= MANH_LITERAL;
452 
453 	return(0);
454 }
455 
456 static int
457 man_SM_pre(MAN_ARGS)
458 {
459 
460 	print_otag(h, TAG_SMALL, 0, NULL);
461 	if (MAN_SB == n->tok)
462 		print_otag(h, TAG_B, 0, NULL);
463 	return(1);
464 }
465 
466 static int
467 man_SS_pre(MAN_ARGS)
468 {
469 	struct htmlpair	 tag;
470 
471 	if (MAN_BLOCK == n->type) {
472 		mh->fl &= ~MANH_LITERAL;
473 		PAIR_CLASS_INIT(&tag, "subsection");
474 		print_otag(h, TAG_DIV, 1, &tag);
475 		return(1);
476 	} else if (MAN_BODY == n->type)
477 		return(1);
478 
479 	print_otag(h, TAG_H2, 0, NULL);
480 	return(1);
481 }
482 
483 static int
484 man_PP_pre(MAN_ARGS)
485 {
486 
487 	if (MAN_HEAD == n->type)
488 		return(0);
489 	else if (MAN_BLOCK == n->type)
490 		print_bvspace(h, n);
491 
492 	return(1);
493 }
494 
495 static int
496 man_IP_pre(MAN_ARGS)
497 {
498 	const struct man_node	*nn;
499 
500 	if (MAN_BODY == n->type) {
501 		print_otag(h, TAG_DD, 0, NULL);
502 		return(1);
503 	} else if (MAN_HEAD != n->type) {
504 		print_otag(h, TAG_DL, 0, NULL);
505 		return(1);
506 	}
507 
508 	/* FIXME: width specification. */
509 
510 	print_otag(h, TAG_DT, 0, NULL);
511 
512 	/* For IP, only print the first header element. */
513 
514 	if (MAN_IP == n->tok && n->child)
515 		print_man_node(man, n->child, mh, h);
516 
517 	/* For TP, only print next-line header elements. */
518 
519 	if (MAN_TP == n->tok) {
520 		nn = n->child;
521 		while (NULL != nn && 0 == (MAN_LINE & nn->flags))
522 			nn = nn->next;
523 		while (NULL != nn) {
524 			print_man_node(man, nn, mh, h);
525 			nn = nn->next;
526 		}
527 	}
528 
529 	return(0);
530 }
531 
532 static int
533 man_HP_pre(MAN_ARGS)
534 {
535 	struct htmlpair	 tag[2];
536 	struct roffsu	 su;
537 	const struct man_node *np;
538 
539 	if (MAN_HEAD == n->type)
540 		return(0);
541 	else if (MAN_BLOCK != n->type)
542 		return(1);
543 
544 	np = n->head->child;
545 
546 	if (NULL == np || ! a2width(np, &su))
547 		SCALE_HS_INIT(&su, INDENT);
548 
549 	bufinit(h);
550 
551 	print_bvspace(h, n);
552 	bufcat_su(h, "margin-left", &su);
553 	su.scale = -su.scale;
554 	bufcat_su(h, "text-indent", &su);
555 	PAIR_STYLE_INIT(&tag[0], h);
556 	PAIR_CLASS_INIT(&tag[1], "spacer");
557 	print_otag(h, TAG_DIV, 2, tag);
558 	return(1);
559 }
560 
561 static int
562 man_OP_pre(MAN_ARGS)
563 {
564 	struct tag	*tt;
565 	struct htmlpair	 tag;
566 
567 	print_text(h, "[");
568 	h->flags |= HTML_NOSPACE;
569 	PAIR_CLASS_INIT(&tag, "opt");
570 	tt = print_otag(h, TAG_SPAN, 1, &tag);
571 
572 	if (NULL != (n = n->child)) {
573 		print_otag(h, TAG_B, 0, NULL);
574 		print_text(h, n->string);
575 	}
576 
577 	print_stagq(h, tt);
578 
579 	if (NULL != n && NULL != n->next) {
580 		print_otag(h, TAG_I, 0, NULL);
581 		print_text(h, n->next->string);
582 	}
583 
584 	print_stagq(h, tt);
585 	h->flags |= HTML_NOSPACE;
586 	print_text(h, "]");
587 	return(0);
588 }
589 
590 static int
591 man_B_pre(MAN_ARGS)
592 {
593 
594 	print_otag(h, TAG_B, 0, NULL);
595 	return(1);
596 }
597 
598 static int
599 man_I_pre(MAN_ARGS)
600 {
601 
602 	print_otag(h, TAG_I, 0, NULL);
603 	return(1);
604 }
605 
606 static int
607 man_literal_pre(MAN_ARGS)
608 {
609 
610 	if (MAN_fi == n->tok || MAN_EE == n->tok) {
611 		print_otag(h, TAG_BR, 0, NULL);
612 		mh->fl &= ~MANH_LITERAL;
613 	} else
614 		mh->fl |= MANH_LITERAL;
615 
616 	return(0);
617 }
618 
619 static int
620 man_in_pre(MAN_ARGS)
621 {
622 
623 	print_otag(h, TAG_BR, 0, NULL);
624 	return(0);
625 }
626 
627 static int
628 man_ign_pre(MAN_ARGS)
629 {
630 
631 	return(0);
632 }
633 
634 static int
635 man_RS_pre(MAN_ARGS)
636 {
637 	struct htmlpair	 tag;
638 	struct roffsu	 su;
639 
640 	if (MAN_HEAD == n->type)
641 		return(0);
642 	else if (MAN_BODY == n->type)
643 		return(1);
644 
645 	SCALE_HS_INIT(&su, INDENT);
646 	if (n->head->child)
647 		a2width(n->head->child, &su);
648 
649 	bufinit(h);
650 	bufcat_su(h, "margin-left", &su);
651 	PAIR_STYLE_INIT(&tag, h);
652 	print_otag(h, TAG_DIV, 1, &tag);
653 	return(1);
654 }
655 
656 static int
657 man_UR_pre(MAN_ARGS)
658 {
659 	struct htmlpair		 tag[2];
660 
661 	n = n->child;
662 	assert(MAN_HEAD == n->type);
663 	if (n->nchild) {
664 		assert(MAN_TEXT == n->child->type);
665 		PAIR_CLASS_INIT(&tag[0], "link-ext");
666 		PAIR_HREF_INIT(&tag[1], n->child->string);
667 		print_otag(h, TAG_A, 2, tag);
668 	}
669 
670 	assert(MAN_BODY == n->next->type);
671 	if (n->next->nchild)
672 		n = n->next;
673 
674 	print_man_nodelist(man, n->child, mh, h);
675 
676 	return(0);
677 }
678