xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision d59bb9942320b767f2a19aaa7690c8c6e30b724c)
1 /*	$OpenBSD: man_html.c,v 1.86 2017/02/05 18:13:28 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013, 2014, 2015, 2017 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 AUTHORS DISCLAIM ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 <sys/types.h>
19 
20 #include <assert.h>
21 #include <ctype.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 
26 #include "mandoc_aux.h"
27 #include "roff.h"
28 #include "man.h"
29 #include "out.h"
30 #include "html.h"
31 #include "main.h"
32 
33 /* FIXME: have PD set the default vspace width. */
34 
35 #define	INDENT		  5
36 
37 #define	MAN_ARGS	  const struct roff_meta *man, \
38 			  const struct roff_node *n, \
39 			  struct html *h
40 
41 struct	htmlman {
42 	int		(*pre)(MAN_ARGS);
43 	int		(*post)(MAN_ARGS);
44 };
45 
46 static	void		  print_bvspace(struct html *,
47 				const struct roff_node *);
48 static	void		  print_man_head(MAN_ARGS);
49 static	void		  print_man_nodelist(MAN_ARGS);
50 static	void		  print_man_node(MAN_ARGS);
51 static	int		  fillmode(struct html *, int);
52 static	int		  a2width(const struct roff_node *,
53 				struct roffsu *);
54 static	int		  man_B_pre(MAN_ARGS);
55 static	int		  man_HP_pre(MAN_ARGS);
56 static	int		  man_IP_pre(MAN_ARGS);
57 static	int		  man_I_pre(MAN_ARGS);
58 static	int		  man_OP_pre(MAN_ARGS);
59 static	int		  man_PP_pre(MAN_ARGS);
60 static	int		  man_RS_pre(MAN_ARGS);
61 static	int		  man_SH_pre(MAN_ARGS);
62 static	int		  man_SM_pre(MAN_ARGS);
63 static	int		  man_SS_pre(MAN_ARGS);
64 static	int		  man_UR_pre(MAN_ARGS);
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	void		  man_root_post(MAN_ARGS);
70 static	void		  man_root_pre(MAN_ARGS);
71 
72 static	const struct htmlman mans[MAN_MAX] = {
73 	{ man_br_pre, NULL }, /* br */
74 	{ NULL, NULL }, /* TH */
75 	{ man_SH_pre, NULL }, /* SH */
76 	{ man_SS_pre, NULL }, /* SS */
77 	{ man_IP_pre, NULL }, /* TP */
78 	{ man_PP_pre, NULL }, /* LP */
79 	{ man_PP_pre, NULL }, /* PP */
80 	{ man_PP_pre, NULL }, /* P */
81 	{ man_IP_pre, NULL }, /* IP */
82 	{ man_HP_pre, NULL }, /* HP */
83 	{ man_SM_pre, NULL }, /* SM */
84 	{ man_SM_pre, NULL }, /* SB */
85 	{ man_alt_pre, NULL }, /* BI */
86 	{ man_alt_pre, NULL }, /* IB */
87 	{ man_alt_pre, NULL }, /* BR */
88 	{ man_alt_pre, NULL }, /* RB */
89 	{ NULL, NULL }, /* R */
90 	{ man_B_pre, NULL }, /* B */
91 	{ man_I_pre, NULL }, /* I */
92 	{ man_alt_pre, NULL }, /* IR */
93 	{ man_alt_pre, NULL }, /* RI */
94 	{ man_br_pre, NULL }, /* sp */
95 	{ NULL, NULL }, /* nf */
96 	{ NULL, NULL }, /* fi */
97 	{ NULL, NULL }, /* RE */
98 	{ man_RS_pre, NULL }, /* RS */
99 	{ man_ign_pre, NULL }, /* DT */
100 	{ man_ign_pre, NULL }, /* UC */
101 	{ man_ign_pre, NULL }, /* PD */
102 	{ man_ign_pre, NULL }, /* AT */
103 	{ man_in_pre, NULL }, /* in */
104 	{ man_ign_pre, NULL }, /* ft */
105 	{ man_OP_pre, NULL }, /* OP */
106 	{ NULL, NULL }, /* EX */
107 	{ NULL, NULL }, /* EE */
108 	{ man_UR_pre, NULL }, /* UR */
109 	{ NULL, NULL }, /* UE */
110 	{ man_ign_pre, NULL }, /* ll */
111 };
112 
113 
114 /*
115  * Printing leading vertical space before a block.
116  * This is used for the paragraph macros.
117  * The rules are pretty simple, since there's very little nesting going
118  * on here.  Basically, if we're the first within another block (SS/SH),
119  * then don't emit vertical space.  If we are (RS), then do.  If not the
120  * first, print it.
121  */
122 static void
123 print_bvspace(struct html *h, const struct roff_node *n)
124 {
125 
126 	if (n->body && n->body->child)
127 		if (n->body->child->type == ROFFT_TBL)
128 			return;
129 
130 	if (n->parent->type == ROFFT_ROOT || n->parent->tok != MAN_RS)
131 		if (NULL == n->prev)
132 			return;
133 
134 	print_paragraph(h);
135 }
136 
137 void
138 html_man(void *arg, const struct roff_man *man)
139 {
140 	struct html	*h;
141 	struct tag	*t;
142 
143 	h = (struct html *)arg;
144 
145 	if ((h->oflags & HTML_FRAGMENT) == 0) {
146 		print_gen_decls(h);
147 		print_otag(h, TAG_HTML, "");
148 		t = print_otag(h, TAG_HEAD, "");
149 		print_man_head(&man->meta, man->first, h);
150 		print_tagq(h, t);
151 		print_otag(h, TAG_BODY, "");
152 	}
153 
154 	man_root_pre(&man->meta, man->first, h);
155 	t = print_otag(h, TAG_DIV, "c", "manual-text");
156 	print_man_nodelist(&man->meta, man->first->child, h);
157 	print_tagq(h, t);
158 	man_root_post(&man->meta, man->first, h);
159 	print_tagq(h, NULL);
160 }
161 
162 static void
163 print_man_head(MAN_ARGS)
164 {
165 	char	*cp;
166 
167 	print_gen_head(h);
168 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
169 	print_otag(h, TAG_TITLE, "");
170 	print_text(h, cp);
171 	free(cp);
172 }
173 
174 static void
175 print_man_nodelist(MAN_ARGS)
176 {
177 
178 	while (n != NULL) {
179 		print_man_node(man, n, h);
180 		n = n->next;
181 	}
182 }
183 
184 static void
185 print_man_node(MAN_ARGS)
186 {
187 	static int	 want_fillmode = MAN_fi;
188 	static int	 save_fillmode;
189 
190 	struct tag	*t;
191 	int		 child;
192 
193 	/*
194 	 * Handle fill mode switch requests up front,
195 	 * they would just cause trouble in the subsequent code.
196 	 */
197 
198 	switch (n->tok) {
199 	case MAN_nf:
200 	case MAN_EX:
201 		want_fillmode = MAN_nf;
202 		return;
203 	case MAN_fi:
204 	case MAN_EE:
205 		want_fillmode = MAN_fi;
206 		if (fillmode(h, 0) == MAN_fi)
207 			print_otag(h, TAG_BR, "");
208 		return;
209 	default:
210 		break;
211 	}
212 
213 	/* Set up fill mode for the upcoming node. */
214 
215 	switch (n->type) {
216 	case ROFFT_BLOCK:
217 		save_fillmode = 0;
218 		/* Some block macros suspend or cancel .nf. */
219 		switch (n->tok) {
220 		case MAN_TP:  /* Tagged paragraphs		*/
221 		case MAN_IP:  /* temporarily disable .nf	*/
222 		case MAN_HP:  /* for the head.			*/
223 			save_fillmode = want_fillmode;
224 			/* FALLTHROUGH */
225 		case MAN_SH:  /* Section headers		*/
226 		case MAN_SS:  /* permanently cancel .nf.	*/
227 			want_fillmode = MAN_fi;
228 			/* FALLTHROUGH */
229 		case MAN_PP:  /* These have no head.		*/
230 		case MAN_LP:  /* They will simply		*/
231 		case MAN_P:   /* reopen .nf in the body.	*/
232 		case MAN_RS:
233 		case MAN_UR:
234 			fillmode(h, MAN_fi);
235 			break;
236 		default:
237 			break;
238 		}
239 		break;
240 	case ROFFT_TBL:
241 		fillmode(h, MAN_fi);
242 		break;
243 	case ROFFT_ELEM:
244 		/*
245 		 * Some in-line macros produce tags and/or text
246 		 * in the handler, so they require fill mode to be
247 		 * configured up front just like for text nodes.
248 		 * For the others, keep the traditional approach
249 		 * of doing the same, for now.
250 		 */
251 		fillmode(h, want_fillmode);
252 		break;
253 	case ROFFT_TEXT:
254 		if (fillmode(h, want_fillmode) == MAN_fi &&
255 		    want_fillmode == MAN_fi &&
256 		    n->flags & NODE_LINE && *n->string == ' ')
257 			print_otag(h, TAG_BR, "");
258 		if (*n->string != '\0')
259 			break;
260 		print_paragraph(h);
261 		return;
262 	default:
263 		break;
264 	}
265 
266 	/* Produce output for this node. */
267 
268 	child = 1;
269 	switch (n->type) {
270 	case ROFFT_TEXT:
271 		t = h->tag;
272 		print_text(h, n->string);
273 		break;
274 	case ROFFT_EQN:
275 		t = h->tag;
276 		print_eqn(h, n->eqn);
277 		break;
278 	case ROFFT_TBL:
279 		/*
280 		 * This will take care of initialising all of the table
281 		 * state data for the first table, then tearing it down
282 		 * for the last one.
283 		 */
284 		print_tbl(h, n->span);
285 		return;
286 	default:
287 		/*
288 		 * Close out scope of font prior to opening a macro
289 		 * scope.
290 		 */
291 		if (HTMLFONT_NONE != h->metac) {
292 			h->metal = h->metac;
293 			h->metac = HTMLFONT_NONE;
294 		}
295 
296 		/*
297 		 * Close out the current table, if it's open, and unset
298 		 * the "meta" table state.  This will be reopened on the
299 		 * next table element.
300 		 */
301 		if (h->tblt)
302 			print_tblclose(h);
303 
304 		t = h->tag;
305 		if (mans[n->tok].pre)
306 			child = (*mans[n->tok].pre)(man, n, h);
307 
308 		/* Some block macros resume .nf in the body. */
309 		if (save_fillmode && n->type == ROFFT_BODY)
310 			want_fillmode = save_fillmode;
311 
312 		break;
313 	}
314 
315 	if (child && n->child)
316 		print_man_nodelist(man, n->child, h);
317 
318 	/* This will automatically close out any font scope. */
319 	print_stagq(h, t);
320 
321 	if (fillmode(h, 0) == MAN_nf &&
322 	    n->next != NULL && n->next->flags & NODE_LINE)
323 		print_endline(h);
324 }
325 
326 /*
327  * MAN_nf switches to no-fill mode, MAN_fi to fill mode.
328  * Other arguments do not switch.
329  * The old mode is returned.
330  */
331 static int
332 fillmode(struct html *h, int want)
333 {
334 	struct tag	*pre;
335 	int		 had;
336 
337 	for (pre = h->tag; pre != NULL; pre = pre->next)
338 		if (pre->tag == TAG_PRE)
339 			break;
340 
341 	had = pre == NULL ? MAN_fi : MAN_nf;
342 
343 	if (want && want != had) {
344 		if (want == MAN_nf)
345 			print_otag(h, TAG_PRE, "");
346 		else
347 			print_tagq(h, pre);
348 	}
349 	return had;
350 }
351 
352 static int
353 a2width(const struct roff_node *n, struct roffsu *su)
354 {
355 
356 	if (n->type != ROFFT_TEXT)
357 		return 0;
358 	if (a2roffsu(n->string, su, SCALE_EN))
359 		return 1;
360 
361 	return 0;
362 }
363 
364 static void
365 man_root_pre(MAN_ARGS)
366 {
367 	struct tag	*t, *tt;
368 	char		*title;
369 
370 	assert(man->title);
371 	assert(man->msec);
372 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
373 
374 	t = print_otag(h, TAG_TABLE, "c", "head");
375 	tt = print_otag(h, TAG_TR, "");
376 
377 	print_otag(h, TAG_TD, "c", "head-ltitle");
378 	print_text(h, title);
379 	print_stagq(h, tt);
380 
381 	print_otag(h, TAG_TD, "c", "head-vol");
382 	if (NULL != man->vol)
383 		print_text(h, man->vol);
384 	print_stagq(h, tt);
385 
386 	print_otag(h, TAG_TD, "c", "head-rtitle");
387 	print_text(h, title);
388 	print_tagq(h, t);
389 	free(title);
390 }
391 
392 static void
393 man_root_post(MAN_ARGS)
394 {
395 	struct tag	*t, *tt;
396 
397 	t = print_otag(h, TAG_TABLE, "c", "foot");
398 	tt = print_otag(h, TAG_TR, "");
399 
400 	print_otag(h, TAG_TD, "c", "foot-date");
401 	print_text(h, man->date);
402 	print_stagq(h, tt);
403 
404 	print_otag(h, TAG_TD, "c", "foot-os");
405 	if (man->os)
406 		print_text(h, man->os);
407 	print_tagq(h, t);
408 }
409 
410 
411 static int
412 man_br_pre(MAN_ARGS)
413 {
414 	struct roffsu	 su;
415 
416 	SCALE_VS_INIT(&su, 1);
417 
418 	if (MAN_sp == n->tok) {
419 		if (NULL != (n = n->child))
420 			if ( ! a2roffsu(n->string, &su, SCALE_VS))
421 				su.scale = 1.0;
422 	} else
423 		su.scale = 0.0;
424 
425 	print_otag(h, TAG_DIV, "suh", &su);
426 
427 	/* So the div isn't empty: */
428 	print_text(h, "\\~");
429 
430 	return 0;
431 }
432 
433 static int
434 man_SH_pre(MAN_ARGS)
435 {
436 	if (n->type == ROFFT_HEAD)
437 		print_otag(h, TAG_H1, "c", "Sh");
438 	return 1;
439 }
440 
441 static int
442 man_alt_pre(MAN_ARGS)
443 {
444 	const struct roff_node	*nn;
445 	int		 i;
446 	enum htmltag	 fp;
447 	struct tag	*t;
448 
449 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
450 		switch (n->tok) {
451 		case MAN_BI:
452 			fp = i % 2 ? TAG_I : TAG_B;
453 			break;
454 		case MAN_IB:
455 			fp = i % 2 ? TAG_B : TAG_I;
456 			break;
457 		case MAN_RI:
458 			fp = i % 2 ? TAG_I : TAG_MAX;
459 			break;
460 		case MAN_IR:
461 			fp = i % 2 ? TAG_MAX : TAG_I;
462 			break;
463 		case MAN_BR:
464 			fp = i % 2 ? TAG_MAX : TAG_B;
465 			break;
466 		case MAN_RB:
467 			fp = i % 2 ? TAG_B : TAG_MAX;
468 			break;
469 		default:
470 			abort();
471 		}
472 
473 		if (i)
474 			h->flags |= HTML_NOSPACE;
475 
476 		if (fp != TAG_MAX)
477 			t = print_otag(h, fp, "");
478 
479 		print_text(h, nn->string);
480 
481 		if (fp != TAG_MAX)
482 			print_tagq(h, t);
483 	}
484 	return 0;
485 }
486 
487 static int
488 man_SM_pre(MAN_ARGS)
489 {
490 	print_otag(h, TAG_SMALL, "");
491 	if (MAN_SB == n->tok)
492 		print_otag(h, TAG_B, "");
493 	return 1;
494 }
495 
496 static int
497 man_SS_pre(MAN_ARGS)
498 {
499 	if (n->type == ROFFT_HEAD)
500 		print_otag(h, TAG_H2, "c", "Ss");
501 	return 1;
502 }
503 
504 static int
505 man_PP_pre(MAN_ARGS)
506 {
507 
508 	if (n->type == ROFFT_HEAD)
509 		return 0;
510 	else if (n->type == ROFFT_BLOCK)
511 		print_bvspace(h, n);
512 
513 	return 1;
514 }
515 
516 static int
517 man_IP_pre(MAN_ARGS)
518 {
519 	const struct roff_node	*nn;
520 
521 	if (n->type == ROFFT_BODY) {
522 		print_otag(h, TAG_DD, "c", "It-tag");
523 		return 1;
524 	} else if (n->type != ROFFT_HEAD) {
525 		print_otag(h, TAG_DL, "c", "Bl-tag");
526 		return 1;
527 	}
528 
529 	/* FIXME: width specification. */
530 
531 	print_otag(h, TAG_DT, "c", "It-tag");
532 
533 	/* For IP, only print the first header element. */
534 
535 	if (MAN_IP == n->tok && n->child)
536 		print_man_node(man, n->child, h);
537 
538 	/* For TP, only print next-line header elements. */
539 
540 	if (MAN_TP == n->tok) {
541 		nn = n->child;
542 		while (NULL != nn && 0 == (NODE_LINE & nn->flags))
543 			nn = nn->next;
544 		while (NULL != nn) {
545 			print_man_node(man, nn, h);
546 			nn = nn->next;
547 		}
548 	}
549 
550 	return 0;
551 }
552 
553 static int
554 man_HP_pre(MAN_ARGS)
555 {
556 	struct roffsu	 sum, sui;
557 	const struct roff_node *np;
558 
559 	if (n->type == ROFFT_HEAD)
560 		return 0;
561 	else if (n->type != ROFFT_BLOCK)
562 		return 1;
563 
564 	np = n->head->child;
565 
566 	if (np == NULL || !a2width(np, &sum))
567 		SCALE_HS_INIT(&sum, INDENT);
568 
569 	sui.unit = sum.unit;
570 	sui.scale = -sum.scale;
571 
572 	print_bvspace(h, n);
573 	print_otag(h, TAG_DIV, "csului", "Pp", &sum, &sui);
574 	return 1;
575 }
576 
577 static int
578 man_OP_pre(MAN_ARGS)
579 {
580 	struct tag	*tt;
581 
582 	print_text(h, "[");
583 	h->flags |= HTML_NOSPACE;
584 	tt = print_otag(h, TAG_SPAN, "c", "Op");
585 
586 	if (NULL != (n = n->child)) {
587 		print_otag(h, TAG_B, "");
588 		print_text(h, n->string);
589 	}
590 
591 	print_stagq(h, tt);
592 
593 	if (NULL != n && NULL != n->next) {
594 		print_otag(h, TAG_I, "");
595 		print_text(h, n->next->string);
596 	}
597 
598 	print_stagq(h, tt);
599 	h->flags |= HTML_NOSPACE;
600 	print_text(h, "]");
601 	return 0;
602 }
603 
604 static int
605 man_B_pre(MAN_ARGS)
606 {
607 	print_otag(h, TAG_B, "");
608 	return 1;
609 }
610 
611 static int
612 man_I_pre(MAN_ARGS)
613 {
614 	print_otag(h, TAG_I, "");
615 	return 1;
616 }
617 
618 static int
619 man_in_pre(MAN_ARGS)
620 {
621 	print_otag(h, TAG_BR, "");
622 	return 0;
623 }
624 
625 static int
626 man_ign_pre(MAN_ARGS)
627 {
628 
629 	return 0;
630 }
631 
632 static int
633 man_RS_pre(MAN_ARGS)
634 {
635 	struct roffsu	 su;
636 
637 	if (n->type == ROFFT_HEAD)
638 		return 0;
639 	else if (n->type == ROFFT_BODY)
640 		return 1;
641 
642 	SCALE_HS_INIT(&su, INDENT);
643 	if (n->head->child)
644 		a2width(n->head->child, &su);
645 
646 	print_otag(h, TAG_DIV, "sul", &su);
647 	return 1;
648 }
649 
650 static int
651 man_UR_pre(MAN_ARGS)
652 {
653 	n = n->child;
654 	assert(n->type == ROFFT_HEAD);
655 	if (n->child != NULL) {
656 		assert(n->child->type == ROFFT_TEXT);
657 		print_otag(h, TAG_A, "ch", "Lk", n->child->string);
658 	}
659 
660 	assert(n->next->type == ROFFT_BODY);
661 	if (n->next->child != NULL)
662 		n = n->next;
663 
664 	print_man_nodelist(man, n->child, h);
665 
666 	return 0;
667 }
668