xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision d1df930ffab53da22f3324c32bed7ac5709915e6)
1 /*	$OpenBSD: man_html.c,v 1.109 2018/08/18 02:03:41 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013,2014,2015,2017,2018 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 "mandoc.h"
28 #include "roff.h"
29 #include "man.h"
30 #include "out.h"
31 #include "html.h"
32 #include "main.h"
33 
34 /* FIXME: have PD set the default vspace width. */
35 
36 #define	MAN_ARGS	  const struct roff_meta *man, \
37 			  const struct roff_node *n, \
38 			  struct html *h
39 
40 struct	man_html_act {
41 	int		(*pre)(MAN_ARGS);
42 	int		(*post)(MAN_ARGS);
43 };
44 
45 static	void		  print_bvspace(struct html *,
46 				const struct roff_node *);
47 static	void		  print_man_head(const struct roff_meta *,
48 				struct html *);
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		  man_B_pre(MAN_ARGS);
53 static	int		  man_HP_pre(MAN_ARGS);
54 static	int		  man_IP_pre(MAN_ARGS);
55 static	int		  man_I_pre(MAN_ARGS);
56 static	int		  man_OP_pre(MAN_ARGS);
57 static	int		  man_PP_pre(MAN_ARGS);
58 static	int		  man_RS_pre(MAN_ARGS);
59 static	int		  man_SH_pre(MAN_ARGS);
60 static	int		  man_SM_pre(MAN_ARGS);
61 static	int		  man_SS_pre(MAN_ARGS);
62 static	int		  man_SY_pre(MAN_ARGS);
63 static	int		  man_UR_pre(MAN_ARGS);
64 static	int		  man_alt_pre(MAN_ARGS);
65 static	int		  man_ign_pre(MAN_ARGS);
66 static	int		  man_in_pre(MAN_ARGS);
67 static	void		  man_root_post(const struct roff_meta *,
68 				struct html *);
69 static	void		  man_root_pre(const struct roff_meta *,
70 				struct html *);
71 
72 static	const struct man_html_act man_html_acts[MAN_MAX - MAN_TH] = {
73 	{ NULL, NULL }, /* TH */
74 	{ man_SH_pre, NULL }, /* SH */
75 	{ man_SS_pre, NULL }, /* SS */
76 	{ man_IP_pre, NULL }, /* TP */
77 	{ man_IP_pre, NULL }, /* TQ */
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 	{ NULL, NULL }, /* nf */
95 	{ NULL, NULL }, /* fi */
96 	{ NULL, NULL }, /* RE */
97 	{ man_RS_pre, NULL }, /* RS */
98 	{ man_ign_pre, NULL }, /* DT */
99 	{ man_ign_pre, NULL }, /* UC */
100 	{ man_ign_pre, NULL }, /* PD */
101 	{ man_ign_pre, NULL }, /* AT */
102 	{ man_in_pre, NULL }, /* in */
103 	{ man_SY_pre, NULL }, /* SY */
104 	{ NULL, NULL }, /* YS */
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_UR_pre, NULL }, /* MT */
111 	{ NULL, NULL }, /* ME */
112 };
113 
114 
115 /*
116  * Printing leading vertical space before a block.
117  * This is used for the paragraph macros.
118  * The rules are pretty simple, since there's very little nesting going
119  * on here.  Basically, if we're the first within another block (SS/SH),
120  * then don't emit vertical space.  If we are (RS), then do.  If not the
121  * first, print it.
122  */
123 static void
124 print_bvspace(struct html *h, const struct roff_node *n)
125 {
126 
127 	if (n->body && n->body->child)
128 		if (n->body->child->type == ROFFT_TBL)
129 			return;
130 
131 	if (n->parent->type == ROFFT_ROOT || n->parent->tok != MAN_RS)
132 		if (NULL == n->prev)
133 			return;
134 
135 	print_paragraph(h);
136 }
137 
138 void
139 html_man(void *arg, const struct roff_man *man)
140 {
141 	struct html		*h;
142 	struct roff_node	*n;
143 	struct tag		*t;
144 
145 	h = (struct html *)arg;
146 	n = man->first->child;
147 
148 	if ((h->oflags & HTML_FRAGMENT) == 0) {
149 		print_gen_decls(h);
150 		print_otag(h, TAG_HTML, "");
151 		if (n->type == ROFFT_COMMENT)
152 			print_gen_comment(h, n);
153 		t = print_otag(h, TAG_HEAD, "");
154 		print_man_head(&man->meta, h);
155 		print_tagq(h, t);
156 		print_otag(h, TAG_BODY, "");
157 	}
158 
159 	man_root_pre(&man->meta, h);
160 	t = print_otag(h, TAG_DIV, "c", "manual-text");
161 	print_man_nodelist(&man->meta, n, h);
162 	print_tagq(h, t);
163 	man_root_post(&man->meta, h);
164 	print_tagq(h, NULL);
165 }
166 
167 static void
168 print_man_head(const struct roff_meta *man, struct html *h)
169 {
170 	char	*cp;
171 
172 	print_gen_head(h);
173 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
174 	print_otag(h, TAG_TITLE, "");
175 	print_text(h, cp);
176 	free(cp);
177 }
178 
179 static void
180 print_man_nodelist(MAN_ARGS)
181 {
182 
183 	while (n != NULL) {
184 		print_man_node(man, n, h);
185 		n = n->next;
186 	}
187 }
188 
189 static void
190 print_man_node(MAN_ARGS)
191 {
192 	static int	 want_fillmode = MAN_fi;
193 	static int	 save_fillmode;
194 
195 	struct tag	*t;
196 	int		 child;
197 
198 	/*
199 	 * Handle fill mode switch requests up front,
200 	 * they would just cause trouble in the subsequent code.
201 	 */
202 
203 	switch (n->tok) {
204 	case MAN_nf:
205 	case MAN_EX:
206 		want_fillmode = MAN_nf;
207 		return;
208 	case MAN_fi:
209 	case MAN_EE:
210 		want_fillmode = MAN_fi;
211 		if (fillmode(h, 0) == MAN_fi)
212 			print_otag(h, TAG_BR, "");
213 		return;
214 	default:
215 		break;
216 	}
217 
218 	/* Set up fill mode for the upcoming node. */
219 
220 	switch (n->type) {
221 	case ROFFT_BLOCK:
222 		save_fillmode = 0;
223 		/* Some block macros suspend or cancel .nf. */
224 		switch (n->tok) {
225 		case MAN_TP:  /* Tagged paragraphs		*/
226 		case MAN_IP:  /* temporarily disable .nf	*/
227 		case MAN_HP:  /* for the head.			*/
228 			save_fillmode = want_fillmode;
229 			/* FALLTHROUGH */
230 		case MAN_SH:  /* Section headers		*/
231 		case MAN_SS:  /* permanently cancel .nf.	*/
232 			want_fillmode = MAN_fi;
233 			/* FALLTHROUGH */
234 		case MAN_PP:  /* These have no head.		*/
235 		case MAN_LP:  /* They will simply		*/
236 		case MAN_P:   /* reopen .nf in the body.	*/
237 		case MAN_RS:
238 		case MAN_UR:
239 		case MAN_MT:
240 			fillmode(h, MAN_fi);
241 			break;
242 		default:
243 			break;
244 		}
245 		break;
246 	case ROFFT_TBL:
247 		fillmode(h, MAN_fi);
248 		break;
249 	case ROFFT_ELEM:
250 		/*
251 		 * Some in-line macros produce tags and/or text
252 		 * in the handler, so they require fill mode to be
253 		 * configured up front just like for text nodes.
254 		 * For the others, keep the traditional approach
255 		 * of doing the same, for now.
256 		 */
257 		fillmode(h, want_fillmode);
258 		break;
259 	case ROFFT_TEXT:
260 		if (fillmode(h, want_fillmode) == MAN_fi &&
261 		    want_fillmode == MAN_fi &&
262 		    n->flags & NODE_LINE && *n->string == ' ' &&
263 		    (h->flags & HTML_NONEWLINE) == 0)
264 			print_otag(h, TAG_BR, "");
265 		if (*n->string != '\0')
266 			break;
267 		print_paragraph(h);
268 		return;
269 	case ROFFT_COMMENT:
270 		return;
271 	default:
272 		break;
273 	}
274 
275 	/* Produce output for this node. */
276 
277 	child = 1;
278 	switch (n->type) {
279 	case ROFFT_TEXT:
280 		t = h->tag;
281 		print_text(h, n->string);
282 		break;
283 	case ROFFT_EQN:
284 		t = h->tag;
285 		print_eqn(h, n->eqn);
286 		break;
287 	case ROFFT_TBL:
288 		/*
289 		 * This will take care of initialising all of the table
290 		 * state data for the first table, then tearing it down
291 		 * for the last one.
292 		 */
293 		print_tbl(h, n->span);
294 		return;
295 	default:
296 		/*
297 		 * Close out scope of font prior to opening a macro
298 		 * scope.
299 		 */
300 		if (HTMLFONT_NONE != h->metac) {
301 			h->metal = h->metac;
302 			h->metac = HTMLFONT_NONE;
303 		}
304 
305 		/*
306 		 * Close out the current table, if it's open, and unset
307 		 * the "meta" table state.  This will be reopened on the
308 		 * next table element.
309 		 */
310 		if (h->tblt)
311 			print_tblclose(h);
312 
313 		t = h->tag;
314 		if (n->tok < ROFF_MAX) {
315 			roff_html_pre(h, n);
316 			child = 0;
317 			break;
318 		}
319 
320 		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
321 		if (man_html_acts[n->tok - MAN_TH].pre != NULL)
322 			child = (*man_html_acts[n->tok - MAN_TH].pre)(man,
323 			    n, h);
324 
325 		/* Some block macros resume .nf in the body. */
326 		if (save_fillmode && n->type == ROFFT_BODY)
327 			want_fillmode = save_fillmode;
328 
329 		break;
330 	}
331 
332 	if (child && n->child)
333 		print_man_nodelist(man, n->child, h);
334 
335 	/* This will automatically close out any font scope. */
336 	print_stagq(h, t);
337 
338 	if (fillmode(h, 0) == MAN_nf &&
339 	    n->next != NULL && n->next->flags & NODE_LINE)
340 		print_endline(h);
341 }
342 
343 /*
344  * MAN_nf switches to no-fill mode, MAN_fi to fill mode.
345  * Other arguments do not switch.
346  * The old mode is returned.
347  */
348 static int
349 fillmode(struct html *h, int want)
350 {
351 	struct tag	*pre;
352 	int		 had;
353 
354 	for (pre = h->tag; pre != NULL; pre = pre->next)
355 		if (pre->tag == TAG_PRE)
356 			break;
357 
358 	had = pre == NULL ? MAN_fi : MAN_nf;
359 
360 	if (want && want != had) {
361 		if (want == MAN_nf)
362 			print_otag(h, TAG_PRE, "");
363 		else
364 			print_tagq(h, pre);
365 	}
366 	return had;
367 }
368 
369 static void
370 man_root_pre(const struct roff_meta *man, struct html *h)
371 {
372 	struct tag	*t, *tt;
373 	char		*title;
374 
375 	assert(man->title);
376 	assert(man->msec);
377 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
378 
379 	t = print_otag(h, TAG_TABLE, "c", "head");
380 	tt = print_otag(h, TAG_TR, "");
381 
382 	print_otag(h, TAG_TD, "c", "head-ltitle");
383 	print_text(h, title);
384 	print_stagq(h, tt);
385 
386 	print_otag(h, TAG_TD, "c", "head-vol");
387 	if (NULL != man->vol)
388 		print_text(h, man->vol);
389 	print_stagq(h, tt);
390 
391 	print_otag(h, TAG_TD, "c", "head-rtitle");
392 	print_text(h, title);
393 	print_tagq(h, t);
394 	free(title);
395 }
396 
397 static void
398 man_root_post(const struct roff_meta *man, struct html *h)
399 {
400 	struct tag	*t, *tt;
401 
402 	t = print_otag(h, TAG_TABLE, "c", "foot");
403 	tt = print_otag(h, TAG_TR, "");
404 
405 	print_otag(h, TAG_TD, "c", "foot-date");
406 	print_text(h, man->date);
407 	print_stagq(h, tt);
408 
409 	print_otag(h, TAG_TD, "c", "foot-os");
410 	if (man->os)
411 		print_text(h, man->os);
412 	print_tagq(h, t);
413 }
414 
415 static int
416 man_SH_pre(MAN_ARGS)
417 {
418 	char	*id;
419 
420 	if (n->type == ROFFT_HEAD) {
421 		id = html_make_id(n, 1);
422 		print_otag(h, TAG_H1, "cTi", "Sh", id);
423 		if (id != NULL)
424 			print_otag(h, TAG_A, "chR", "permalink", id);
425 	}
426 	return 1;
427 }
428 
429 static int
430 man_alt_pre(MAN_ARGS)
431 {
432 	const struct roff_node	*nn;
433 	int		 i;
434 	enum htmltag	 fp;
435 	struct tag	*t;
436 
437 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
438 		switch (n->tok) {
439 		case MAN_BI:
440 			fp = i % 2 ? TAG_I : TAG_B;
441 			break;
442 		case MAN_IB:
443 			fp = i % 2 ? TAG_B : TAG_I;
444 			break;
445 		case MAN_RI:
446 			fp = i % 2 ? TAG_I : TAG_MAX;
447 			break;
448 		case MAN_IR:
449 			fp = i % 2 ? TAG_MAX : TAG_I;
450 			break;
451 		case MAN_BR:
452 			fp = i % 2 ? TAG_MAX : TAG_B;
453 			break;
454 		case MAN_RB:
455 			fp = i % 2 ? TAG_B : TAG_MAX;
456 			break;
457 		default:
458 			abort();
459 		}
460 
461 		if (i)
462 			h->flags |= HTML_NOSPACE;
463 
464 		if (fp != TAG_MAX)
465 			t = print_otag(h, fp, "");
466 
467 		print_text(h, nn->string);
468 
469 		if (fp != TAG_MAX)
470 			print_tagq(h, t);
471 	}
472 	return 0;
473 }
474 
475 static int
476 man_SM_pre(MAN_ARGS)
477 {
478 	print_otag(h, TAG_SMALL, "");
479 	if (MAN_SB == n->tok)
480 		print_otag(h, TAG_B, "");
481 	return 1;
482 }
483 
484 static int
485 man_SS_pre(MAN_ARGS)
486 {
487 	char	*id;
488 
489 	if (n->type == ROFFT_HEAD) {
490 		id = html_make_id(n, 1);
491 		print_otag(h, TAG_H2, "cTi", "Ss", id);
492 		if (id != NULL)
493 			print_otag(h, TAG_A, "chR", "permalink", id);
494 	}
495 	return 1;
496 }
497 
498 static int
499 man_PP_pre(MAN_ARGS)
500 {
501 
502 	if (n->type == ROFFT_HEAD)
503 		return 0;
504 	else if (n->type == ROFFT_BLOCK)
505 		print_bvspace(h, n);
506 
507 	return 1;
508 }
509 
510 static int
511 man_IP_pre(MAN_ARGS)
512 {
513 	const struct roff_node	*nn;
514 
515 	if (n->type == ROFFT_BODY) {
516 		print_otag(h, TAG_DD, "");
517 		return 1;
518 	} else if (n->type != ROFFT_HEAD) {
519 		print_otag(h, TAG_DL, "c", "Bl-tag");
520 		return 1;
521 	}
522 
523 	print_otag(h, TAG_DT, "");
524 
525 	switch(n->tok) {
526 	case MAN_IP:  /* Only print the first header element. */
527 		if (n->child != NULL)
528 			print_man_node(man, n->child, h);
529 		break;
530 	case MAN_TP:  /* Only print next-line header elements. */
531 	case MAN_TQ:
532 		nn = n->child;
533 		while (nn != NULL && (NODE_LINE & nn->flags) == 0)
534 			nn = nn->next;
535 		while (nn != NULL) {
536 			print_man_node(man, nn, h);
537 			nn = nn->next;
538 		}
539 		break;
540 	default:
541 		abort();
542 	}
543 
544 	return 0;
545 }
546 
547 static int
548 man_HP_pre(MAN_ARGS)
549 {
550 	if (n->type == ROFFT_HEAD)
551 		return 0;
552 
553 	if (n->type == ROFFT_BLOCK) {
554 		print_bvspace(h, n);
555 		print_otag(h, TAG_DIV, "c", "HP");
556 	}
557 	return 1;
558 }
559 
560 static int
561 man_OP_pre(MAN_ARGS)
562 {
563 	struct tag	*tt;
564 
565 	print_text(h, "[");
566 	h->flags |= HTML_NOSPACE;
567 	tt = print_otag(h, TAG_SPAN, "c", "Op");
568 
569 	if (NULL != (n = n->child)) {
570 		print_otag(h, TAG_B, "");
571 		print_text(h, n->string);
572 	}
573 
574 	print_stagq(h, tt);
575 
576 	if (NULL != n && NULL != n->next) {
577 		print_otag(h, TAG_I, "");
578 		print_text(h, n->next->string);
579 	}
580 
581 	print_stagq(h, tt);
582 	h->flags |= HTML_NOSPACE;
583 	print_text(h, "]");
584 	return 0;
585 }
586 
587 static int
588 man_B_pre(MAN_ARGS)
589 {
590 	print_otag(h, TAG_B, "");
591 	return 1;
592 }
593 
594 static int
595 man_I_pre(MAN_ARGS)
596 {
597 	print_otag(h, TAG_I, "");
598 	return 1;
599 }
600 
601 static int
602 man_in_pre(MAN_ARGS)
603 {
604 	print_otag(h, TAG_BR, "");
605 	return 0;
606 }
607 
608 static int
609 man_ign_pre(MAN_ARGS)
610 {
611 
612 	return 0;
613 }
614 
615 static int
616 man_RS_pre(MAN_ARGS)
617 {
618 	if (n->type == ROFFT_HEAD)
619 		return 0;
620 	if (n->type == ROFFT_BLOCK)
621 		print_otag(h, TAG_DIV, "c", "Bd-indent");
622 	return 1;
623 }
624 
625 static int
626 man_SY_pre(MAN_ARGS)
627 {
628 	switch (n->type) {
629 	case ROFFT_BLOCK:
630 		print_otag(h, TAG_TABLE, "c", "Nm");
631 		print_otag(h, TAG_TR, "");
632 		break;
633 	case ROFFT_HEAD:
634 		print_otag(h, TAG_TD, "");
635 		print_otag(h, TAG_CODE, "cT", "Nm");
636 		break;
637 	case ROFFT_BODY:
638 		print_otag(h, TAG_TD, "");
639 		break;
640 	default:
641 		abort();
642 	}
643 	return 1;
644 }
645 
646 static int
647 man_UR_pre(MAN_ARGS)
648 {
649 	char *cp;
650 	n = n->child;
651 	assert(n->type == ROFFT_HEAD);
652 	if (n->child != NULL) {
653 		assert(n->child->type == ROFFT_TEXT);
654 		if (n->tok == MAN_MT) {
655 			mandoc_asprintf(&cp, "mailto:%s", n->child->string);
656 			print_otag(h, TAG_A, "cTh", "Mt", cp);
657 			free(cp);
658 		} else
659 			print_otag(h, TAG_A, "cTh", "Lk", n->child->string);
660 	}
661 
662 	assert(n->next->type == ROFFT_BODY);
663 	if (n->next->child != NULL)
664 		n = n->next;
665 
666 	print_man_nodelist(man, n->child, h);
667 
668 	return 0;
669 }
670