xref: /openbsd/usr.bin/mandoc/man_html.c (revision 8f48bc46) 
1 /* $OpenBSD: man_html.c,v 1.135 2022/07/05 21:25:23 schwarze Exp $ */
2 /*
3  * Copyright (c) 2013-2015,2017-2020,2022 Ingo Schwarze <schwarze@openbsd.org>
4  * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
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  * HTML formatter for man(7) used by mandoc(1).
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 "mandoc.h"
30 #include "roff.h"
31 #include "man.h"
32 #include "out.h"
33 #include "html.h"
34 #include "main.h"
35 
36 #define	MAN_ARGS	  const struct roff_meta *man, \
37 			  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_man_head(const struct roff_meta *,
46 				struct html *);
47 static	void		  print_man_nodelist(MAN_ARGS);
48 static	void		  print_man_node(MAN_ARGS);
49 static	char		  list_continues(const struct roff_node *,
50 				const struct roff_node *);
51 static	int		  man_B_pre(MAN_ARGS);
52 static	int		  man_IP_pre(MAN_ARGS);
53 static	int		  man_I_pre(MAN_ARGS);
54 static	int		  man_OP_pre(MAN_ARGS);
55 static	int		  man_PP_pre(MAN_ARGS);
56 static	int		  man_RS_pre(MAN_ARGS);
57 static	int		  man_SH_pre(MAN_ARGS);
58 static	int		  man_SM_pre(MAN_ARGS);
59 static	int		  man_SY_pre(MAN_ARGS);
60 static	int		  man_UR_pre(MAN_ARGS);
61 static	int		  man_abort_pre(MAN_ARGS);
62 static	int		  man_alt_pre(MAN_ARGS);
63 static	int		  man_ign_pre(MAN_ARGS);
64 static	int		  man_in_pre(MAN_ARGS);
65 static	void		  man_root_post(const struct roff_meta *,
66 				struct html *);
67 static	void		  man_root_pre(const struct roff_meta *,
68 				struct html *);
69 
70 static	const struct man_html_act man_html_acts[MAN_MAX - MAN_TH] = {
71 	{ NULL, NULL }, /* TH */
72 	{ man_SH_pre, NULL }, /* SH */
73 	{ man_SH_pre, NULL }, /* SS */
74 	{ man_IP_pre, NULL }, /* TP */
75 	{ man_IP_pre, NULL }, /* TQ */
76 	{ man_abort_pre, NULL }, /* LP */
77 	{ man_PP_pre, NULL }, /* PP */
78 	{ man_abort_pre, NULL }, /* P */
79 	{ man_IP_pre, NULL }, /* IP */
80 	{ man_PP_pre, NULL }, /* HP */
81 	{ man_SM_pre, NULL }, /* SM */
82 	{ man_SM_pre, NULL }, /* SB */
83 	{ man_alt_pre, NULL }, /* BI */
84 	{ man_alt_pre, NULL }, /* IB */
85 	{ man_alt_pre, NULL }, /* BR */
86 	{ man_alt_pre, NULL }, /* RB */
87 	{ NULL, NULL }, /* R */
88 	{ man_B_pre, NULL }, /* B */
89 	{ man_I_pre, NULL }, /* I */
90 	{ man_alt_pre, NULL }, /* IR */
91 	{ man_alt_pre, NULL }, /* RI */
92 	{ NULL, NULL }, /* RE */
93 	{ man_RS_pre, NULL }, /* RS */
94 	{ man_ign_pre, NULL }, /* DT */
95 	{ man_ign_pre, NULL }, /* UC */
96 	{ man_ign_pre, NULL }, /* PD */
97 	{ man_ign_pre, NULL }, /* AT */
98 	{ man_in_pre, NULL }, /* in */
99 	{ man_SY_pre, NULL }, /* SY */
100 	{ NULL, NULL }, /* YS */
101 	{ man_OP_pre, NULL }, /* OP */
102 	{ NULL, NULL }, /* EX */
103 	{ NULL, NULL }, /* EE */
104 	{ man_UR_pre, NULL }, /* UR */
105 	{ NULL, NULL }, /* UE */
106 	{ man_UR_pre, NULL }, /* MT */
107 	{ NULL, NULL }, /* ME */
108 };
109 
110 
111 void
112 html_man(void *arg, const struct roff_meta *man)
113 {
114 	struct html		*h;
115 	struct roff_node	*n;
116 	struct tag		*t;
117 
118 	h = (struct html *)arg;
119 	n = man->first->child;
120 
121 	if ((h->oflags & HTML_FRAGMENT) == 0) {
122 		print_gen_decls(h);
123 		print_otag(h, TAG_HTML, "");
124 		t = print_otag(h, TAG_HEAD, "");
125 		print_man_head(man, h);
126 		print_tagq(h, t);
127 		if (n != NULL && n->type == ROFFT_COMMENT)
128 			print_gen_comment(h, n);
129 		print_otag(h, TAG_BODY, "");
130 	}
131 
132 	man_root_pre(man, h);
133 	t = print_otag(h, TAG_MAIN, "c", "manual-text");
134 	print_man_nodelist(man, n, h);
135 	print_tagq(h, t);
136 	man_root_post(man, h);
137 	print_tagq(h, NULL);
138 }
139 
140 static void
141 print_man_head(const struct roff_meta *man, struct html *h)
142 {
143 	char	*cp;
144 
145 	print_gen_head(h);
146 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
147 	print_otag(h, TAG_TITLE, "");
148 	print_text(h, cp);
149 	free(cp);
150 }
151 
152 static void
153 print_man_nodelist(MAN_ARGS)
154 {
155 	while (n != NULL) {
156 		print_man_node(man, n, h);
157 		n = n->next;
158 	}
159 }
160 
161 static void
162 print_man_node(MAN_ARGS)
163 {
164 	struct tag	*t;
165 	int		 child;
166 
167 	if (n->type == ROFFT_COMMENT || n->flags & NODE_NOPRT)
168 		return;
169 
170 	if ((n->flags & NODE_NOFILL) == 0)
171 		html_fillmode(h, ROFF_fi);
172 	else if (html_fillmode(h, ROFF_nf) == ROFF_nf &&
173 	    n->tok != ROFF_fi && n->flags & NODE_LINE &&
174 	    (n->prev == NULL || n->prev->tok != MAN_YS))
175 		print_endline(h);
176 
177 	child = 1;
178 	switch (n->type) {
179 	case ROFFT_TEXT:
180 		if (*n->string == '\0') {
181 			print_endline(h);
182 			return;
183 		}
184 		if (*n->string == ' ' && n->flags & NODE_LINE &&
185 		    (h->flags & HTML_NONEWLINE) == 0)
186 			print_otag(h, TAG_BR, "");
187 		else if (n->flags & NODE_DELIMC)
188 			h->flags |= HTML_NOSPACE;
189 		t = h->tag;
190 		t->refcnt++;
191 		print_text(h, n->string);
192 		break;
193 	case ROFFT_EQN:
194 		t = h->tag;
195 		t->refcnt++;
196 		print_eqn(h, n->eqn);
197 		break;
198 	case ROFFT_TBL:
199 		/*
200 		 * This will take care of initialising all of the table
201 		 * state data for the first table, then tearing it down
202 		 * for the last one.
203 		 */
204 		print_tbl(h, n->span);
205 		return;
206 	default:
207 		/*
208 		 * Close out scope of font prior to opening a macro
209 		 * scope.
210 		 */
211 		if (h->metac != ESCAPE_FONTROMAN) {
212 			h->metal = h->metac;
213 			h->metac = ESCAPE_FONTROMAN;
214 		}
215 
216 		/*
217 		 * Close out the current table, if it's open, and unset
218 		 * the "meta" table state.  This will be reopened on the
219 		 * next table element.
220 		 */
221 		if (h->tblt != NULL)
222 			print_tblclose(h);
223 		t = h->tag;
224 		t->refcnt++;
225 		if (n->tok < ROFF_MAX) {
226 			roff_html_pre(h, n);
227 			t->refcnt--;
228 			print_stagq(h, t);
229 			return;
230 		}
231 		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
232 		if (man_html_acts[n->tok - MAN_TH].pre != NULL)
233 			child = (*man_html_acts[n->tok - MAN_TH].pre)(man,
234 			    n, h);
235 		break;
236 	}
237 
238 	if (child && n->child != NULL)
239 		print_man_nodelist(man, n->child, h);
240 
241 	/* This will automatically close out any font scope. */
242 	t->refcnt--;
243 	if (n->type == ROFFT_BLOCK &&
244 	    (n->tok == MAN_IP || n->tok == MAN_TP || n->tok == MAN_TQ)) {
245 		t = h->tag;
246 		while (t->tag != TAG_DL && t->tag != TAG_UL)
247 			t = t->next;
248 		/*
249 		 * Close the list if no further item of the same type
250 		 * follows; otherwise, close the item only.
251 		 */
252 		if (list_continues(n, roff_node_next(n)) == '\0') {
253 			print_tagq(h, t);
254 			t = NULL;
255 		}
256 	}
257 	if (t != NULL)
258 		print_stagq(h, t);
259 }
260 
261 static void
262 man_root_pre(const struct roff_meta *man, struct html *h)
263 {
264 	struct tag	*t;
265 	char		*title;
266 
267 	assert(man->title);
268 	assert(man->msec);
269 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
270 
271 	t = print_otag(h, TAG_DIV, "cr?", "head", "doc-pageheader",
272 	    "aria-label", "manual header line");
273 
274 	print_otag(h, TAG_SPAN, "c", "head-ltitle");
275 	print_text(h, title);
276 	print_stagq(h, t);
277 
278 	print_otag(h, TAG_SPAN, "c", "head-vol");
279 	if (man->vol != NULL)
280 		print_text(h, man->vol);
281 	print_stagq(h, t);
282 
283 	print_otag(h, TAG_SPAN, "c", "head-rtitle");
284 	print_text(h, title);
285 	print_tagq(h, t);
286 	free(title);
287 }
288 
289 static void
290 man_root_post(const struct roff_meta *man, struct html *h)
291 {
292 	struct tag	*t;
293 
294 	t = print_otag(h, TAG_DIV, "cr?", "foot", "doc-pagefooter",
295 	    "aria-label", "manual footer line");
296 
297 	print_otag(h, TAG_SPAN, "c", "foot-left");
298 	print_stagq(h, t);
299 
300 	print_otag(h, TAG_SPAN, "c", "foot-date");
301 	print_text(h, man->date);
302 	print_stagq(h, t);
303 
304 	print_otag(h, TAG_SPAN, "c", "foot-os");
305 	if (man->os != NULL)
306 		print_text(h, man->os);
307 	print_tagq(h, t);
308 }
309 
310 static int
311 man_SH_pre(MAN_ARGS)
312 {
313 	const char	*class;
314 	enum htmltag	 tag;
315 
316 	if (n->tok == MAN_SH) {
317 		tag = TAG_H1;
318 		class = "Sh";
319 	} else {
320 		tag = TAG_H2;
321 		class = "Ss";
322 	}
323 	switch (n->type) {
324 	case ROFFT_BLOCK:
325 		html_close_paragraph(h);
326 		print_otag(h, TAG_SECTION, "c", class);
327 		break;
328 	case ROFFT_HEAD:
329 		print_otag_id(h, tag, class, n);
330 		break;
331 	case ROFFT_BODY:
332 		break;
333 	default:
334 		abort();
335 	}
336 	return 1;
337 }
338 
339 static int
340 man_alt_pre(MAN_ARGS)
341 {
342 	const struct roff_node	*nn;
343 	struct tag	*t;
344 	int		 i;
345 	enum htmltag	 fp;
346 
347 	for (i = 0, nn = n->child; nn != NULL; nn = nn->next, i++) {
348 		switch (n->tok) {
349 		case MAN_BI:
350 			fp = i % 2 ? TAG_I : TAG_B;
351 			break;
352 		case MAN_IB:
353 			fp = i % 2 ? TAG_B : TAG_I;
354 			break;
355 		case MAN_RI:
356 			fp = i % 2 ? TAG_I : TAG_MAX;
357 			break;
358 		case MAN_IR:
359 			fp = i % 2 ? TAG_MAX : TAG_I;
360 			break;
361 		case MAN_BR:
362 			fp = i % 2 ? TAG_MAX : TAG_B;
363 			break;
364 		case MAN_RB:
365 			fp = i % 2 ? TAG_B : TAG_MAX;
366 			break;
367 		default:
368 			abort();
369 		}
370 
371 		if (i)
372 			h->flags |= HTML_NOSPACE;
373 
374 		if (fp != TAG_MAX)
375 			t = print_otag(h, fp, "");
376 
377 		print_text(h, nn->string);
378 
379 		if (fp != TAG_MAX)
380 			print_tagq(h, t);
381 	}
382 	return 0;
383 }
384 
385 static int
386 man_SM_pre(MAN_ARGS)
387 {
388 	print_otag(h, TAG_SMALL, "");
389 	if (n->tok == MAN_SB)
390 		print_otag(h, TAG_B, "");
391 	return 1;
392 }
393 
394 static int
395 man_PP_pre(MAN_ARGS)
396 {
397 	switch (n->type) {
398 	case ROFFT_BLOCK:
399 		html_close_paragraph(h);
400 		break;
401 	case ROFFT_HEAD:
402 		return 0;
403 	case ROFFT_BODY:
404 		if (n->child != NULL &&
405 		    (n->child->flags & NODE_NOFILL) == 0)
406 			print_otag(h, TAG_P, "c",
407 			    n->tok == MAN_PP ? "Pp" : "Pp HP");
408 		break;
409 	default:
410 		abort();
411 	}
412 	return 1;
413 }
414 
415 static char
416 list_continues(const struct roff_node *n1, const struct roff_node *n2)
417 {
418 	const char *s1, *s2;
419 	char c1, c2;
420 
421 	if (n1 == NULL || n1->type != ROFFT_BLOCK ||
422 	    n2 == NULL || n2->type != ROFFT_BLOCK)
423 		return '\0';
424 	if ((n1->tok == MAN_TP || n1->tok == MAN_TQ) &&
425 	    (n2->tok == MAN_TP || n2->tok == MAN_TQ))
426 		return ' ';
427 	if (n1->tok != MAN_IP || n2->tok != MAN_IP)
428 		return '\0';
429 	n1 = n1->head->child;
430 	n2 = n2->head->child;
431 	s1 = n1 == NULL ? "" : n1->string;
432 	s2 = n2 == NULL ? "" : n2->string;
433 	c1 = strcmp(s1, "*") == 0 ? '*' :
434 	     strcmp(s1, "\\-") == 0 ? '-' :
435 	     strcmp(s1, "\\(bu") == 0 ? 'b' : ' ';
436 	c2 = strcmp(s2, "*") == 0 ? '*' :
437 	     strcmp(s2, "\\-") == 0 ? '-' :
438 	     strcmp(s2, "\\(bu") == 0 ? 'b' : ' ';
439 	return c1 != c2 ? '\0' : c1 == 'b' ? '*' : c1;
440 }
441 
442 static int
443 man_IP_pre(MAN_ARGS)
444 {
445 	struct roff_node	*nn;
446 	const char		*list_class;
447 	enum htmltag		 list_elem, body_elem;
448 	char			 list_type;
449 
450 	nn = n->type == ROFFT_BLOCK ? n : n->parent;
451 	list_type = list_continues(roff_node_prev(nn), nn);
452 	if (list_type == '\0') {
453 		/* Start a new list. */
454 		list_type = list_continues(nn, roff_node_next(nn));
455 		if (list_type == '\0')
456 			list_type = ' ';
457 		switch (list_type) {
458 		case ' ':
459 			list_class = "Bl-tag";
460 			list_elem = TAG_DL;
461 			break;
462 		case '*':
463 			list_class = "Bl-bullet";
464 			list_elem = TAG_UL;
465 			break;
466 		case '-':
467 			list_class = "Bl-dash";
468 			list_elem = TAG_UL;
469 			break;
470 		default:
471 			abort();
472 		}
473 	} else {
474 		/* Continue a list that was started earlier. */
475 		list_class = NULL;
476 		list_elem = TAG_MAX;
477 	}
478 	body_elem = list_type == ' ' ? TAG_DD : TAG_LI;
479 
480 	switch (n->type) {
481 	case ROFFT_BLOCK:
482 		html_close_paragraph(h);
483 		if (list_elem != TAG_MAX)
484 			print_otag(h, list_elem, "c", list_class);
485 		return 1;
486 	case ROFFT_HEAD:
487 		if (body_elem == TAG_LI)
488 			return 0;
489 		print_otag_id(h, TAG_DT, NULL, n);
490 		break;
491 	case ROFFT_BODY:
492 		print_otag(h, body_elem, "");
493 		return 1;
494 	default:
495 		abort();
496 	}
497 	switch(n->tok) {
498 	case MAN_IP:  /* Only print the first header element. */
499 		if (n->child != NULL)
500 			print_man_node(man, n->child, h);
501 		break;
502 	case MAN_TP:  /* Only print next-line header elements. */
503 	case MAN_TQ:
504 		nn = n->child;
505 		while (nn != NULL && (NODE_LINE & nn->flags) == 0)
506 			nn = nn->next;
507 		while (nn != NULL) {
508 			print_man_node(man, nn, h);
509 			nn = nn->next;
510 		}
511 		break;
512 	default:
513 		abort();
514 	}
515 	return 0;
516 }
517 
518 static int
519 man_OP_pre(MAN_ARGS)
520 {
521 	struct tag	*tt;
522 
523 	print_text(h, "[");
524 	h->flags |= HTML_NOSPACE;
525 	tt = print_otag(h, TAG_SPAN, "c", "Op");
526 
527 	if ((n = n->child) != NULL) {
528 		print_otag(h, TAG_B, "");
529 		print_text(h, n->string);
530 	}
531 
532 	print_stagq(h, tt);
533 
534 	if (n != NULL && n->next != NULL) {
535 		print_otag(h, TAG_I, "");
536 		print_text(h, n->next->string);
537 	}
538 
539 	print_stagq(h, tt);
540 	h->flags |= HTML_NOSPACE;
541 	print_text(h, "]");
542 	return 0;
543 }
544 
545 static int
546 man_B_pre(MAN_ARGS)
547 {
548 	print_otag(h, TAG_B, "");
549 	return 1;
550 }
551 
552 static int
553 man_I_pre(MAN_ARGS)
554 {
555 	print_otag(h, TAG_I, "");
556 	return 1;
557 }
558 
559 static int
560 man_in_pre(MAN_ARGS)
561 {
562 	print_otag(h, TAG_BR, "");
563 	return 0;
564 }
565 
566 static int
567 man_ign_pre(MAN_ARGS)
568 {
569 	return 0;
570 }
571 
572 static int
573 man_RS_pre(MAN_ARGS)
574 {
575 	switch (n->type) {
576 	case ROFFT_BLOCK:
577 		html_close_paragraph(h);
578 		break;
579 	case ROFFT_HEAD:
580 		return 0;
581 	case ROFFT_BODY:
582 		print_otag(h, TAG_DIV, "c", "Bd-indent");
583 		break;
584 	default:
585 		abort();
586 	}
587 	return 1;
588 }
589 
590 static int
591 man_SY_pre(MAN_ARGS)
592 {
593 	switch (n->type) {
594 	case ROFFT_BLOCK:
595 		html_close_paragraph(h);
596 		print_otag(h, TAG_TABLE, "c", "Nm");
597 		print_otag(h, TAG_TR, "");
598 		break;
599 	case ROFFT_HEAD:
600 		print_otag(h, TAG_TD, "");
601 		print_otag(h, TAG_CODE, "c", "Nm");
602 		break;
603 	case ROFFT_BODY:
604 		print_otag(h, TAG_TD, "");
605 		break;
606 	default:
607 		abort();
608 	}
609 	return 1;
610 }
611 
612 static int
613 man_UR_pre(MAN_ARGS)
614 {
615 	char *cp;
616 
617 	n = n->child;
618 	assert(n->type == ROFFT_HEAD);
619 	if (n->child != NULL) {
620 		assert(n->child->type == ROFFT_TEXT);
621 		if (n->tok == MAN_MT) {
622 			mandoc_asprintf(&cp, "mailto:%s", n->child->string);
623 			print_otag(h, TAG_A, "ch", "Mt", cp);
624 			free(cp);
625 		} else
626 			print_otag(h, TAG_A, "ch", "Lk", n->child->string);
627 	}
628 
629 	assert(n->next->type == ROFFT_BODY);
630 	if (n->next->child != NULL)
631 		n = n->next;
632 
633 	print_man_nodelist(man, n->child, h);
634 	return 0;
635 }
636 
637 static int
638 man_abort_pre(MAN_ARGS)
639 {
640 	abort();
641 }
642