xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision ae3cb403620ab940fbaabb3055fac045a63d56b7)
1 /*	$OpenBSD: man_html.c,v 1.98 2017/06/25 07:23:53 bentley 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 "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	INDENT		  5
37 
38 #define	MAN_ARGS	  const struct roff_meta *man, \
39 			  const struct roff_node *n, \
40 			  struct html *h
41 
42 struct	htmlman {
43 	int		(*pre)(MAN_ARGS);
44 	int		(*post)(MAN_ARGS);
45 };
46 
47 static	void		  print_bvspace(struct html *,
48 				const struct roff_node *);
49 static	void		  print_man_head(MAN_ARGS);
50 static	void		  print_man_nodelist(MAN_ARGS);
51 static	void		  print_man_node(MAN_ARGS);
52 static	int		  fillmode(struct html *, int);
53 static	int		  a2width(const struct roff_node *,
54 				struct roffsu *);
55 static	int		  man_B_pre(MAN_ARGS);
56 static	int		  man_HP_pre(MAN_ARGS);
57 static	int		  man_IP_pre(MAN_ARGS);
58 static	int		  man_I_pre(MAN_ARGS);
59 static	int		  man_OP_pre(MAN_ARGS);
60 static	int		  man_PP_pre(MAN_ARGS);
61 static	int		  man_RS_pre(MAN_ARGS);
62 static	int		  man_SH_pre(MAN_ARGS);
63 static	int		  man_SM_pre(MAN_ARGS);
64 static	int		  man_SS_pre(MAN_ARGS);
65 static	int		  man_UR_pre(MAN_ARGS);
66 static	int		  man_alt_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 - 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_PP_pre, NULL }, /* LP */
78 	{ man_PP_pre, NULL }, /* PP */
79 	{ man_PP_pre, NULL }, /* P */
80 	{ man_IP_pre, NULL }, /* IP */
81 	{ man_HP_pre, NULL }, /* HP */
82 	{ man_SM_pre, NULL }, /* SM */
83 	{ man_SM_pre, NULL }, /* SB */
84 	{ man_alt_pre, NULL }, /* BI */
85 	{ man_alt_pre, NULL }, /* IB */
86 	{ man_alt_pre, NULL }, /* BR */
87 	{ man_alt_pre, NULL }, /* RB */
88 	{ NULL, NULL }, /* R */
89 	{ man_B_pre, NULL }, /* B */
90 	{ man_I_pre, NULL }, /* I */
91 	{ man_alt_pre, NULL }, /* IR */
92 	{ man_alt_pre, NULL }, /* RI */
93 	{ NULL, NULL }, /* nf */
94 	{ NULL, NULL }, /* fi */
95 	{ NULL, NULL }, /* RE */
96 	{ man_RS_pre, NULL }, /* RS */
97 	{ man_ign_pre, NULL }, /* DT */
98 	{ man_ign_pre, NULL }, /* UC */
99 	{ man_ign_pre, NULL }, /* PD */
100 	{ man_ign_pre, NULL }, /* AT */
101 	{ man_in_pre, NULL }, /* in */
102 	{ man_OP_pre, NULL }, /* OP */
103 	{ NULL, NULL }, /* EX */
104 	{ NULL, NULL }, /* EE */
105 	{ man_UR_pre, NULL }, /* UR */
106 	{ NULL, NULL }, /* UE */
107 	{ man_UR_pre, NULL }, /* MT */
108 	{ NULL, NULL }, /* ME */
109 };
110 static	const struct htmlman *const mans = __mans - MAN_TH;
111 
112 
113 /*
114  * Printing leading vertical space before a block.
115  * This is used for the paragraph macros.
116  * The rules are pretty simple, since there's very little nesting going
117  * on here.  Basically, if we're the first within another block (SS/SH),
118  * then don't emit vertical space.  If we are (RS), then do.  If not the
119  * first, print it.
120  */
121 static void
122 print_bvspace(struct html *h, const struct roff_node *n)
123 {
124 
125 	if (n->body && n->body->child)
126 		if (n->body->child->type == ROFFT_TBL)
127 			return;
128 
129 	if (n->parent->type == ROFFT_ROOT || n->parent->tok != MAN_RS)
130 		if (NULL == n->prev)
131 			return;
132 
133 	print_paragraph(h);
134 }
135 
136 void
137 html_man(void *arg, const struct roff_man *man)
138 {
139 	struct html	*h;
140 	struct tag	*t;
141 
142 	h = (struct html *)arg;
143 
144 	if ((h->oflags & HTML_FRAGMENT) == 0) {
145 		print_gen_decls(h);
146 		print_otag(h, TAG_HTML, "");
147 		t = print_otag(h, TAG_HEAD, "");
148 		print_man_head(&man->meta, man->first, h);
149 		print_tagq(h, t);
150 		print_otag(h, TAG_BODY, "");
151 	}
152 
153 	man_root_pre(&man->meta, man->first, h);
154 	t = print_otag(h, TAG_DIV, "c", "manual-text");
155 	print_man_nodelist(&man->meta, man->first->child, h);
156 	print_tagq(h, t);
157 	man_root_post(&man->meta, man->first, h);
158 	print_tagq(h, NULL);
159 }
160 
161 static void
162 print_man_head(MAN_ARGS)
163 {
164 	char	*cp;
165 
166 	print_gen_head(h);
167 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
168 	print_otag(h, TAG_TITLE, "");
169 	print_text(h, cp);
170 	free(cp);
171 }
172 
173 static void
174 print_man_nodelist(MAN_ARGS)
175 {
176 
177 	while (n != NULL) {
178 		print_man_node(man, n, h);
179 		n = n->next;
180 	}
181 }
182 
183 static void
184 print_man_node(MAN_ARGS)
185 {
186 	static int	 want_fillmode = MAN_fi;
187 	static int	 save_fillmode;
188 
189 	struct tag	*t;
190 	int		 child;
191 
192 	/*
193 	 * Handle fill mode switch requests up front,
194 	 * they would just cause trouble in the subsequent code.
195 	 */
196 
197 	switch (n->tok) {
198 	case MAN_nf:
199 	case MAN_EX:
200 		want_fillmode = MAN_nf;
201 		return;
202 	case MAN_fi:
203 	case MAN_EE:
204 		want_fillmode = MAN_fi;
205 		if (fillmode(h, 0) == MAN_fi)
206 			print_otag(h, TAG_BR, "");
207 		return;
208 	default:
209 		break;
210 	}
211 
212 	/* Set up fill mode for the upcoming node. */
213 
214 	switch (n->type) {
215 	case ROFFT_BLOCK:
216 		save_fillmode = 0;
217 		/* Some block macros suspend or cancel .nf. */
218 		switch (n->tok) {
219 		case MAN_TP:  /* Tagged paragraphs		*/
220 		case MAN_IP:  /* temporarily disable .nf	*/
221 		case MAN_HP:  /* for the head.			*/
222 			save_fillmode = want_fillmode;
223 			/* FALLTHROUGH */
224 		case MAN_SH:  /* Section headers		*/
225 		case MAN_SS:  /* permanently cancel .nf.	*/
226 			want_fillmode = MAN_fi;
227 			/* FALLTHROUGH */
228 		case MAN_PP:  /* These have no head.		*/
229 		case MAN_LP:  /* They will simply		*/
230 		case MAN_P:   /* reopen .nf in the body.	*/
231 		case MAN_RS:
232 		case MAN_UR:
233 		case MAN_MT:
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 		    (h->flags & HTML_NONEWLINE) == 0)
258 			print_otag(h, TAG_BR, "");
259 		if (*n->string != '\0')
260 			break;
261 		print_paragraph(h);
262 		return;
263 	default:
264 		break;
265 	}
266 
267 	/* Produce output for this node. */
268 
269 	child = 1;
270 	switch (n->type) {
271 	case ROFFT_TEXT:
272 		t = h->tag;
273 		print_text(h, n->string);
274 		break;
275 	case ROFFT_EQN:
276 		t = h->tag;
277 		print_eqn(h, n->eqn);
278 		break;
279 	case ROFFT_TBL:
280 		/*
281 		 * This will take care of initialising all of the table
282 		 * state data for the first table, then tearing it down
283 		 * for the last one.
284 		 */
285 		print_tbl(h, n->span);
286 		return;
287 	default:
288 		/*
289 		 * Close out scope of font prior to opening a macro
290 		 * scope.
291 		 */
292 		if (HTMLFONT_NONE != h->metac) {
293 			h->metal = h->metac;
294 			h->metac = HTMLFONT_NONE;
295 		}
296 
297 		/*
298 		 * Close out the current table, if it's open, and unset
299 		 * the "meta" table state.  This will be reopened on the
300 		 * next table element.
301 		 */
302 		if (h->tblt)
303 			print_tblclose(h);
304 
305 		t = h->tag;
306 		if (n->tok < ROFF_MAX) {
307 			roff_html_pre(h, n);
308 			child = 0;
309 			break;
310 		}
311 
312 		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
313 		if (mans[n->tok].pre)
314 			child = (*mans[n->tok].pre)(man, n, h);
315 
316 		/* Some block macros resume .nf in the body. */
317 		if (save_fillmode && n->type == ROFFT_BODY)
318 			want_fillmode = save_fillmode;
319 
320 		break;
321 	}
322 
323 	if (child && n->child)
324 		print_man_nodelist(man, n->child, h);
325 
326 	/* This will automatically close out any font scope. */
327 	print_stagq(h, t);
328 
329 	if (fillmode(h, 0) == MAN_nf &&
330 	    n->next != NULL && n->next->flags & NODE_LINE)
331 		print_endline(h);
332 }
333 
334 /*
335  * MAN_nf switches to no-fill mode, MAN_fi to fill mode.
336  * Other arguments do not switch.
337  * The old mode is returned.
338  */
339 static int
340 fillmode(struct html *h, int want)
341 {
342 	struct tag	*pre;
343 	int		 had;
344 
345 	for (pre = h->tag; pre != NULL; pre = pre->next)
346 		if (pre->tag == TAG_PRE)
347 			break;
348 
349 	had = pre == NULL ? MAN_fi : MAN_nf;
350 
351 	if (want && want != had) {
352 		if (want == MAN_nf)
353 			print_otag(h, TAG_PRE, "");
354 		else
355 			print_tagq(h, pre);
356 	}
357 	return had;
358 }
359 
360 static int
361 a2width(const struct roff_node *n, struct roffsu *su)
362 {
363 	if (n->type != ROFFT_TEXT)
364 		return 0;
365 	return a2roffsu(n->string, su, SCALE_EN) != NULL;
366 }
367 
368 static void
369 man_root_pre(MAN_ARGS)
370 {
371 	struct tag	*t, *tt;
372 	char		*title;
373 
374 	assert(man->title);
375 	assert(man->msec);
376 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
377 
378 	t = print_otag(h, TAG_TABLE, "c", "head");
379 	tt = print_otag(h, TAG_TR, "");
380 
381 	print_otag(h, TAG_TD, "c", "head-ltitle");
382 	print_text(h, title);
383 	print_stagq(h, tt);
384 
385 	print_otag(h, TAG_TD, "c", "head-vol");
386 	if (NULL != man->vol)
387 		print_text(h, man->vol);
388 	print_stagq(h, tt);
389 
390 	print_otag(h, TAG_TD, "c", "head-rtitle");
391 	print_text(h, title);
392 	print_tagq(h, t);
393 	free(title);
394 }
395 
396 static void
397 man_root_post(MAN_ARGS)
398 {
399 	struct tag	*t, *tt;
400 
401 	t = print_otag(h, TAG_TABLE, "c", "foot");
402 	tt = print_otag(h, TAG_TR, "");
403 
404 	print_otag(h, TAG_TD, "c", "foot-date");
405 	print_text(h, man->date);
406 	print_stagq(h, tt);
407 
408 	print_otag(h, TAG_TD, "c", "foot-os");
409 	if (man->os)
410 		print_text(h, man->os);
411 	print_tagq(h, t);
412 }
413 
414 static int
415 man_SH_pre(MAN_ARGS)
416 {
417 	char	*id;
418 
419 	if (n->type == ROFFT_HEAD) {
420 		id = html_make_id(n);
421 		print_otag(h, TAG_H1, "cTi", "Sh", id);
422 		if (id != NULL)
423 			print_otag(h, TAG_A, "chR", "selflink", id);
424 		free(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);
491 		print_otag(h, TAG_H2, "cTi", "Ss", id);
492 		if (id != NULL)
493 			print_otag(h, TAG_A, "chR", "selflink", id);
494 		free(id);
495 	}
496 	return 1;
497 }
498 
499 static int
500 man_PP_pre(MAN_ARGS)
501 {
502 
503 	if (n->type == ROFFT_HEAD)
504 		return 0;
505 	else if (n->type == ROFFT_BLOCK)
506 		print_bvspace(h, n);
507 
508 	return 1;
509 }
510 
511 static int
512 man_IP_pre(MAN_ARGS)
513 {
514 	const struct roff_node	*nn;
515 
516 	if (n->type == ROFFT_BODY) {
517 		print_otag(h, TAG_DD, "c", "It-tag");
518 		return 1;
519 	} else if (n->type != ROFFT_HEAD) {
520 		print_otag(h, TAG_DL, "c", "Bl-tag");
521 		return 1;
522 	}
523 
524 	/* FIXME: width specification. */
525 
526 	print_otag(h, TAG_DT, "c", "It-tag");
527 
528 	/* For IP, only print the first header element. */
529 
530 	if (MAN_IP == n->tok && n->child)
531 		print_man_node(man, n->child, h);
532 
533 	/* For TP, only print next-line header elements. */
534 
535 	if (MAN_TP == n->tok) {
536 		nn = n->child;
537 		while (NULL != nn && 0 == (NODE_LINE & nn->flags))
538 			nn = nn->next;
539 		while (NULL != nn) {
540 			print_man_node(man, nn, h);
541 			nn = nn->next;
542 		}
543 	}
544 
545 	return 0;
546 }
547 
548 static int
549 man_HP_pre(MAN_ARGS)
550 {
551 	struct roffsu	 sum, sui;
552 	const struct roff_node *np;
553 
554 	if (n->type == ROFFT_HEAD)
555 		return 0;
556 	else if (n->type != ROFFT_BLOCK)
557 		return 1;
558 
559 	np = n->head->child;
560 
561 	if (np == NULL || !a2width(np, &sum))
562 		SCALE_HS_INIT(&sum, INDENT);
563 
564 	sui.unit = sum.unit;
565 	sui.scale = -sum.scale;
566 
567 	print_bvspace(h, n);
568 	print_otag(h, TAG_DIV, "csului", "Pp", &sum, &sui);
569 	return 1;
570 }
571 
572 static int
573 man_OP_pre(MAN_ARGS)
574 {
575 	struct tag	*tt;
576 
577 	print_text(h, "[");
578 	h->flags |= HTML_NOSPACE;
579 	tt = print_otag(h, TAG_SPAN, "c", "Op");
580 
581 	if (NULL != (n = n->child)) {
582 		print_otag(h, TAG_B, "");
583 		print_text(h, n->string);
584 	}
585 
586 	print_stagq(h, tt);
587 
588 	if (NULL != n && NULL != n->next) {
589 		print_otag(h, TAG_I, "");
590 		print_text(h, n->next->string);
591 	}
592 
593 	print_stagq(h, tt);
594 	h->flags |= HTML_NOSPACE;
595 	print_text(h, "]");
596 	return 0;
597 }
598 
599 static int
600 man_B_pre(MAN_ARGS)
601 {
602 	print_otag(h, TAG_B, "");
603 	return 1;
604 }
605 
606 static int
607 man_I_pre(MAN_ARGS)
608 {
609 	print_otag(h, TAG_I, "");
610 	return 1;
611 }
612 
613 static int
614 man_in_pre(MAN_ARGS)
615 {
616 	print_otag(h, TAG_BR, "");
617 	return 0;
618 }
619 
620 static int
621 man_ign_pre(MAN_ARGS)
622 {
623 
624 	return 0;
625 }
626 
627 static int
628 man_RS_pre(MAN_ARGS)
629 {
630 	struct roffsu	 su;
631 
632 	if (n->type == ROFFT_HEAD)
633 		return 0;
634 	else if (n->type == ROFFT_BODY)
635 		return 1;
636 
637 	SCALE_HS_INIT(&su, INDENT);
638 	if (n->head->child)
639 		a2width(n->head->child, &su);
640 
641 	print_otag(h, TAG_DIV, "sul", &su);
642 	return 1;
643 }
644 
645 static int
646 man_UR_pre(MAN_ARGS)
647 {
648 	char *cp;
649 	n = n->child;
650 	assert(n->type == ROFFT_HEAD);
651 	if (n->child != NULL) {
652 		assert(n->child->type == ROFFT_TEXT);
653 		if (n->tok == MAN_MT) {
654 			mandoc_asprintf(&cp, "mailto:%s", n->child->string);
655 			print_otag(h, TAG_A, "cTh", "Mt", cp);
656 			free(cp);
657 		} else
658 			print_otag(h, TAG_A, "cTh", "Lk", n->child->string);
659 	}
660 
661 	assert(n->next->type == ROFFT_BODY);
662 	if (n->next->child != NULL)
663 		n = n->next;
664 
665 	print_man_nodelist(man, n->child, h);
666 
667 	return 0;
668 }
669