xref: /openbsd-src/usr.bin/mandoc/man_html.c (revision f933361f20df4def12f9bc8391f68d6bfad3bb75)
1 /*	$OpenBSD: man_html.c,v 1.96 2017/06/08 12:54:40 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_ign_pre(MAN_ARGS);
67 static	int		  man_in_pre(MAN_ARGS);
68 static	void		  man_root_post(MAN_ARGS);
69 static	void		  man_root_pre(MAN_ARGS);
70 
71 static	const struct htmlman __mans[MAN_MAX - MAN_TH] = {
72 	{ NULL, NULL }, /* TH */
73 	{ man_SH_pre, NULL }, /* SH */
74 	{ man_SS_pre, NULL }, /* SS */
75 	{ man_IP_pre, NULL }, /* TP */
76 	{ man_PP_pre, NULL }, /* LP */
77 	{ man_PP_pre, NULL }, /* PP */
78 	{ man_PP_pre, NULL }, /* P */
79 	{ man_IP_pre, NULL }, /* IP */
80 	{ man_HP_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 }, /* nf */
93 	{ NULL, NULL }, /* fi */
94 	{ NULL, NULL }, /* RE */
95 	{ man_RS_pre, NULL }, /* RS */
96 	{ man_ign_pre, NULL }, /* DT */
97 	{ man_ign_pre, NULL }, /* UC */
98 	{ man_ign_pre, NULL }, /* PD */
99 	{ man_ign_pre, NULL }, /* AT */
100 	{ man_in_pre, NULL }, /* in */
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 };
107 static	const struct htmlman *const mans = __mans - MAN_TH;
108 
109 
110 /*
111  * Printing leading vertical space before a block.
112  * This is used for the paragraph macros.
113  * The rules are pretty simple, since there's very little nesting going
114  * on here.  Basically, if we're the first within another block (SS/SH),
115  * then don't emit vertical space.  If we are (RS), then do.  If not the
116  * first, print it.
117  */
118 static void
119 print_bvspace(struct html *h, const struct roff_node *n)
120 {
121 
122 	if (n->body && n->body->child)
123 		if (n->body->child->type == ROFFT_TBL)
124 			return;
125 
126 	if (n->parent->type == ROFFT_ROOT || n->parent->tok != MAN_RS)
127 		if (NULL == n->prev)
128 			return;
129 
130 	print_paragraph(h);
131 }
132 
133 void
134 html_man(void *arg, const struct roff_man *man)
135 {
136 	struct html	*h;
137 	struct tag	*t;
138 
139 	h = (struct html *)arg;
140 
141 	if ((h->oflags & HTML_FRAGMENT) == 0) {
142 		print_gen_decls(h);
143 		print_otag(h, TAG_HTML, "");
144 		t = print_otag(h, TAG_HEAD, "");
145 		print_man_head(&man->meta, man->first, h);
146 		print_tagq(h, t);
147 		print_otag(h, TAG_BODY, "");
148 	}
149 
150 	man_root_pre(&man->meta, man->first, h);
151 	t = print_otag(h, TAG_DIV, "c", "manual-text");
152 	print_man_nodelist(&man->meta, man->first->child, h);
153 	print_tagq(h, t);
154 	man_root_post(&man->meta, man->first, h);
155 	print_tagq(h, NULL);
156 }
157 
158 static void
159 print_man_head(MAN_ARGS)
160 {
161 	char	*cp;
162 
163 	print_gen_head(h);
164 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
165 	print_otag(h, TAG_TITLE, "");
166 	print_text(h, cp);
167 	free(cp);
168 }
169 
170 static void
171 print_man_nodelist(MAN_ARGS)
172 {
173 
174 	while (n != NULL) {
175 		print_man_node(man, n, h);
176 		n = n->next;
177 	}
178 }
179 
180 static void
181 print_man_node(MAN_ARGS)
182 {
183 	static int	 want_fillmode = MAN_fi;
184 	static int	 save_fillmode;
185 
186 	struct tag	*t;
187 	int		 child;
188 
189 	/*
190 	 * Handle fill mode switch requests up front,
191 	 * they would just cause trouble in the subsequent code.
192 	 */
193 
194 	switch (n->tok) {
195 	case MAN_nf:
196 	case MAN_EX:
197 		want_fillmode = MAN_nf;
198 		return;
199 	case MAN_fi:
200 	case MAN_EE:
201 		want_fillmode = MAN_fi;
202 		if (fillmode(h, 0) == MAN_fi)
203 			print_otag(h, TAG_BR, "");
204 		return;
205 	default:
206 		break;
207 	}
208 
209 	/* Set up fill mode for the upcoming node. */
210 
211 	switch (n->type) {
212 	case ROFFT_BLOCK:
213 		save_fillmode = 0;
214 		/* Some block macros suspend or cancel .nf. */
215 		switch (n->tok) {
216 		case MAN_TP:  /* Tagged paragraphs		*/
217 		case MAN_IP:  /* temporarily disable .nf	*/
218 		case MAN_HP:  /* for the head.			*/
219 			save_fillmode = want_fillmode;
220 			/* FALLTHROUGH */
221 		case MAN_SH:  /* Section headers		*/
222 		case MAN_SS:  /* permanently cancel .nf.	*/
223 			want_fillmode = MAN_fi;
224 			/* FALLTHROUGH */
225 		case MAN_PP:  /* These have no head.		*/
226 		case MAN_LP:  /* They will simply		*/
227 		case MAN_P:   /* reopen .nf in the body.	*/
228 		case MAN_RS:
229 		case MAN_UR:
230 			fillmode(h, MAN_fi);
231 			break;
232 		default:
233 			break;
234 		}
235 		break;
236 	case ROFFT_TBL:
237 		fillmode(h, MAN_fi);
238 		break;
239 	case ROFFT_ELEM:
240 		/*
241 		 * Some in-line macros produce tags and/or text
242 		 * in the handler, so they require fill mode to be
243 		 * configured up front just like for text nodes.
244 		 * For the others, keep the traditional approach
245 		 * of doing the same, for now.
246 		 */
247 		fillmode(h, want_fillmode);
248 		break;
249 	case ROFFT_TEXT:
250 		if (fillmode(h, want_fillmode) == MAN_fi &&
251 		    want_fillmode == MAN_fi &&
252 		    n->flags & NODE_LINE && *n->string == ' ' &&
253 		    (h->flags & HTML_NONEWLINE) == 0)
254 			print_otag(h, TAG_BR, "");
255 		if (*n->string != '\0')
256 			break;
257 		print_paragraph(h);
258 		return;
259 	default:
260 		break;
261 	}
262 
263 	/* Produce output for this node. */
264 
265 	child = 1;
266 	switch (n->type) {
267 	case ROFFT_TEXT:
268 		t = h->tag;
269 		print_text(h, n->string);
270 		break;
271 	case ROFFT_EQN:
272 		t = h->tag;
273 		print_eqn(h, n->eqn);
274 		break;
275 	case ROFFT_TBL:
276 		/*
277 		 * This will take care of initialising all of the table
278 		 * state data for the first table, then tearing it down
279 		 * for the last one.
280 		 */
281 		print_tbl(h, n->span);
282 		return;
283 	default:
284 		/*
285 		 * Close out scope of font prior to opening a macro
286 		 * scope.
287 		 */
288 		if (HTMLFONT_NONE != h->metac) {
289 			h->metal = h->metac;
290 			h->metac = HTMLFONT_NONE;
291 		}
292 
293 		/*
294 		 * Close out the current table, if it's open, and unset
295 		 * the "meta" table state.  This will be reopened on the
296 		 * next table element.
297 		 */
298 		if (h->tblt)
299 			print_tblclose(h);
300 
301 		t = h->tag;
302 		if (n->tok < ROFF_MAX) {
303 			roff_html_pre(h, n);
304 			child = 0;
305 			break;
306 		}
307 
308 		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
309 		if (mans[n->tok].pre)
310 			child = (*mans[n->tok].pre)(man, n, h);
311 
312 		/* Some block macros resume .nf in the body. */
313 		if (save_fillmode && n->type == ROFFT_BODY)
314 			want_fillmode = save_fillmode;
315 
316 		break;
317 	}
318 
319 	if (child && n->child)
320 		print_man_nodelist(man, n->child, h);
321 
322 	/* This will automatically close out any font scope. */
323 	print_stagq(h, t);
324 
325 	if (fillmode(h, 0) == MAN_nf &&
326 	    n->next != NULL && n->next->flags & NODE_LINE)
327 		print_endline(h);
328 }
329 
330 /*
331  * MAN_nf switches to no-fill mode, MAN_fi to fill mode.
332  * Other arguments do not switch.
333  * The old mode is returned.
334  */
335 static int
336 fillmode(struct html *h, int want)
337 {
338 	struct tag	*pre;
339 	int		 had;
340 
341 	for (pre = h->tag; pre != NULL; pre = pre->next)
342 		if (pre->tag == TAG_PRE)
343 			break;
344 
345 	had = pre == NULL ? MAN_fi : MAN_nf;
346 
347 	if (want && want != had) {
348 		if (want == MAN_nf)
349 			print_otag(h, TAG_PRE, "");
350 		else
351 			print_tagq(h, pre);
352 	}
353 	return had;
354 }
355 
356 static int
357 a2width(const struct roff_node *n, struct roffsu *su)
358 {
359 	if (n->type != ROFFT_TEXT)
360 		return 0;
361 	return a2roffsu(n->string, su, SCALE_EN) != NULL;
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 static int
411 man_SH_pre(MAN_ARGS)
412 {
413 	char	*id;
414 
415 	if (n->type == ROFFT_HEAD) {
416 		id = html_make_id(n);
417 		print_otag(h, TAG_H1, "cTi", "Sh", id);
418 		if (id != NULL)
419 			print_otag(h, TAG_A, "chR", "selflink", id);
420 		free(id);
421 	}
422 	return 1;
423 }
424 
425 static int
426 man_alt_pre(MAN_ARGS)
427 {
428 	const struct roff_node	*nn;
429 	int		 i;
430 	enum htmltag	 fp;
431 	struct tag	*t;
432 
433 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
434 		switch (n->tok) {
435 		case MAN_BI:
436 			fp = i % 2 ? TAG_I : TAG_B;
437 			break;
438 		case MAN_IB:
439 			fp = i % 2 ? TAG_B : TAG_I;
440 			break;
441 		case MAN_RI:
442 			fp = i % 2 ? TAG_I : TAG_MAX;
443 			break;
444 		case MAN_IR:
445 			fp = i % 2 ? TAG_MAX : TAG_I;
446 			break;
447 		case MAN_BR:
448 			fp = i % 2 ? TAG_MAX : TAG_B;
449 			break;
450 		case MAN_RB:
451 			fp = i % 2 ? TAG_B : TAG_MAX;
452 			break;
453 		default:
454 			abort();
455 		}
456 
457 		if (i)
458 			h->flags |= HTML_NOSPACE;
459 
460 		if (fp != TAG_MAX)
461 			t = print_otag(h, fp, "");
462 
463 		print_text(h, nn->string);
464 
465 		if (fp != TAG_MAX)
466 			print_tagq(h, t);
467 	}
468 	return 0;
469 }
470 
471 static int
472 man_SM_pre(MAN_ARGS)
473 {
474 	print_otag(h, TAG_SMALL, "");
475 	if (MAN_SB == n->tok)
476 		print_otag(h, TAG_B, "");
477 	return 1;
478 }
479 
480 static int
481 man_SS_pre(MAN_ARGS)
482 {
483 	char	*id;
484 
485 	if (n->type == ROFFT_HEAD) {
486 		id = html_make_id(n);
487 		print_otag(h, TAG_H2, "cTi", "Ss", id);
488 		if (id != NULL)
489 			print_otag(h, TAG_A, "chR", "selflink", id);
490 		free(id);
491 	}
492 	return 1;
493 }
494 
495 static int
496 man_PP_pre(MAN_ARGS)
497 {
498 
499 	if (n->type == ROFFT_HEAD)
500 		return 0;
501 	else if (n->type == ROFFT_BLOCK)
502 		print_bvspace(h, n);
503 
504 	return 1;
505 }
506 
507 static int
508 man_IP_pre(MAN_ARGS)
509 {
510 	const struct roff_node	*nn;
511 
512 	if (n->type == ROFFT_BODY) {
513 		print_otag(h, TAG_DD, "c", "It-tag");
514 		return 1;
515 	} else if (n->type != ROFFT_HEAD) {
516 		print_otag(h, TAG_DL, "c", "Bl-tag");
517 		return 1;
518 	}
519 
520 	/* FIXME: width specification. */
521 
522 	print_otag(h, TAG_DT, "c", "It-tag");
523 
524 	/* For IP, only print the first header element. */
525 
526 	if (MAN_IP == n->tok && n->child)
527 		print_man_node(man, n->child, h);
528 
529 	/* For TP, only print next-line header elements. */
530 
531 	if (MAN_TP == n->tok) {
532 		nn = n->child;
533 		while (NULL != nn && 0 == (NODE_LINE & nn->flags))
534 			nn = nn->next;
535 		while (NULL != nn) {
536 			print_man_node(man, nn, h);
537 			nn = nn->next;
538 		}
539 	}
540 
541 	return 0;
542 }
543 
544 static int
545 man_HP_pre(MAN_ARGS)
546 {
547 	struct roffsu	 sum, sui;
548 	const struct roff_node *np;
549 
550 	if (n->type == ROFFT_HEAD)
551 		return 0;
552 	else if (n->type != ROFFT_BLOCK)
553 		return 1;
554 
555 	np = n->head->child;
556 
557 	if (np == NULL || !a2width(np, &sum))
558 		SCALE_HS_INIT(&sum, INDENT);
559 
560 	sui.unit = sum.unit;
561 	sui.scale = -sum.scale;
562 
563 	print_bvspace(h, n);
564 	print_otag(h, TAG_DIV, "csului", "Pp", &sum, &sui);
565 	return 1;
566 }
567 
568 static int
569 man_OP_pre(MAN_ARGS)
570 {
571 	struct tag	*tt;
572 
573 	print_text(h, "[");
574 	h->flags |= HTML_NOSPACE;
575 	tt = print_otag(h, TAG_SPAN, "c", "Op");
576 
577 	if (NULL != (n = n->child)) {
578 		print_otag(h, TAG_B, "");
579 		print_text(h, n->string);
580 	}
581 
582 	print_stagq(h, tt);
583 
584 	if (NULL != n && NULL != n->next) {
585 		print_otag(h, TAG_I, "");
586 		print_text(h, n->next->string);
587 	}
588 
589 	print_stagq(h, tt);
590 	h->flags |= HTML_NOSPACE;
591 	print_text(h, "]");
592 	return 0;
593 }
594 
595 static int
596 man_B_pre(MAN_ARGS)
597 {
598 	print_otag(h, TAG_B, "");
599 	return 1;
600 }
601 
602 static int
603 man_I_pre(MAN_ARGS)
604 {
605 	print_otag(h, TAG_I, "");
606 	return 1;
607 }
608 
609 static int
610 man_in_pre(MAN_ARGS)
611 {
612 	print_otag(h, TAG_BR, "");
613 	return 0;
614 }
615 
616 static int
617 man_ign_pre(MAN_ARGS)
618 {
619 
620 	return 0;
621 }
622 
623 static int
624 man_RS_pre(MAN_ARGS)
625 {
626 	struct roffsu	 su;
627 
628 	if (n->type == ROFFT_HEAD)
629 		return 0;
630 	else if (n->type == ROFFT_BODY)
631 		return 1;
632 
633 	SCALE_HS_INIT(&su, INDENT);
634 	if (n->head->child)
635 		a2width(n->head->child, &su);
636 
637 	print_otag(h, TAG_DIV, "sul", &su);
638 	return 1;
639 }
640 
641 static int
642 man_UR_pre(MAN_ARGS)
643 {
644 	n = n->child;
645 	assert(n->type == ROFFT_HEAD);
646 	if (n->child != NULL) {
647 		assert(n->child->type == ROFFT_TEXT);
648 		print_otag(h, TAG_A, "cTh", "Lk", n->child->string);
649 	}
650 
651 	assert(n->next->type == ROFFT_BODY);
652 	if (n->next->child != NULL)
653 		n = n->next;
654 
655 	print_man_nodelist(man, n->child, h);
656 
657 	return 0;
658 }
659