xref: /illumos-gate/usr/src/cmd/mandoc/man_html.c (revision c7a0678d)
1 /*	$Id: man_html.c,v 1.145 2017/06/25 11:42:02 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 "config.h"
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 /* FIXME: have PD set the default vspace width. */
37 
38 #define	INDENT		  5
39 
40 #define	MAN_ARGS	  const struct roff_meta *man, \
41 			  const struct roff_node *n, \
42 			  struct html *h
43 
44 struct	htmlman {
45 	int		(*pre)(MAN_ARGS);
46 	int		(*post)(MAN_ARGS);
47 };
48 
49 static	void		  print_bvspace(struct html *,
50 				const struct roff_node *);
51 static	void		  print_man_head(MAN_ARGS);
52 static	void		  print_man_nodelist(MAN_ARGS);
53 static	void		  print_man_node(MAN_ARGS);
54 static	int		  fillmode(struct html *, int);
55 static	int		  a2width(const struct roff_node *,
56 				struct roffsu *);
57 static	int		  man_B_pre(MAN_ARGS);
58 static	int		  man_HP_pre(MAN_ARGS);
59 static	int		  man_IP_pre(MAN_ARGS);
60 static	int		  man_I_pre(MAN_ARGS);
61 static	int		  man_OP_pre(MAN_ARGS);
62 static	int		  man_PP_pre(MAN_ARGS);
63 static	int		  man_RS_pre(MAN_ARGS);
64 static	int		  man_SH_pre(MAN_ARGS);
65 static	int		  man_SM_pre(MAN_ARGS);
66 static	int		  man_SS_pre(MAN_ARGS);
67 static	int		  man_UR_pre(MAN_ARGS);
68 static	int		  man_alt_pre(MAN_ARGS);
69 static	int		  man_ign_pre(MAN_ARGS);
70 static	int		  man_in_pre(MAN_ARGS);
71 static	void		  man_root_post(MAN_ARGS);
72 static	void		  man_root_pre(MAN_ARGS);
73 
74 static	const struct htmlman __mans[MAN_MAX - MAN_TH] = {
75 	{ NULL, NULL }, /* TH */
76 	{ man_SH_pre, NULL }, /* SH */
77 	{ man_SS_pre, NULL }, /* SS */
78 	{ man_IP_pre, NULL }, /* TP */
79 	{ man_PP_pre, NULL }, /* LP */
80 	{ man_PP_pre, NULL }, /* PP */
81 	{ man_PP_pre, NULL }, /* P */
82 	{ man_IP_pre, NULL }, /* IP */
83 	{ man_HP_pre, NULL }, /* HP */
84 	{ man_SM_pre, NULL }, /* SM */
85 	{ man_SM_pre, NULL }, /* SB */
86 	{ man_alt_pre, NULL }, /* BI */
87 	{ man_alt_pre, NULL }, /* IB */
88 	{ man_alt_pre, NULL }, /* BR */
89 	{ man_alt_pre, NULL }, /* RB */
90 	{ NULL, NULL }, /* R */
91 	{ man_B_pre, NULL }, /* B */
92 	{ man_I_pre, NULL }, /* I */
93 	{ man_alt_pre, NULL }, /* IR */
94 	{ man_alt_pre, NULL }, /* RI */
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_OP_pre, NULL }, /* OP */
105 	{ NULL, NULL }, /* EX */
106 	{ NULL, NULL }, /* EE */
107 	{ man_UR_pre, NULL }, /* UR */
108 	{ NULL, NULL }, /* UE */
109 	{ man_UR_pre, NULL }, /* MT */
110 	{ NULL, NULL }, /* ME */
111 };
112 static	const struct htmlman *const mans = __mans - MAN_TH;
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 tag	*t;
143 
144 	h = (struct html *)arg;
145 
146 	if ((h->oflags & HTML_FRAGMENT) == 0) {
147 		print_gen_decls(h);
148 		print_otag(h, TAG_HTML, "");
149 		t = print_otag(h, TAG_HEAD, "");
150 		print_man_head(&man->meta, man->first, h);
151 		print_tagq(h, t);
152 		print_otag(h, TAG_BODY, "");
153 	}
154 
155 	man_root_pre(&man->meta, man->first, h);
156 	t = print_otag(h, TAG_DIV, "c", "manual-text");
157 	print_man_nodelist(&man->meta, man->first->child, h);
158 	print_tagq(h, t);
159 	man_root_post(&man->meta, man->first, h);
160 	print_tagq(h, NULL);
161 }
162 
163 static void
164 print_man_head(MAN_ARGS)
165 {
166 	char	*cp;
167 
168 	print_gen_head(h);
169 	mandoc_asprintf(&cp, "%s(%s)", man->title, man->msec);
170 	print_otag(h, TAG_TITLE, "");
171 	print_text(h, cp);
172 	free(cp);
173 }
174 
175 static void
176 print_man_nodelist(MAN_ARGS)
177 {
178 
179 	while (n != NULL) {
180 		print_man_node(man, n, h);
181 		n = n->next;
182 	}
183 }
184 
185 static void
186 print_man_node(MAN_ARGS)
187 {
188 	static int	 want_fillmode = MAN_fi;
189 	static int	 save_fillmode;
190 
191 	struct tag	*t;
192 	int		 child;
193 
194 	/*
195 	 * Handle fill mode switch requests up front,
196 	 * they would just cause trouble in the subsequent code.
197 	 */
198 
199 	switch (n->tok) {
200 	case MAN_nf:
201 	case MAN_EX:
202 		want_fillmode = MAN_nf;
203 		return;
204 	case MAN_fi:
205 	case MAN_EE:
206 		want_fillmode = MAN_fi;
207 		if (fillmode(h, 0) == MAN_fi)
208 			print_otag(h, TAG_BR, "");
209 		return;
210 	default:
211 		break;
212 	}
213 
214 	/* Set up fill mode for the upcoming node. */
215 
216 	switch (n->type) {
217 	case ROFFT_BLOCK:
218 		save_fillmode = 0;
219 		/* Some block macros suspend or cancel .nf. */
220 		switch (n->tok) {
221 		case MAN_TP:  /* Tagged paragraphs		*/
222 		case MAN_IP:  /* temporarily disable .nf	*/
223 		case MAN_HP:  /* for the head.			*/
224 			save_fillmode = want_fillmode;
225 			/* FALLTHROUGH */
226 		case MAN_SH:  /* Section headers		*/
227 		case MAN_SS:  /* permanently cancel .nf.	*/
228 			want_fillmode = MAN_fi;
229 			/* FALLTHROUGH */
230 		case MAN_PP:  /* These have no head.		*/
231 		case MAN_LP:  /* They will simply		*/
232 		case MAN_P:   /* reopen .nf in the body.	*/
233 		case MAN_RS:
234 		case MAN_UR:
235 		case MAN_MT:
236 			fillmode(h, MAN_fi);
237 			break;
238 		default:
239 			break;
240 		}
241 		break;
242 	case ROFFT_TBL:
243 		fillmode(h, MAN_fi);
244 		break;
245 	case ROFFT_ELEM:
246 		/*
247 		 * Some in-line macros produce tags and/or text
248 		 * in the handler, so they require fill mode to be
249 		 * configured up front just like for text nodes.
250 		 * For the others, keep the traditional approach
251 		 * of doing the same, for now.
252 		 */
253 		fillmode(h, want_fillmode);
254 		break;
255 	case ROFFT_TEXT:
256 		if (fillmode(h, want_fillmode) == MAN_fi &&
257 		    want_fillmode == MAN_fi &&
258 		    n->flags & NODE_LINE && *n->string == ' ' &&
259 		    (h->flags & HTML_NONEWLINE) == 0)
260 			print_otag(h, TAG_BR, "");
261 		if (*n->string != '\0')
262 			break;
263 		print_paragraph(h);
264 		return;
265 	default:
266 		break;
267 	}
268 
269 	/* Produce output for this node. */
270 
271 	child = 1;
272 	switch (n->type) {
273 	case ROFFT_TEXT:
274 		t = h->tag;
275 		print_text(h, n->string);
276 		break;
277 	case ROFFT_EQN:
278 		t = h->tag;
279 		print_eqn(h, n->eqn);
280 		break;
281 	case ROFFT_TBL:
282 		/*
283 		 * This will take care of initialising all of the table
284 		 * state data for the first table, then tearing it down
285 		 * for the last one.
286 		 */
287 		print_tbl(h, n->span);
288 		return;
289 	default:
290 		/*
291 		 * Close out scope of font prior to opening a macro
292 		 * scope.
293 		 */
294 		if (HTMLFONT_NONE != h->metac) {
295 			h->metal = h->metac;
296 			h->metac = HTMLFONT_NONE;
297 		}
298 
299 		/*
300 		 * Close out the current table, if it's open, and unset
301 		 * the "meta" table state.  This will be reopened on the
302 		 * next table element.
303 		 */
304 		if (h->tblt)
305 			print_tblclose(h);
306 
307 		t = h->tag;
308 		if (n->tok < ROFF_MAX) {
309 			roff_html_pre(h, n);
310 			child = 0;
311 			break;
312 		}
313 
314 		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
315 		if (mans[n->tok].pre)
316 			child = (*mans[n->tok].pre)(man, n, h);
317 
318 		/* Some block macros resume .nf in the body. */
319 		if (save_fillmode && n->type == ROFFT_BODY)
320 			want_fillmode = save_fillmode;
321 
322 		break;
323 	}
324 
325 	if (child && n->child)
326 		print_man_nodelist(man, n->child, h);
327 
328 	/* This will automatically close out any font scope. */
329 	print_stagq(h, t);
330 
331 	if (fillmode(h, 0) == MAN_nf &&
332 	    n->next != NULL && n->next->flags & NODE_LINE)
333 		print_endline(h);
334 }
335 
336 /*
337  * MAN_nf switches to no-fill mode, MAN_fi to fill mode.
338  * Other arguments do not switch.
339  * The old mode is returned.
340  */
341 static int
342 fillmode(struct html *h, int want)
343 {
344 	struct tag	*pre;
345 	int		 had;
346 
347 	for (pre = h->tag; pre != NULL; pre = pre->next)
348 		if (pre->tag == TAG_PRE)
349 			break;
350 
351 	had = pre == NULL ? MAN_fi : MAN_nf;
352 
353 	if (want && want != had) {
354 		if (want == MAN_nf)
355 			print_otag(h, TAG_PRE, "");
356 		else
357 			print_tagq(h, pre);
358 	}
359 	return had;
360 }
361 
362 static int
363 a2width(const struct roff_node *n, struct roffsu *su)
364 {
365 	if (n->type != ROFFT_TEXT)
366 		return 0;
367 	return a2roffsu(n->string, su, SCALE_EN) != NULL;
368 }
369 
370 static void
371 man_root_pre(MAN_ARGS)
372 {
373 	struct tag	*t, *tt;
374 	char		*title;
375 
376 	assert(man->title);
377 	assert(man->msec);
378 	mandoc_asprintf(&title, "%s(%s)", man->title, man->msec);
379 
380 	t = print_otag(h, TAG_TABLE, "c", "head");
381 	tt = print_otag(h, TAG_TR, "");
382 
383 	print_otag(h, TAG_TD, "c", "head-ltitle");
384 	print_text(h, title);
385 	print_stagq(h, tt);
386 
387 	print_otag(h, TAG_TD, "c", "head-vol");
388 	if (NULL != man->vol)
389 		print_text(h, man->vol);
390 	print_stagq(h, tt);
391 
392 	print_otag(h, TAG_TD, "c", "head-rtitle");
393 	print_text(h, title);
394 	print_tagq(h, t);
395 	free(title);
396 }
397 
398 static void
399 man_root_post(MAN_ARGS)
400 {
401 	struct tag	*t, *tt;
402 
403 	t = print_otag(h, TAG_TABLE, "c", "foot");
404 	tt = print_otag(h, TAG_TR, "");
405 
406 	print_otag(h, TAG_TD, "c", "foot-date");
407 	print_text(h, man->date);
408 	print_stagq(h, tt);
409 
410 	print_otag(h, TAG_TD, "c", "foot-os");
411 	if (man->os)
412 		print_text(h, man->os);
413 	print_tagq(h, t);
414 }
415 
416 static int
417 man_SH_pre(MAN_ARGS)
418 {
419 	char	*id;
420 
421 	if (n->type == ROFFT_HEAD) {
422 		id = html_make_id(n);
423 		print_otag(h, TAG_H1, "cTi", "Sh", id);
424 		if (id != NULL)
425 			print_otag(h, TAG_A, "chR", "selflink", id);
426 		free(id);
427 	}
428 	return 1;
429 }
430 
431 static int
432 man_alt_pre(MAN_ARGS)
433 {
434 	const struct roff_node	*nn;
435 	int		 i;
436 	enum htmltag	 fp;
437 	struct tag	*t;
438 
439 	for (i = 0, nn = n->child; nn; nn = nn->next, i++) {
440 		switch (n->tok) {
441 		case MAN_BI:
442 			fp = i % 2 ? TAG_I : TAG_B;
443 			break;
444 		case MAN_IB:
445 			fp = i % 2 ? TAG_B : TAG_I;
446 			break;
447 		case MAN_RI:
448 			fp = i % 2 ? TAG_I : TAG_MAX;
449 			break;
450 		case MAN_IR:
451 			fp = i % 2 ? TAG_MAX : TAG_I;
452 			break;
453 		case MAN_BR:
454 			fp = i % 2 ? TAG_MAX : TAG_B;
455 			break;
456 		case MAN_RB:
457 			fp = i % 2 ? TAG_B : TAG_MAX;
458 			break;
459 		default:
460 			abort();
461 		}
462 
463 		if (i)
464 			h->flags |= HTML_NOSPACE;
465 
466 		if (fp != TAG_MAX)
467 			t = print_otag(h, fp, "");
468 
469 		print_text(h, nn->string);
470 
471 		if (fp != TAG_MAX)
472 			print_tagq(h, t);
473 	}
474 	return 0;
475 }
476 
477 static int
478 man_SM_pre(MAN_ARGS)
479 {
480 	print_otag(h, TAG_SMALL, "");
481 	if (MAN_SB == n->tok)
482 		print_otag(h, TAG_B, "");
483 	return 1;
484 }
485 
486 static int
487 man_SS_pre(MAN_ARGS)
488 {
489 	char	*id;
490 
491 	if (n->type == ROFFT_HEAD) {
492 		id = html_make_id(n);
493 		print_otag(h, TAG_H2, "cTi", "Ss", id);
494 		if (id != NULL)
495 			print_otag(h, TAG_A, "chR", "selflink", id);
496 		free(id);
497 	}
498 	return 1;
499 }
500 
501 static int
502 man_PP_pre(MAN_ARGS)
503 {
504 
505 	if (n->type == ROFFT_HEAD)
506 		return 0;
507 	else if (n->type == ROFFT_BLOCK)
508 		print_bvspace(h, n);
509 
510 	return 1;
511 }
512 
513 static int
514 man_IP_pre(MAN_ARGS)
515 {
516 	const struct roff_node	*nn;
517 
518 	if (n->type == ROFFT_BODY) {
519 		print_otag(h, TAG_DD, "c", "It-tag");
520 		return 1;
521 	} else if (n->type != ROFFT_HEAD) {
522 		print_otag(h, TAG_DL, "c", "Bl-tag");
523 		return 1;
524 	}
525 
526 	/* FIXME: width specification. */
527 
528 	print_otag(h, TAG_DT, "c", "It-tag");
529 
530 	/* For IP, only print the first header element. */
531 
532 	if (MAN_IP == n->tok && n->child)
533 		print_man_node(man, n->child, h);
534 
535 	/* For TP, only print next-line header elements. */
536 
537 	if (MAN_TP == n->tok) {
538 		nn = n->child;
539 		while (NULL != nn && 0 == (NODE_LINE & nn->flags))
540 			nn = nn->next;
541 		while (NULL != nn) {
542 			print_man_node(man, nn, h);
543 			nn = nn->next;
544 		}
545 	}
546 
547 	return 0;
548 }
549 
550 static int
551 man_HP_pre(MAN_ARGS)
552 {
553 	struct roffsu	 sum, sui;
554 	const struct roff_node *np;
555 
556 	if (n->type == ROFFT_HEAD)
557 		return 0;
558 	else if (n->type != ROFFT_BLOCK)
559 		return 1;
560 
561 	np = n->head->child;
562 
563 	if (np == NULL || !a2width(np, &sum))
564 		SCALE_HS_INIT(&sum, INDENT);
565 
566 	sui.unit = sum.unit;
567 	sui.scale = -sum.scale;
568 
569 	print_bvspace(h, n);
570 	print_otag(h, TAG_DIV, "csului", "Pp", &sum, &sui);
571 	return 1;
572 }
573 
574 static int
575 man_OP_pre(MAN_ARGS)
576 {
577 	struct tag	*tt;
578 
579 	print_text(h, "[");
580 	h->flags |= HTML_NOSPACE;
581 	tt = print_otag(h, TAG_SPAN, "c", "Op");
582 
583 	if (NULL != (n = n->child)) {
584 		print_otag(h, TAG_B, "");
585 		print_text(h, n->string);
586 	}
587 
588 	print_stagq(h, tt);
589 
590 	if (NULL != n && NULL != n->next) {
591 		print_otag(h, TAG_I, "");
592 		print_text(h, n->next->string);
593 	}
594 
595 	print_stagq(h, tt);
596 	h->flags |= HTML_NOSPACE;
597 	print_text(h, "]");
598 	return 0;
599 }
600 
601 static int
602 man_B_pre(MAN_ARGS)
603 {
604 	print_otag(h, TAG_B, "");
605 	return 1;
606 }
607 
608 static int
609 man_I_pre(MAN_ARGS)
610 {
611 	print_otag(h, TAG_I, "");
612 	return 1;
613 }
614 
615 static int
616 man_in_pre(MAN_ARGS)
617 {
618 	print_otag(h, TAG_BR, "");
619 	return 0;
620 }
621 
622 static int
623 man_ign_pre(MAN_ARGS)
624 {
625 
626 	return 0;
627 }
628 
629 static int
630 man_RS_pre(MAN_ARGS)
631 {
632 	struct roffsu	 su;
633 
634 	if (n->type == ROFFT_HEAD)
635 		return 0;
636 	else if (n->type == ROFFT_BODY)
637 		return 1;
638 
639 	SCALE_HS_INIT(&su, INDENT);
640 	if (n->head->child)
641 		a2width(n->head->child, &su);
642 
643 	print_otag(h, TAG_DIV, "sul", &su);
644 	return 1;
645 }
646 
647 static int
648 man_UR_pre(MAN_ARGS)
649 {
650 	char *cp;
651 	n = n->child;
652 	assert(n->type == ROFFT_HEAD);
653 	if (n->child != NULL) {
654 		assert(n->child->type == ROFFT_TEXT);
655 		if (n->tok == MAN_MT) {
656 			mandoc_asprintf(&cp, "mailto:%s", n->child->string);
657 			print_otag(h, TAG_A, "cTh", "Mt", cp);
658 			free(cp);
659 		} else
660 			print_otag(h, TAG_A, "cTh", "Lk", n->child->string);
661 	}
662 
663 	assert(n->next->type == ROFFT_BODY);
664 	if (n->next->child != NULL)
665 		n = n->next;
666 
667 	print_man_nodelist(man, n->child, h);
668 
669 	return 0;
670 }
671