1/*	$Id: man.c,v 1.115 2012/01/03 15:16:24 kristaps Exp $ */
2/*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 *
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
8 *
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 */
17#ifdef HAVE_CONFIG_H
18#include "config.h"
19#endif
20
21#include <sys/types.h>
22
23#include <assert.h>
24#include <stdarg.h>
25#include <stdlib.h>
26#include <stdio.h>
27#include <string.h>
28
29#include "man.h"
30#include "mandoc.h"
31#include "libman.h"
32#include "libmandoc.h"
33
34const	char *const __man_macronames[MAN_MAX] = {
35	"br",		"TH",		"SH",		"SS",
36	"TP", 		"LP",		"PP",		"P",
37	"IP",		"HP",		"SM",		"SB",
38	"BI",		"IB",		"BR",		"RB",
39	"R",		"B",		"I",		"IR",
40	"RI",		"na",		"sp",		"nf",
41	"fi",		"RE",		"RS",		"DT",
42	"UC",		"PD",		"AT",		"in",
43	"ft",		"OP"
44	};
45
46const	char * const *man_macronames = __man_macronames;
47
48static	struct man_node	*man_node_alloc(struct man *, int, int,
49				enum man_type, enum mant);
50static	int		 man_node_append(struct man *,
51				struct man_node *);
52static	void		 man_node_free(struct man_node *);
53static	void		 man_node_unlink(struct man *,
54				struct man_node *);
55static	int		 man_ptext(struct man *, int, char *, int);
56static	int		 man_pmacro(struct man *, int, char *, int);
57static	void		 man_free1(struct man *);
58static	void		 man_alloc1(struct man *);
59static	int		 man_descope(struct man *, int, int);
60
61
62const struct man_node *
63man_node(const struct man *m)
64{
65
66	assert( ! (MAN_HALT & m->flags));
67	return(m->first);
68}
69
70
71const struct man_meta *
72man_meta(const struct man *m)
73{
74
75	assert( ! (MAN_HALT & m->flags));
76	return(&m->meta);
77}
78
79
80void
81man_reset(struct man *man)
82{
83
84	man_free1(man);
85	man_alloc1(man);
86}
87
88
89void
90man_free(struct man *man)
91{
92
93	man_free1(man);
94	free(man);
95}
96
97
98struct man *
99man_alloc(struct roff *roff, struct mparse *parse)
100{
101	struct man	*p;
102
103	p = mandoc_calloc(1, sizeof(struct man));
104
105	man_hash_init();
106	p->parse = parse;
107	p->roff = roff;
108
109	man_alloc1(p);
110	return(p);
111}
112
113
114int
115man_endparse(struct man *m)
116{
117
118	assert( ! (MAN_HALT & m->flags));
119	if (man_macroend(m))
120		return(1);
121	m->flags |= MAN_HALT;
122	return(0);
123}
124
125
126int
127man_parseln(struct man *m, int ln, char *buf, int offs)
128{
129
130	m->flags |= MAN_NEWLINE;
131
132	assert( ! (MAN_HALT & m->flags));
133
134	return (mandoc_getcontrol(buf, &offs) ?
135			man_pmacro(m, ln, buf, offs) :
136			man_ptext(m, ln, buf, offs));
137}
138
139
140static void
141man_free1(struct man *man)
142{
143
144	if (man->first)
145		man_node_delete(man, man->first);
146	if (man->meta.title)
147		free(man->meta.title);
148	if (man->meta.source)
149		free(man->meta.source);
150	if (man->meta.date)
151		free(man->meta.date);
152	if (man->meta.vol)
153		free(man->meta.vol);
154	if (man->meta.msec)
155		free(man->meta.msec);
156}
157
158
159static void
160man_alloc1(struct man *m)
161{
162
163	memset(&m->meta, 0, sizeof(struct man_meta));
164	m->flags = 0;
165	m->last = mandoc_calloc(1, sizeof(struct man_node));
166	m->first = m->last;
167	m->last->type = MAN_ROOT;
168	m->last->tok = MAN_MAX;
169	m->next = MAN_NEXT_CHILD;
170}
171
172
173static int
174man_node_append(struct man *man, struct man_node *p)
175{
176
177	assert(man->last);
178	assert(man->first);
179	assert(MAN_ROOT != p->type);
180
181	switch (man->next) {
182	case (MAN_NEXT_SIBLING):
183		man->last->next = p;
184		p->prev = man->last;
185		p->parent = man->last->parent;
186		break;
187	case (MAN_NEXT_CHILD):
188		man->last->child = p;
189		p->parent = man->last;
190		break;
191	default:
192		abort();
193		/* NOTREACHED */
194	}
195
196	assert(p->parent);
197	p->parent->nchild++;
198
199	if ( ! man_valid_pre(man, p))
200		return(0);
201
202	switch (p->type) {
203	case (MAN_HEAD):
204		assert(MAN_BLOCK == p->parent->type);
205		p->parent->head = p;
206		break;
207	case (MAN_TAIL):
208		assert(MAN_BLOCK == p->parent->type);
209		p->parent->tail = p;
210		break;
211	case (MAN_BODY):
212		assert(MAN_BLOCK == p->parent->type);
213		p->parent->body = p;
214		break;
215	default:
216		break;
217	}
218
219	man->last = p;
220
221	switch (p->type) {
222	case (MAN_TBL):
223		/* FALLTHROUGH */
224	case (MAN_TEXT):
225		if ( ! man_valid_post(man))
226			return(0);
227		break;
228	default:
229		break;
230	}
231
232	return(1);
233}
234
235
236static struct man_node *
237man_node_alloc(struct man *m, int line, int pos,
238		enum man_type type, enum mant tok)
239{
240	struct man_node *p;
241
242	p = mandoc_calloc(1, sizeof(struct man_node));
243	p->line = line;
244	p->pos = pos;
245	p->type = type;
246	p->tok = tok;
247
248	if (MAN_NEWLINE & m->flags)
249		p->flags |= MAN_LINE;
250	m->flags &= ~MAN_NEWLINE;
251	return(p);
252}
253
254
255int
256man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
257{
258	struct man_node *p;
259
260	p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
261	if ( ! man_node_append(m, p))
262		return(0);
263	m->next = MAN_NEXT_CHILD;
264	return(1);
265}
266
267
268int
269man_tail_alloc(struct man *m, int line, int pos, enum mant tok)
270{
271	struct man_node *p;
272
273	p = man_node_alloc(m, line, pos, MAN_TAIL, tok);
274	if ( ! man_node_append(m, p))
275		return(0);
276	m->next = MAN_NEXT_CHILD;
277	return(1);
278}
279
280
281int
282man_head_alloc(struct man *m, int line, int pos, enum mant tok)
283{
284	struct man_node *p;
285
286	p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
287	if ( ! man_node_append(m, p))
288		return(0);
289	m->next = MAN_NEXT_CHILD;
290	return(1);
291}
292
293
294int
295man_body_alloc(struct man *m, int line, int pos, enum mant tok)
296{
297	struct man_node *p;
298
299	p = man_node_alloc(m, line, pos, MAN_BODY, tok);
300	if ( ! man_node_append(m, p))
301		return(0);
302	m->next = MAN_NEXT_CHILD;
303	return(1);
304}
305
306
307int
308man_block_alloc(struct man *m, int line, int pos, enum mant tok)
309{
310	struct man_node *p;
311
312	p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
313	if ( ! man_node_append(m, p))
314		return(0);
315	m->next = MAN_NEXT_CHILD;
316	return(1);
317}
318
319int
320man_word_alloc(struct man *m, int line, int pos, const char *word)
321{
322	struct man_node	*n;
323
324	n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
325	n->string = roff_strdup(m->roff, word);
326
327	if ( ! man_node_append(m, n))
328		return(0);
329
330	m->next = MAN_NEXT_SIBLING;
331	return(1);
332}
333
334
335/*
336 * Free all of the resources held by a node.  This does NOT unlink a
337 * node from its context; for that, see man_node_unlink().
338 */
339static void
340man_node_free(struct man_node *p)
341{
342
343	if (p->string)
344		free(p->string);
345	free(p);
346}
347
348
349void
350man_node_delete(struct man *m, struct man_node *p)
351{
352
353	while (p->child)
354		man_node_delete(m, p->child);
355
356	man_node_unlink(m, p);
357	man_node_free(p);
358}
359
360int
361man_addeqn(struct man *m, const struct eqn *ep)
362{
363	struct man_node	*n;
364
365	assert( ! (MAN_HALT & m->flags));
366
367	n = man_node_alloc(m, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
368	n->eqn = ep;
369
370	if ( ! man_node_append(m, n))
371		return(0);
372
373	m->next = MAN_NEXT_SIBLING;
374	return(man_descope(m, ep->ln, ep->pos));
375}
376
377int
378man_addspan(struct man *m, const struct tbl_span *sp)
379{
380	struct man_node	*n;
381
382	assert( ! (MAN_HALT & m->flags));
383
384	n = man_node_alloc(m, sp->line, 0, MAN_TBL, MAN_MAX);
385	n->span = sp;
386
387	if ( ! man_node_append(m, n))
388		return(0);
389
390	m->next = MAN_NEXT_SIBLING;
391	return(man_descope(m, sp->line, 0));
392}
393
394static int
395man_descope(struct man *m, int line, int offs)
396{
397	/*
398	 * Co-ordinate what happens with having a next-line scope open:
399	 * first close out the element scope (if applicable), then close
400	 * out the block scope (also if applicable).
401	 */
402
403	if (MAN_ELINE & m->flags) {
404		m->flags &= ~MAN_ELINE;
405		if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
406			return(0);
407	}
408
409	if ( ! (MAN_BLINE & m->flags))
410		return(1);
411	m->flags &= ~MAN_BLINE;
412
413	if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
414		return(0);
415	return(man_body_alloc(m, line, offs, m->last->tok));
416}
417
418static int
419man_ptext(struct man *m, int line, char *buf, int offs)
420{
421	int		 i;
422
423	/* Literal free-form text whitespace is preserved. */
424
425	if (MAN_LITERAL & m->flags) {
426		if ( ! man_word_alloc(m, line, offs, buf + offs))
427			return(0);
428		return(man_descope(m, line, offs));
429	}
430
431	/* Pump blank lines directly into the backend. */
432
433	for (i = offs; ' ' == buf[i]; i++)
434		/* Skip leading whitespace. */ ;
435
436	if ('\0' == buf[i]) {
437		/* Allocate a blank entry. */
438		if ( ! man_word_alloc(m, line, offs, ""))
439			return(0);
440		return(man_descope(m, line, offs));
441	}
442
443	/*
444	 * Warn if the last un-escaped character is whitespace. Then
445	 * strip away the remaining spaces (tabs stay!).
446	 */
447
448	i = (int)strlen(buf);
449	assert(i);
450
451	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
452		if (i > 1 && '\\' != buf[i - 2])
453			man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
454
455		for (--i; i && ' ' == buf[i]; i--)
456			/* Spin back to non-space. */ ;
457
458		/* Jump ahead of escaped whitespace. */
459		i += '\\' == buf[i] ? 2 : 1;
460
461		buf[i] = '\0';
462	}
463
464	if ( ! man_word_alloc(m, line, offs, buf + offs))
465		return(0);
466
467	/*
468	 * End-of-sentence check.  If the last character is an unescaped
469	 * EOS character, then flag the node as being the end of a
470	 * sentence.  The front-end will know how to interpret this.
471	 */
472
473	assert(i);
474	if (mandoc_eos(buf, (size_t)i, 0))
475		m->last->flags |= MAN_EOS;
476
477	return(man_descope(m, line, offs));
478}
479
480static int
481man_pmacro(struct man *m, int ln, char *buf, int offs)
482{
483	int		 i, ppos;
484	enum mant	 tok;
485	char		 mac[5];
486	struct man_node	*n;
487
488	if ('"' == buf[offs]) {
489		man_pmsg(m, ln, offs, MANDOCERR_BADCOMMENT);
490		return(1);
491	} else if ('\0' == buf[offs])
492		return(1);
493
494	ppos = offs;
495
496	/*
497	 * Copy the first word into a nil-terminated buffer.
498	 * Stop copying when a tab, space, or eoln is encountered.
499	 */
500
501	i = 0;
502	while (i < 4 && '\0' != buf[offs] &&
503			' ' != buf[offs] && '\t' != buf[offs])
504		mac[i++] = buf[offs++];
505
506	mac[i] = '\0';
507
508	tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
509
510	if (MAN_MAX == tok) {
511		mandoc_vmsg(MANDOCERR_MACRO, m->parse, ln,
512				ppos, "%s", buf + ppos - 1);
513		return(1);
514	}
515
516	/* The macro is sane.  Jump to the next word. */
517
518	while (buf[offs] && ' ' == buf[offs])
519		offs++;
520
521	/*
522	 * Trailing whitespace.  Note that tabs are allowed to be passed
523	 * into the parser as "text", so we only warn about spaces here.
524	 */
525
526	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
527		man_pmsg(m, ln, offs - 1, MANDOCERR_EOLNSPACE);
528
529	/*
530	 * Remove prior ELINE macro, as it's being clobbered by a new
531	 * macro.  Note that NSCOPED macros do not close out ELINE
532	 * macros---they don't print text---so we let those slip by.
533	 */
534
535	if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
536			m->flags & MAN_ELINE) {
537		n = m->last;
538		assert(MAN_TEXT != n->type);
539
540		/* Remove repeated NSCOPED macros causing ELINE. */
541
542		if (MAN_NSCOPED & man_macros[n->tok].flags)
543			n = n->parent;
544
545		mandoc_vmsg(MANDOCERR_LINESCOPE, m->parse, n->line,
546		    n->pos, "%s breaks %s", man_macronames[tok],
547		    man_macronames[n->tok]);
548
549		man_node_delete(m, n);
550		m->flags &= ~MAN_ELINE;
551	}
552
553	/*
554	 * Remove prior BLINE macro that is being clobbered.
555	 */
556	if ((m->flags & MAN_BLINE) &&
557	    (MAN_BSCOPE & man_macros[tok].flags)) {
558		n = m->last;
559
560		/* Might be a text node like 8 in
561		 * .TP 8
562		 * .SH foo
563		 */
564		if (MAN_TEXT == n->type)
565			n = n->parent;
566
567		/* Remove element that didn't end BLINE, if any. */
568		if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
569			n = n->parent;
570
571		assert(MAN_HEAD == n->type);
572		n = n->parent;
573		assert(MAN_BLOCK == n->type);
574		assert(MAN_SCOPED & man_macros[n->tok].flags);
575
576		mandoc_vmsg(MANDOCERR_LINESCOPE, m->parse, n->line,
577		    n->pos, "%s breaks %s", man_macronames[tok],
578		    man_macronames[n->tok]);
579
580		man_node_delete(m, n);
581		m->flags &= ~MAN_BLINE;
582	}
583
584	/*
585	 * Save the fact that we're in the next-line for a block.  In
586	 * this way, embedded roff instructions can "remember" state
587	 * when they exit.
588	 */
589
590	if (MAN_BLINE & m->flags)
591		m->flags |= MAN_BPLINE;
592
593	/* Call to handler... */
594
595	assert(man_macros[tok].fp);
596	if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &offs, buf))
597		goto err;
598
599	/*
600	 * We weren't in a block-line scope when entering the
601	 * above-parsed macro, so return.
602	 */
603
604	if ( ! (MAN_BPLINE & m->flags)) {
605		m->flags &= ~MAN_ILINE;
606		return(1);
607	}
608	m->flags &= ~MAN_BPLINE;
609
610	/*
611	 * If we're in a block scope, then allow this macro to slip by
612	 * without closing scope around it.
613	 */
614
615	if (MAN_ILINE & m->flags) {
616		m->flags &= ~MAN_ILINE;
617		return(1);
618	}
619
620	/*
621	 * If we've opened a new next-line element scope, then return
622	 * now, as the next line will close out the block scope.
623	 */
624
625	if (MAN_ELINE & m->flags)
626		return(1);
627
628	/* Close out the block scope opened in the prior line.  */
629
630	assert(MAN_BLINE & m->flags);
631	m->flags &= ~MAN_BLINE;
632
633	if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
634		return(0);
635	return(man_body_alloc(m, ln, ppos, m->last->tok));
636
637err:	/* Error out. */
638
639	m->flags |= MAN_HALT;
640	return(0);
641}
642
643/*
644 * Unlink a node from its context.  If "m" is provided, the last parse
645 * point will also be adjusted accordingly.
646 */
647static void
648man_node_unlink(struct man *m, struct man_node *n)
649{
650
651	/* Adjust siblings. */
652
653	if (n->prev)
654		n->prev->next = n->next;
655	if (n->next)
656		n->next->prev = n->prev;
657
658	/* Adjust parent. */
659
660	if (n->parent) {
661		n->parent->nchild--;
662		if (n->parent->child == n)
663			n->parent->child = n->prev ? n->prev : n->next;
664	}
665
666	/* Adjust parse point, if applicable. */
667
668	if (m && m->last == n) {
669		/*XXX: this can occur when bailing from validation. */
670		/*assert(NULL == n->next);*/
671		if (n->prev) {
672			m->last = n->prev;
673			m->next = MAN_NEXT_SIBLING;
674		} else {
675			m->last = n->parent;
676			m->next = MAN_NEXT_CHILD;
677		}
678	}
679
680	if (m && m->first == n)
681		m->first = NULL;
682}
683
684const struct mparse *
685man_mparse(const struct man *m)
686{
687
688	assert(m && m->parse);
689	return(m->parse);
690}
691