mdoc.c revision 1.122 
1/*	$OpenBSD: mdoc.c,v 1.122 2014/11/28 06:26:46 schwarze Exp $ */
2/*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012, 2013, 2014 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 AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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 <stdarg.h>
23#include <stdio.h>
24#include <stdlib.h>
25#include <string.h>
26#include <time.h>
27
28#include "mdoc.h"
29#include "mandoc.h"
30#include "mandoc_aux.h"
31#include "libmdoc.h"
32#include "libmandoc.h"
33
34const	char *const __mdoc_macronames[MDOC_MAX + 1] = {
35	"Ap",		"Dd",		"Dt",		"Os",
36	"Sh",		"Ss",		"Pp",		"D1",
37	"Dl",		"Bd",		"Ed",		"Bl",
38	"El",		"It",		"Ad",		"An",
39	"Ar",		"Cd",		"Cm",		"Dv",
40	"Er",		"Ev",		"Ex",		"Fa",
41	"Fd",		"Fl",		"Fn",		"Ft",
42	"Ic",		"In",		"Li",		"Nd",
43	"Nm",		"Op",		"Ot",		"Pa",
44	"Rv",		"St",		"Va",		"Vt",
45	"Xr",		"%A",		"%B",		"%D",
46	"%I",		"%J",		"%N",		"%O",
47	"%P",		"%R",		"%T",		"%V",
48	"Ac",		"Ao",		"Aq",		"At",
49	"Bc",		"Bf",		"Bo",		"Bq",
50	"Bsx",		"Bx",		"Db",		"Dc",
51	"Do",		"Dq",		"Ec",		"Ef",
52	"Em",		"Eo",		"Fx",		"Ms",
53	"No",		"Ns",		"Nx",		"Ox",
54	"Pc",		"Pf",		"Po",		"Pq",
55	"Qc",		"Ql",		"Qo",		"Qq",
56	"Re",		"Rs",		"Sc",		"So",
57	"Sq",		"Sm",		"Sx",		"Sy",
58	"Tn",		"Ux",		"Xc",		"Xo",
59	"Fo",		"Fc",		"Oo",		"Oc",
60	"Bk",		"Ek",		"Bt",		"Hf",
61	"Fr",		"Ud",		"Lb",		"Lp",
62	"Lk",		"Mt",		"Brq",		"Bro",
63	"Brc",		"%C",		"Es",		"En",
64	"Dx",		"%Q",		"br",		"sp",
65	"%U",		"Ta",		"ll",		"text",
66	};
67
68const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {
69	"split",		"nosplit",		"ragged",
70	"unfilled",		"literal",		"file",
71	"offset",		"bullet",		"dash",
72	"hyphen",		"item",			"enum",
73	"tag",			"diag",			"hang",
74	"ohang",		"inset",		"column",
75	"width",		"compact",		"std",
76	"filled",		"words",		"emphasis",
77	"symbolic",		"nested",		"centered"
78	};
79
80const	char * const *mdoc_macronames = __mdoc_macronames;
81const	char * const *mdoc_argnames = __mdoc_argnames;
82
83static	void		  mdoc_node_free(struct mdoc_node *);
84static	void		  mdoc_node_unlink(struct mdoc *,
85				struct mdoc_node *);
86static	void		  mdoc_free1(struct mdoc *);
87static	void		  mdoc_alloc1(struct mdoc *);
88static	struct mdoc_node *node_alloc(struct mdoc *, int, int,
89				enum mdoct, enum mdoc_type);
90static	void		  node_append(struct mdoc *, struct mdoc_node *);
91static	int		  mdoc_ptext(struct mdoc *, int, char *, int);
92static	int		  mdoc_pmacro(struct mdoc *, int, char *, int);
93
94
95const struct mdoc_node *
96mdoc_node(const struct mdoc *mdoc)
97{
98
99	return(mdoc->first);
100}
101
102const struct mdoc_meta *
103mdoc_meta(const struct mdoc *mdoc)
104{
105
106	return(&mdoc->meta);
107}
108
109/*
110 * Frees volatile resources (parse tree, meta-data, fields).
111 */
112static void
113mdoc_free1(struct mdoc *mdoc)
114{
115
116	if (mdoc->first)
117		mdoc_node_delete(mdoc, mdoc->first);
118	free(mdoc->meta.msec);
119	free(mdoc->meta.vol);
120	free(mdoc->meta.arch);
121	free(mdoc->meta.date);
122	free(mdoc->meta.title);
123	free(mdoc->meta.os);
124	free(mdoc->meta.name);
125}
126
127/*
128 * Allocate all volatile resources (parse tree, meta-data, fields).
129 */
130static void
131mdoc_alloc1(struct mdoc *mdoc)
132{
133
134	memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
135	mdoc->flags = 0;
136	mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
137	mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
138	mdoc->first = mdoc->last;
139	mdoc->last->type = MDOC_ROOT;
140	mdoc->last->tok = MDOC_MAX;
141	mdoc->next = MDOC_NEXT_CHILD;
142}
143
144/*
145 * Free up volatile resources (see mdoc_free1()) then re-initialises the
146 * data with mdoc_alloc1().  After invocation, parse data has been reset
147 * and the parser is ready for re-invocation on a new tree; however,
148 * cross-parse non-volatile data is kept intact.
149 */
150void
151mdoc_reset(struct mdoc *mdoc)
152{
153
154	mdoc_free1(mdoc);
155	mdoc_alloc1(mdoc);
156}
157
158/*
159 * Completely free up all volatile and non-volatile parse resources.
160 * After invocation, the pointer is no longer usable.
161 */
162void
163mdoc_free(struct mdoc *mdoc)
164{
165
166	mdoc_free1(mdoc);
167	free(mdoc);
168}
169
170/*
171 * Allocate volatile and non-volatile parse resources.
172 */
173struct mdoc *
174mdoc_alloc(struct roff *roff, struct mparse *parse,
175	const char *defos, int quick)
176{
177	struct mdoc	*p;
178
179	p = mandoc_calloc(1, sizeof(struct mdoc));
180
181	p->parse = parse;
182	p->defos = defos;
183	p->quick = quick;
184	p->roff = roff;
185
186	mdoc_hash_init();
187	mdoc_alloc1(p);
188	return(p);
189}
190
191int
192mdoc_endparse(struct mdoc *mdoc)
193{
194
195	mdoc_macroend(mdoc);
196	return(1);
197}
198
199void
200mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
201{
202	struct mdoc_node *n;
203
204	n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
205	n->eqn = ep;
206	if (ep->ln > mdoc->last->line)
207		n->flags |= MDOC_LINE;
208	node_append(mdoc, n);
209	mdoc->next = MDOC_NEXT_SIBLING;
210}
211
212void
213mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
214{
215	struct mdoc_node *n;
216
217	n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
218	n->span = sp;
219	node_append(mdoc, n);
220	mdoc->next = MDOC_NEXT_SIBLING;
221}
222
223/*
224 * Main parse routine.  Parses a single line -- really just hands off to
225 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
226 */
227int
228mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
229{
230
231	if (mdoc->last->type != MDOC_EQN || ln > mdoc->last->line)
232		mdoc->flags |= MDOC_NEWLINE;
233
234	/*
235	 * Let the roff nS register switch SYNOPSIS mode early,
236	 * such that the parser knows at all times
237	 * whether this mode is on or off.
238	 * Note that this mode is also switched by the Sh macro.
239	 */
240	if (roff_getreg(mdoc->roff, "nS"))
241		mdoc->flags |= MDOC_SYNOPSIS;
242	else
243		mdoc->flags &= ~MDOC_SYNOPSIS;
244
245	return(roff_getcontrol(mdoc->roff, buf, &offs) ?
246	    mdoc_pmacro(mdoc, ln, buf, offs) :
247	    mdoc_ptext(mdoc, ln, buf, offs));
248}
249
250void
251mdoc_macro(MACRO_PROT_ARGS)
252{
253	assert(tok < MDOC_MAX);
254
255	if (mdoc->flags & MDOC_PBODY) {
256		if (tok == MDOC_Dt) {
257			mandoc_vmsg(MANDOCERR_DT_LATE,
258			    mdoc->parse, line, ppos,
259			    "Dt %s", buf + *pos);
260			return;
261		}
262	} else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
263		if (mdoc->meta.title == NULL) {
264			mandoc_vmsg(MANDOCERR_DT_NOTITLE,
265			    mdoc->parse, line, ppos, "%s %s",
266			    mdoc_macronames[tok], buf + *pos);
267			mdoc->meta.title = mandoc_strdup("UNTITLED");
268		}
269		if (NULL == mdoc->meta.vol)
270			mdoc->meta.vol = mandoc_strdup("LOCAL");
271		mdoc->flags |= MDOC_PBODY;
272	}
273	(*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
274}
275
276
277static void
278node_append(struct mdoc *mdoc, struct mdoc_node *p)
279{
280
281	assert(mdoc->last);
282	assert(mdoc->first);
283	assert(MDOC_ROOT != p->type);
284
285	switch (mdoc->next) {
286	case MDOC_NEXT_SIBLING:
287		mdoc->last->next = p;
288		p->prev = mdoc->last;
289		p->parent = mdoc->last->parent;
290		break;
291	case MDOC_NEXT_CHILD:
292		mdoc->last->child = p;
293		p->parent = mdoc->last;
294		break;
295	default:
296		abort();
297		/* NOTREACHED */
298	}
299
300	p->parent->nchild++;
301
302	/*
303	 * Copy over the normalised-data pointer of our parent.  Not
304	 * everybody has one, but copying a null pointer is fine.
305	 */
306
307	switch (p->type) {
308	case MDOC_BODY:
309		if (ENDBODY_NOT != p->end)
310			break;
311		/* FALLTHROUGH */
312	case MDOC_TAIL:
313		/* FALLTHROUGH */
314	case MDOC_HEAD:
315		p->norm = p->parent->norm;
316		break;
317	default:
318		break;
319	}
320
321	mdoc_valid_pre(mdoc, p);
322
323	switch (p->type) {
324	case MDOC_HEAD:
325		assert(MDOC_BLOCK == p->parent->type);
326		p->parent->head = p;
327		break;
328	case MDOC_TAIL:
329		assert(MDOC_BLOCK == p->parent->type);
330		p->parent->tail = p;
331		break;
332	case MDOC_BODY:
333		if (p->end)
334			break;
335		assert(MDOC_BLOCK == p->parent->type);
336		p->parent->body = p;
337		break;
338	default:
339		break;
340	}
341
342	mdoc->last = p;
343
344	switch (p->type) {
345	case MDOC_TBL:
346		/* FALLTHROUGH */
347	case MDOC_TEXT:
348		mdoc_valid_post(mdoc);
349		break;
350	default:
351		break;
352	}
353}
354
355static struct mdoc_node *
356node_alloc(struct mdoc *mdoc, int line, int pos,
357		enum mdoct tok, enum mdoc_type type)
358{
359	struct mdoc_node *p;
360
361	p = mandoc_calloc(1, sizeof(struct mdoc_node));
362	p->sec = mdoc->lastsec;
363	p->line = line;
364	p->pos = pos;
365	p->lastline = line;
366	p->tok = tok;
367	p->type = type;
368
369	/* Flag analysis. */
370
371	if (MDOC_SYNOPSIS & mdoc->flags)
372		p->flags |= MDOC_SYNPRETTY;
373	else
374		p->flags &= ~MDOC_SYNPRETTY;
375	if (MDOC_NEWLINE & mdoc->flags)
376		p->flags |= MDOC_LINE;
377	mdoc->flags &= ~MDOC_NEWLINE;
378
379	return(p);
380}
381
382void
383mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
384{
385	struct mdoc_node *p;
386
387	p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
388	node_append(mdoc, p);
389	mdoc->next = MDOC_NEXT_CHILD;
390}
391
392struct mdoc_node *
393mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
394{
395	struct mdoc_node *p;
396
397	assert(mdoc->first);
398	assert(mdoc->last);
399	p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
400	node_append(mdoc, p);
401	mdoc->next = MDOC_NEXT_CHILD;
402	return(p);
403}
404
405struct mdoc_node *
406mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
407{
408	struct mdoc_node *p;
409
410	p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
411	node_append(mdoc, p);
412	mdoc->next = MDOC_NEXT_CHILD;
413	return(p);
414}
415
416void
417mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
418		struct mdoc_node *body, enum mdoc_endbody end)
419{
420	struct mdoc_node *p;
421
422	p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
423	p->pending = body;
424	p->norm = body->norm;
425	p->end = end;
426	node_append(mdoc, p);
427	mdoc->next = MDOC_NEXT_SIBLING;
428}
429
430struct mdoc_node *
431mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
432		enum mdoct tok, struct mdoc_arg *args)
433{
434	struct mdoc_node *p;
435
436	p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
437	p->args = args;
438	if (p->args)
439		(args->refcnt)++;
440
441	switch (tok) {
442	case MDOC_Bd:
443		/* FALLTHROUGH */
444	case MDOC_Bf:
445		/* FALLTHROUGH */
446	case MDOC_Bl:
447		/* FALLTHROUGH */
448	case MDOC_En:
449		/* FALLTHROUGH */
450	case MDOC_Rs:
451		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
452		break;
453	default:
454		break;
455	}
456	node_append(mdoc, p);
457	mdoc->next = MDOC_NEXT_CHILD;
458	return(p);
459}
460
461void
462mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
463		enum mdoct tok, struct mdoc_arg *args)
464{
465	struct mdoc_node *p;
466
467	p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
468	p->args = args;
469	if (p->args)
470		(args->refcnt)++;
471
472	switch (tok) {
473	case MDOC_An:
474		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
475		break;
476	default:
477		break;
478	}
479	node_append(mdoc, p);
480	mdoc->next = MDOC_NEXT_CHILD;
481}
482
483void
484mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
485{
486	struct mdoc_node *n;
487
488	n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
489	n->string = roff_strdup(mdoc->roff, p);
490	node_append(mdoc, n);
491	mdoc->next = MDOC_NEXT_SIBLING;
492}
493
494void
495mdoc_word_append(struct mdoc *mdoc, const char *p)
496{
497	struct mdoc_node	*n;
498	char			*addstr, *newstr;
499
500	n = mdoc->last;
501	addstr = roff_strdup(mdoc->roff, p);
502	mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
503	free(addstr);
504	free(n->string);
505	n->string = newstr;
506	mdoc->next = MDOC_NEXT_SIBLING;
507}
508
509static void
510mdoc_node_free(struct mdoc_node *p)
511{
512
513	if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
514		free(p->norm);
515	if (p->string)
516		free(p->string);
517	if (p->args)
518		mdoc_argv_free(p->args);
519	free(p);
520}
521
522static void
523mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
524{
525
526	/* Adjust siblings. */
527
528	if (n->prev)
529		n->prev->next = n->next;
530	if (n->next)
531		n->next->prev = n->prev;
532
533	/* Adjust parent. */
534
535	if (n->parent) {
536		n->parent->nchild--;
537		if (n->parent->child == n)
538			n->parent->child = n->prev ? n->prev : n->next;
539		if (n->parent->last == n)
540			n->parent->last = n->prev ? n->prev : NULL;
541	}
542
543	/* Adjust parse point, if applicable. */
544
545	if (mdoc && mdoc->last == n) {
546		if (n->prev) {
547			mdoc->last = n->prev;
548			mdoc->next = MDOC_NEXT_SIBLING;
549		} else {
550			mdoc->last = n->parent;
551			mdoc->next = MDOC_NEXT_CHILD;
552		}
553	}
554
555	if (mdoc && mdoc->first == n)
556		mdoc->first = NULL;
557}
558
559void
560mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
561{
562
563	while (p->child) {
564		assert(p->nchild);
565		mdoc_node_delete(mdoc, p->child);
566	}
567	assert(0 == p->nchild);
568
569	mdoc_node_unlink(mdoc, p);
570	mdoc_node_free(p);
571}
572
573void
574mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
575{
576
577	mdoc_node_unlink(mdoc, p);
578	node_append(mdoc, p);
579}
580
581/*
582 * Parse free-form text, that is, a line that does not begin with the
583 * control character.
584 */
585static int
586mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
587{
588	char		 *c, *ws, *end;
589	struct mdoc_node *n;
590
591	assert(mdoc->last);
592	n = mdoc->last;
593
594	/*
595	 * Divert directly to list processing if we're encountering a
596	 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
597	 * (a MDOC_BODY means it's already open, in which case we should
598	 * process within its context in the normal way).
599	 */
600
601	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
602	    LIST_column == n->norm->Bl.type) {
603		/* `Bl' is open without any children. */
604		mdoc->flags |= MDOC_FREECOL;
605		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
606		return(1);
607	}
608
609	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
610	    NULL != n->parent &&
611	    MDOC_Bl == n->parent->tok &&
612	    LIST_column == n->parent->norm->Bl.type) {
613		/* `Bl' has block-level `It' children. */
614		mdoc->flags |= MDOC_FREECOL;
615		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
616		return(1);
617	}
618
619	/*
620	 * Search for the beginning of unescaped trailing whitespace (ws)
621	 * and for the first character not to be output (end).
622	 */
623
624	/* FIXME: replace with strcspn(). */
625	ws = NULL;
626	for (c = end = buf + offs; *c; c++) {
627		switch (*c) {
628		case ' ':
629			if (NULL == ws)
630				ws = c;
631			continue;
632		case '\t':
633			/*
634			 * Always warn about trailing tabs,
635			 * even outside literal context,
636			 * where they should be put on the next line.
637			 */
638			if (NULL == ws)
639				ws = c;
640			/*
641			 * Strip trailing tabs in literal context only;
642			 * outside, they affect the next line.
643			 */
644			if (MDOC_LITERAL & mdoc->flags)
645				continue;
646			break;
647		case '\\':
648			/* Skip the escaped character, too, if any. */
649			if (c[1])
650				c++;
651			/* FALLTHROUGH */
652		default:
653			ws = NULL;
654			break;
655		}
656		end = c + 1;
657	}
658	*end = '\0';
659
660	if (ws)
661		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
662		    line, (int)(ws-buf), NULL);
663
664	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
665		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
666		    line, (int)(c - buf), NULL);
667
668		/*
669		 * Insert a `sp' in the case of a blank line.  Technically,
670		 * blank lines aren't allowed, but enough manuals assume this
671		 * behaviour that we want to work around it.
672		 */
673		mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL);
674		mdoc->next = MDOC_NEXT_SIBLING;
675		mdoc_valid_post(mdoc);
676		return(1);
677	}
678
679	mdoc_word_alloc(mdoc, line, offs, buf+offs);
680
681	if (mdoc->flags & MDOC_LITERAL)
682		return(1);
683
684	/*
685	 * End-of-sentence check.  If the last character is an unescaped
686	 * EOS character, then flag the node as being the end of a
687	 * sentence.  The front-end will know how to interpret this.
688	 */
689
690	assert(buf < end);
691
692	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
693		mdoc->last->flags |= MDOC_EOS;
694	return(1);
695}
696
697/*
698 * Parse a macro line, that is, a line beginning with the control
699 * character.
700 */
701static int
702mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
703{
704	struct mdoc_node *n;
705	const char	 *cp;
706	enum mdoct	  tok;
707	int		  i, sv;
708	char		  mac[5];
709
710	sv = offs;
711
712	/*
713	 * Copy the first word into a nil-terminated buffer.
714	 * Stop when a space, tab, escape, or eoln is encountered.
715	 */
716
717	i = 0;
718	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
719		mac[i++] = buf[offs++];
720
721	mac[i] = '\0';
722
723	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
724
725	if (tok == MDOC_MAX) {
726		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
727		    ln, sv, buf + sv - 1);
728		return(1);
729	}
730
731	/* Skip a leading escape sequence or tab. */
732
733	switch (buf[offs]) {
734	case '\\':
735		cp = buf + offs + 1;
736		mandoc_escape(&cp, NULL, NULL);
737		offs = cp - buf;
738		break;
739	case '\t':
740		offs++;
741		break;
742	default:
743		break;
744	}
745
746	/* Jump to the next non-whitespace word. */
747
748	while (buf[offs] && ' ' == buf[offs])
749		offs++;
750
751	/*
752	 * Trailing whitespace.  Note that tabs are allowed to be passed
753	 * into the parser as "text", so we only warn about spaces here.
754	 */
755
756	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
757		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
758		    ln, offs - 1, NULL);
759
760	/*
761	 * If an initial macro or a list invocation, divert directly
762	 * into macro processing.
763	 */
764
765	if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
766		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
767		return(1);
768	}
769
770	n = mdoc->last;
771	assert(mdoc->last);
772
773	/*
774	 * If the first macro of a `Bl -column', open an `It' block
775	 * context around the parsed macro.
776	 */
777
778	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
779	    LIST_column == n->norm->Bl.type) {
780		mdoc->flags |= MDOC_FREECOL;
781		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
782		return(1);
783	}
784
785	/*
786	 * If we're following a block-level `It' within a `Bl -column'
787	 * context (perhaps opened in the above block or in ptext()),
788	 * then open an `It' block context around the parsed macro.
789	 */
790
791	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
792	    NULL != n->parent &&
793	    MDOC_Bl == n->parent->tok &&
794	    LIST_column == n->parent->norm->Bl.type) {
795		mdoc->flags |= MDOC_FREECOL;
796		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
797		return(1);
798	}
799
800	/* Normal processing of a macro. */
801
802	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
803
804	/* In quick mode (for mandocdb), abort after the NAME section. */
805
806	if (mdoc->quick && MDOC_Sh == tok &&
807	    SEC_NAME != mdoc->last->sec)
808		return(2);
809
810	return(1);
811}
812
813enum mdelim
814mdoc_isdelim(const char *p)
815{
816
817	if ('\0' == p[0])
818		return(DELIM_NONE);
819
820	if ('\0' == p[1])
821		switch (p[0]) {
822		case '(':
823			/* FALLTHROUGH */
824		case '[':
825			return(DELIM_OPEN);
826		case '|':
827			return(DELIM_MIDDLE);
828		case '.':
829			/* FALLTHROUGH */
830		case ',':
831			/* FALLTHROUGH */
832		case ';':
833			/* FALLTHROUGH */
834		case ':':
835			/* FALLTHROUGH */
836		case '?':
837			/* FALLTHROUGH */
838		case '!':
839			/* FALLTHROUGH */
840		case ')':
841			/* FALLTHROUGH */
842		case ']':
843			return(DELIM_CLOSE);
844		default:
845			return(DELIM_NONE);
846		}
847
848	if ('\\' != p[0])
849		return(DELIM_NONE);
850
851	if (0 == strcmp(p + 1, "."))
852		return(DELIM_CLOSE);
853	if (0 == strcmp(p + 1, "fR|\\fP"))
854		return(DELIM_MIDDLE);
855
856	return(DELIM_NONE);
857}
858
859void
860mdoc_deroff(char **dest, const struct mdoc_node *n)
861{
862	char	*cp;
863	size_t	 sz;
864
865	if (MDOC_TEXT != n->type) {
866		for (n = n->child; n; n = n->next)
867			mdoc_deroff(dest, n);
868		return;
869	}
870
871	/* Skip leading whitespace. */
872
873	for (cp = n->string; '\0' != *cp; cp++)
874		if (0 == isspace((unsigned char)*cp))
875			break;
876
877	/* Skip trailing whitespace. */
878
879	for (sz = strlen(cp); sz; sz--)
880		if (0 == isspace((unsigned char)cp[sz-1]))
881			break;
882
883	/* Skip empty strings. */
884
885	if (0 == sz)
886		return;
887
888	if (NULL == *dest) {
889		*dest = mandoc_strndup(cp, sz);
890		return;
891	}
892
893	mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
894	free(*dest);
895	*dest = cp;
896}
897