mdoc.c revision 1.111 
1/*	$Id: mdoc.c,v 1.111 2014/08/08 15:32:17 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	int		  node_append(struct mdoc *,
91				struct mdoc_node *);
92#if 0
93static	int		  mdoc_preptext(struct mdoc *, int, char *, int);
94#endif
95static	int		  mdoc_ptext(struct mdoc *, int, char *, int);
96static	int		  mdoc_pmacro(struct mdoc *, int, char *, int);
97
98
99const struct mdoc_node *
100mdoc_node(const struct mdoc *mdoc)
101{
102
103	return(mdoc->first);
104}
105
106const struct mdoc_meta *
107mdoc_meta(const struct mdoc *mdoc)
108{
109
110	return(&mdoc->meta);
111}
112
113/*
114 * Frees volatile resources (parse tree, meta-data, fields).
115 */
116static void
117mdoc_free1(struct mdoc *mdoc)
118{
119
120	if (mdoc->first)
121		mdoc_node_delete(mdoc, mdoc->first);
122	if (mdoc->meta.title)
123		free(mdoc->meta.title);
124	if (mdoc->meta.os)
125		free(mdoc->meta.os);
126	if (mdoc->meta.name)
127		free(mdoc->meta.name);
128	if (mdoc->meta.arch)
129		free(mdoc->meta.arch);
130	if (mdoc->meta.vol)
131		free(mdoc->meta.vol);
132	if (mdoc->meta.msec)
133		free(mdoc->meta.msec);
134	if (mdoc->meta.date)
135		free(mdoc->meta.date);
136}
137
138/*
139 * Allocate all volatile resources (parse tree, meta-data, fields).
140 */
141static void
142mdoc_alloc1(struct mdoc *mdoc)
143{
144
145	memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
146	mdoc->flags = 0;
147	mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
148	mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
149	mdoc->first = mdoc->last;
150	mdoc->last->type = MDOC_ROOT;
151	mdoc->last->tok = MDOC_MAX;
152	mdoc->next = MDOC_NEXT_CHILD;
153}
154
155/*
156 * Free up volatile resources (see mdoc_free1()) then re-initialises the
157 * data with mdoc_alloc1().  After invocation, parse data has been reset
158 * and the parser is ready for re-invocation on a new tree; however,
159 * cross-parse non-volatile data is kept intact.
160 */
161void
162mdoc_reset(struct mdoc *mdoc)
163{
164
165	mdoc_free1(mdoc);
166	mdoc_alloc1(mdoc);
167}
168
169/*
170 * Completely free up all volatile and non-volatile parse resources.
171 * After invocation, the pointer is no longer usable.
172 */
173void
174mdoc_free(struct mdoc *mdoc)
175{
176
177	mdoc_free1(mdoc);
178	free(mdoc);
179}
180
181/*
182 * Allocate volatile and non-volatile parse resources.
183 */
184struct mdoc *
185mdoc_alloc(struct roff *roff, struct mparse *parse,
186	const char *defos, int quick)
187{
188	struct mdoc	*p;
189
190	p = mandoc_calloc(1, sizeof(struct mdoc));
191
192	p->parse = parse;
193	p->defos = defos;
194	p->quick = quick;
195	p->roff = roff;
196
197	mdoc_hash_init();
198	mdoc_alloc1(p);
199	return(p);
200}
201
202int
203mdoc_endparse(struct mdoc *mdoc)
204{
205
206	return(mdoc_macroend(mdoc));
207}
208
209int
210mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
211{
212	struct mdoc_node *n;
213
214	n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
215	n->eqn = ep;
216
217	if ( ! node_append(mdoc, n))
218		return(0);
219
220	mdoc->next = MDOC_NEXT_SIBLING;
221	return(1);
222}
223
224int
225mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
226{
227	struct mdoc_node *n;
228
229	n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
230	n->span = sp;
231
232	if ( ! node_append(mdoc, n))
233		return(0);
234
235	mdoc->next = MDOC_NEXT_SIBLING;
236	return(1);
237}
238
239/*
240 * Main parse routine.  Parses a single line -- really just hands off to
241 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
242 */
243int
244mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
245{
246
247	mdoc->flags |= MDOC_NEWLINE;
248
249	/*
250	 * Let the roff nS register switch SYNOPSIS mode early,
251	 * such that the parser knows at all times
252	 * whether this mode is on or off.
253	 * Note that this mode is also switched by the Sh macro.
254	 */
255	if (roff_getreg(mdoc->roff, "nS"))
256		mdoc->flags |= MDOC_SYNOPSIS;
257	else
258		mdoc->flags &= ~MDOC_SYNOPSIS;
259
260	return(roff_getcontrol(mdoc->roff, buf, &offs) ?
261	    mdoc_pmacro(mdoc, ln, buf, offs) :
262	    mdoc_ptext(mdoc, ln, buf, offs));
263}
264
265int
266mdoc_macro(MACRO_PROT_ARGS)
267{
268	assert(tok < MDOC_MAX);
269
270	/* If we're in the body, deny prologue calls. */
271
272	if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
273	    MDOC_PBODY & mdoc->flags) {
274		mandoc_vmsg(MANDOCERR_PROLOG_ONLY, mdoc->parse,
275		    line, ppos, "%s", mdoc_macronames[tok]);
276		return(1);
277	}
278
279	/* If we're in the prologue, deny "body" macros.  */
280
281	if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
282	     ! (MDOC_PBODY & mdoc->flags)) {
283		mandoc_vmsg(MANDOCERR_PROLOG_BAD, mdoc->parse,
284		    line, ppos, "%s", mdoc_macronames[tok]);
285		if (NULL == mdoc->meta.msec)
286			mdoc->meta.msec = mandoc_strdup("1");
287		if (NULL == mdoc->meta.title)
288			mdoc->meta.title = mandoc_strdup("UNKNOWN");
289		if (NULL == mdoc->meta.vol)
290			mdoc->meta.vol = mandoc_strdup("LOCAL");
291		if (NULL == mdoc->meta.os)
292			mdoc->meta.os = mandoc_strdup("LOCAL");
293		if (NULL == mdoc->meta.date)
294			mdoc->meta.date = mandoc_normdate
295				(mdoc->parse, NULL, line, ppos);
296		mdoc->flags |= MDOC_PBODY;
297	}
298
299	return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
300}
301
302
303static int
304node_append(struct mdoc *mdoc, struct mdoc_node *p)
305{
306
307	assert(mdoc->last);
308	assert(mdoc->first);
309	assert(MDOC_ROOT != p->type);
310
311	switch (mdoc->next) {
312	case MDOC_NEXT_SIBLING:
313		mdoc->last->next = p;
314		p->prev = mdoc->last;
315		p->parent = mdoc->last->parent;
316		break;
317	case MDOC_NEXT_CHILD:
318		mdoc->last->child = p;
319		p->parent = mdoc->last;
320		break;
321	default:
322		abort();
323		/* NOTREACHED */
324	}
325
326	p->parent->nchild++;
327
328	/*
329	 * Copy over the normalised-data pointer of our parent.  Not
330	 * everybody has one, but copying a null pointer is fine.
331	 */
332
333	switch (p->type) {
334	case MDOC_BODY:
335		if (ENDBODY_NOT != p->end)
336			break;
337		/* FALLTHROUGH */
338	case MDOC_TAIL:
339		/* FALLTHROUGH */
340	case MDOC_HEAD:
341		p->norm = p->parent->norm;
342		break;
343	default:
344		break;
345	}
346
347	if ( ! mdoc_valid_pre(mdoc, p))
348		return(0);
349
350	switch (p->type) {
351	case MDOC_HEAD:
352		assert(MDOC_BLOCK == p->parent->type);
353		p->parent->head = p;
354		break;
355	case MDOC_TAIL:
356		assert(MDOC_BLOCK == p->parent->type);
357		p->parent->tail = p;
358		break;
359	case MDOC_BODY:
360		if (p->end)
361			break;
362		assert(MDOC_BLOCK == p->parent->type);
363		p->parent->body = p;
364		break;
365	default:
366		break;
367	}
368
369	mdoc->last = p;
370
371	switch (p->type) {
372	case MDOC_TBL:
373		/* FALLTHROUGH */
374	case MDOC_TEXT:
375		if ( ! mdoc_valid_post(mdoc))
376			return(0);
377		break;
378	default:
379		break;
380	}
381
382	return(1);
383}
384
385static struct mdoc_node *
386node_alloc(struct mdoc *mdoc, int line, int pos,
387		enum mdoct tok, enum mdoc_type type)
388{
389	struct mdoc_node *p;
390
391	p = mandoc_calloc(1, sizeof(struct mdoc_node));
392	p->sec = mdoc->lastsec;
393	p->line = line;
394	p->pos = pos;
395	p->lastline = line;
396	p->tok = tok;
397	p->type = type;
398
399	/* Flag analysis. */
400
401	if (MDOC_SYNOPSIS & mdoc->flags)
402		p->flags |= MDOC_SYNPRETTY;
403	else
404		p->flags &= ~MDOC_SYNPRETTY;
405	if (MDOC_NEWLINE & mdoc->flags)
406		p->flags |= MDOC_LINE;
407	mdoc->flags &= ~MDOC_NEWLINE;
408
409	return(p);
410}
411
412int
413mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
414{
415	struct mdoc_node *p;
416
417	p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
418	if ( ! node_append(mdoc, p))
419		return(0);
420	mdoc->next = MDOC_NEXT_CHILD;
421	return(1);
422}
423
424int
425mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
426{
427	struct mdoc_node *p;
428
429	assert(mdoc->first);
430	assert(mdoc->last);
431
432	p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
433	if ( ! node_append(mdoc, p))
434		return(0);
435	mdoc->next = MDOC_NEXT_CHILD;
436	return(1);
437}
438
439int
440mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
441{
442	struct mdoc_node *p;
443
444	p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
445	if ( ! node_append(mdoc, p))
446		return(0);
447	mdoc->next = MDOC_NEXT_CHILD;
448	return(1);
449}
450
451int
452mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
453		struct mdoc_node *body, enum mdoc_endbody end)
454{
455	struct mdoc_node *p;
456
457	p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
458	p->pending = body;
459	p->norm = body->norm;
460	p->end = end;
461	if ( ! node_append(mdoc, p))
462		return(0);
463	mdoc->next = MDOC_NEXT_SIBLING;
464	return(1);
465}
466
467int
468mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
469		enum mdoct tok, struct mdoc_arg *args)
470{
471	struct mdoc_node *p;
472
473	p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
474	p->args = args;
475	if (p->args)
476		(args->refcnt)++;
477
478	switch (tok) {
479	case MDOC_Bd:
480		/* FALLTHROUGH */
481	case MDOC_Bf:
482		/* FALLTHROUGH */
483	case MDOC_Bl:
484		/* FALLTHROUGH */
485	case MDOC_En:
486		/* FALLTHROUGH */
487	case MDOC_Rs:
488		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
489		break;
490	default:
491		break;
492	}
493
494	if ( ! node_append(mdoc, p))
495		return(0);
496	mdoc->next = MDOC_NEXT_CHILD;
497	return(1);
498}
499
500int
501mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
502		enum mdoct tok, struct mdoc_arg *args)
503{
504	struct mdoc_node *p;
505
506	p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
507	p->args = args;
508	if (p->args)
509		(args->refcnt)++;
510
511	switch (tok) {
512	case MDOC_An:
513		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
514		break;
515	default:
516		break;
517	}
518
519	if ( ! node_append(mdoc, p))
520		return(0);
521	mdoc->next = MDOC_NEXT_CHILD;
522	return(1);
523}
524
525int
526mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
527{
528	struct mdoc_node *n;
529
530	n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
531	n->string = roff_strdup(mdoc->roff, p);
532
533	if ( ! node_append(mdoc, n))
534		return(0);
535
536	mdoc->next = MDOC_NEXT_SIBLING;
537	return(1);
538}
539
540void
541mdoc_word_append(struct mdoc *mdoc, const char *p)
542{
543	struct mdoc_node	*n;
544	char			*addstr, *newstr;
545
546	n = mdoc->last;
547	addstr = roff_strdup(mdoc->roff, p);
548	mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
549	free(addstr);
550	free(n->string);
551	n->string = newstr;
552	mdoc->next = MDOC_NEXT_SIBLING;
553}
554
555static void
556mdoc_node_free(struct mdoc_node *p)
557{
558
559	if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
560		free(p->norm);
561	if (p->string)
562		free(p->string);
563	if (p->args)
564		mdoc_argv_free(p->args);
565	free(p);
566}
567
568static void
569mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
570{
571
572	/* Adjust siblings. */
573
574	if (n->prev)
575		n->prev->next = n->next;
576	if (n->next)
577		n->next->prev = n->prev;
578
579	/* Adjust parent. */
580
581	if (n->parent) {
582		n->parent->nchild--;
583		if (n->parent->child == n)
584			n->parent->child = n->prev ? n->prev : n->next;
585		if (n->parent->last == n)
586			n->parent->last = n->prev ? n->prev : NULL;
587	}
588
589	/* Adjust parse point, if applicable. */
590
591	if (mdoc && mdoc->last == n) {
592		if (n->prev) {
593			mdoc->last = n->prev;
594			mdoc->next = MDOC_NEXT_SIBLING;
595		} else {
596			mdoc->last = n->parent;
597			mdoc->next = MDOC_NEXT_CHILD;
598		}
599	}
600
601	if (mdoc && mdoc->first == n)
602		mdoc->first = NULL;
603}
604
605void
606mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
607{
608
609	while (p->child) {
610		assert(p->nchild);
611		mdoc_node_delete(mdoc, p->child);
612	}
613	assert(0 == p->nchild);
614
615	mdoc_node_unlink(mdoc, p);
616	mdoc_node_free(p);
617}
618
619int
620mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
621{
622
623	mdoc_node_unlink(mdoc, p);
624	return(node_append(mdoc, p));
625}
626
627#if 0
628/*
629 * Pre-treat a text line.
630 * Text lines can consist of equations, which must be handled apart from
631 * the regular text.
632 * Thus, use this function to step through a line checking if it has any
633 * equations embedded in it.
634 * This must handle multiple equations AND equations that do not end at
635 * the end-of-line, i.e., will re-enter in the next roff parse.
636 */
637static int
638mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
639{
640	char		*start, *end;
641	char		 delim;
642
643	while ('\0' != buf[offs]) {
644		/* Mark starting position if eqn is set. */
645		start = NULL;
646		if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
647			if (NULL != (start = strchr(buf + offs, delim)))
648				*start++ = '\0';
649
650		/* Parse text as normal. */
651		if ( ! mdoc_ptext(mdoc, line, buf, offs))
652			return(0);
653
654		/* Continue only if an equation exists. */
655		if (NULL == start)
656			break;
657
658		/* Read past the end of the equation. */
659		offs += start - (buf + offs);
660		assert(start == &buf[offs]);
661		if (NULL != (end = strchr(buf + offs, delim))) {
662			*end++ = '\0';
663			while (' ' == *end)
664				end++;
665		}
666
667		/* Parse the equation itself. */
668		roff_openeqn(mdoc->roff, NULL, line, offs, buf);
669
670		/* Process a finished equation? */
671		if (roff_closeeqn(mdoc->roff))
672			if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
673				return(0);
674		offs += (end - (buf + offs));
675	}
676
677	return(1);
678}
679#endif
680
681/*
682 * Parse free-form text, that is, a line that does not begin with the
683 * control character.
684 */
685static int
686mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
687{
688	char		 *c, *ws, *end;
689	struct mdoc_node *n;
690
691	assert(mdoc->last);
692	n = mdoc->last;
693
694	/*
695	 * Divert directly to list processing if we're encountering a
696	 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
697	 * (a MDOC_BODY means it's already open, in which case we should
698	 * process within its context in the normal way).
699	 */
700
701	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
702	    LIST_column == n->norm->Bl.type) {
703		/* `Bl' is open without any children. */
704		mdoc->flags |= MDOC_FREECOL;
705		return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
706	}
707
708	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
709	    NULL != n->parent &&
710	    MDOC_Bl == n->parent->tok &&
711	    LIST_column == n->parent->norm->Bl.type) {
712		/* `Bl' has block-level `It' children. */
713		mdoc->flags |= MDOC_FREECOL;
714		return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
715	}
716
717	/*
718	 * Search for the beginning of unescaped trailing whitespace (ws)
719	 * and for the first character not to be output (end).
720	 */
721
722	/* FIXME: replace with strcspn(). */
723	ws = NULL;
724	for (c = end = buf + offs; *c; c++) {
725		switch (*c) {
726		case ' ':
727			if (NULL == ws)
728				ws = c;
729			continue;
730		case '\t':
731			/*
732			 * Always warn about trailing tabs,
733			 * even outside literal context,
734			 * where they should be put on the next line.
735			 */
736			if (NULL == ws)
737				ws = c;
738			/*
739			 * Strip trailing tabs in literal context only;
740			 * outside, they affect the next line.
741			 */
742			if (MDOC_LITERAL & mdoc->flags)
743				continue;
744			break;
745		case '\\':
746			/* Skip the escaped character, too, if any. */
747			if (c[1])
748				c++;
749			/* FALLTHROUGH */
750		default:
751			ws = NULL;
752			break;
753		}
754		end = c + 1;
755	}
756	*end = '\0';
757
758	if (ws)
759		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
760		    line, (int)(ws-buf), NULL);
761
762	if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
763		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
764		    line, (int)(c - buf), NULL);
765
766		/*
767		 * Insert a `sp' in the case of a blank line.  Technically,
768		 * blank lines aren't allowed, but enough manuals assume this
769		 * behaviour that we want to work around it.
770		 */
771		if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
772			return(0);
773
774		mdoc->next = MDOC_NEXT_SIBLING;
775
776		return(mdoc_valid_post(mdoc));
777	}
778
779	if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
780		return(0);
781
782	if (MDOC_LITERAL & mdoc->flags)
783		return(1);
784
785	/*
786	 * End-of-sentence check.  If the last character is an unescaped
787	 * EOS character, then flag the node as being the end of a
788	 * sentence.  The front-end will know how to interpret this.
789	 */
790
791	assert(buf < end);
792
793	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
794		mdoc->last->flags |= MDOC_EOS;
795
796	return(1);
797}
798
799/*
800 * Parse a macro line, that is, a line beginning with the control
801 * character.
802 */
803static int
804mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
805{
806	enum mdoct	  tok;
807	int		  i, sv;
808	char		  mac[5];
809	struct mdoc_node *n;
810
811	/* Empty post-control lines are ignored. */
812
813	if ('"' == buf[offs]) {
814		mandoc_msg(MANDOCERR_COMMENT_BAD, mdoc->parse,
815		    ln, offs, NULL);
816		return(1);
817	} else if ('\0' == buf[offs])
818		return(1);
819
820	sv = offs;
821
822	/*
823	 * Copy the first word into a nil-terminated buffer.
824	 * Stop copying when a tab, space, or eoln is encountered.
825	 */
826
827	i = 0;
828	while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
829	    '\t' != buf[offs])
830		mac[i++] = buf[offs++];
831
832	mac[i] = '\0';
833
834	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
835
836	if (MDOC_MAX == tok) {
837		mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
838		    ln, sv, "%s", buf + sv - 1);
839		return(1);
840	}
841
842	/* Disregard the first trailing tab, if applicable. */
843
844	if ('\t' == buf[offs])
845		offs++;
846
847	/* Jump to the next non-whitespace word. */
848
849	while (buf[offs] && ' ' == buf[offs])
850		offs++;
851
852	/*
853	 * Trailing whitespace.  Note that tabs are allowed to be passed
854	 * into the parser as "text", so we only warn about spaces here.
855	 */
856
857	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
858		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
859		    ln, offs - 1, NULL);
860
861	/*
862	 * If an initial macro or a list invocation, divert directly
863	 * into macro processing.
864	 */
865
866	if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok)
867		return(mdoc_macro(mdoc, tok, ln, sv, &offs, buf));
868
869	n = mdoc->last;
870	assert(mdoc->last);
871
872	/*
873	 * If the first macro of a `Bl -column', open an `It' block
874	 * context around the parsed macro.
875	 */
876
877	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
878	    LIST_column == n->norm->Bl.type) {
879		mdoc->flags |= MDOC_FREECOL;
880		return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
881	}
882
883	/*
884	 * If we're following a block-level `It' within a `Bl -column'
885	 * context (perhaps opened in the above block or in ptext()),
886	 * then open an `It' block context around the parsed macro.
887	 */
888
889	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
890	    NULL != n->parent &&
891	    MDOC_Bl == n->parent->tok &&
892	    LIST_column == n->parent->norm->Bl.type) {
893		mdoc->flags |= MDOC_FREECOL;
894		return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
895	}
896
897	/* Normal processing of a macro. */
898
899	if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
900		return(0);
901
902	/* In quick mode (for mandocdb), abort after the NAME section. */
903
904	if (mdoc->quick && MDOC_Sh == tok &&
905	    SEC_NAME != mdoc->last->sec)
906		return(2);
907
908	return(1);
909}
910
911enum mdelim
912mdoc_isdelim(const char *p)
913{
914
915	if ('\0' == p[0])
916		return(DELIM_NONE);
917
918	if ('\0' == p[1])
919		switch (p[0]) {
920		case '(':
921			/* FALLTHROUGH */
922		case '[':
923			return(DELIM_OPEN);
924		case '|':
925			return(DELIM_MIDDLE);
926		case '.':
927			/* FALLTHROUGH */
928		case ',':
929			/* FALLTHROUGH */
930		case ';':
931			/* FALLTHROUGH */
932		case ':':
933			/* FALLTHROUGH */
934		case '?':
935			/* FALLTHROUGH */
936		case '!':
937			/* FALLTHROUGH */
938		case ')':
939			/* FALLTHROUGH */
940		case ']':
941			return(DELIM_CLOSE);
942		default:
943			return(DELIM_NONE);
944		}
945
946	if ('\\' != p[0])
947		return(DELIM_NONE);
948
949	if (0 == strcmp(p + 1, "."))
950		return(DELIM_CLOSE);
951	if (0 == strcmp(p + 1, "fR|\\fP"))
952		return(DELIM_MIDDLE);
953
954	return(DELIM_NONE);
955}
956
957void
958mdoc_deroff(char **dest, const struct mdoc_node *n)
959{
960	char	*cp;
961	size_t	 sz;
962
963	if (MDOC_TEXT != n->type) {
964		for (n = n->child; n; n = n->next)
965			mdoc_deroff(dest, n);
966		return;
967	}
968
969	/* Skip leading whitespace. */
970
971	for (cp = n->string; '\0' != *cp; cp++)
972		if (0 == isspace((unsigned char)*cp))
973			break;
974
975	/* Skip trailing whitespace. */
976
977	for (sz = strlen(cp); sz; sz--)
978		if (0 == isspace((unsigned char)cp[sz-1]))
979			break;
980
981	/* Skip empty strings. */
982
983	if (0 == sz)
984		return;
985
986	if (NULL == *dest) {
987		*dest = mandoc_strndup(cp, sz);
988		return;
989	}
990
991	mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
992	free(*dest);
993	*dest = cp;
994}
995