xref: /netbsd-src/external/bsd/mdocml/dist/mdoc.c (revision c2f76ff004a2cb67efe5b12d97bd3ef7fe89e18d) 
1 /*	$Vendor-Id: mdoc.c,v 1.177 2011/01/03 11:27:33 kristaps Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010 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 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21 
22 #include <sys/types.h>
23 
24 #include <assert.h>
25 #include <stdarg.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <time.h>
30 
31 #include "mandoc.h"
32 #include "libmdoc.h"
33 #include "libmandoc.h"
34 
35 const	char *const __mdoc_macronames[MDOC_MAX] = {
36 	"Ap",		"Dd",		"Dt",		"Os",
37 	"Sh",		"Ss",		"Pp",		"D1",
38 	"Dl",		"Bd",		"Ed",		"Bl",
39 	"El",		"It",		"Ad",		"An",
40 	"Ar",		"Cd",		"Cm",		"Dv",
41 	"Er",		"Ev",		"Ex",		"Fa",
42 	"Fd",		"Fl",		"Fn",		"Ft",
43 	"Ic",		"In",		"Li",		"Nd",
44 	"Nm",		"Op",		"Ot",		"Pa",
45 	"Rv",		"St",		"Va",		"Vt",
46 	/* LINTED */
47 	"Xr",		"%A",		"%B",		"%D",
48 	/* LINTED */
49 	"%I",		"%J",		"%N",		"%O",
50 	/* LINTED */
51 	"%P",		"%R",		"%T",		"%V",
52 	"Ac",		"Ao",		"Aq",		"At",
53 	"Bc",		"Bf",		"Bo",		"Bq",
54 	"Bsx",		"Bx",		"Db",		"Dc",
55 	"Do",		"Dq",		"Ec",		"Ef",
56 	"Em",		"Eo",		"Fx",		"Ms",
57 	"No",		"Ns",		"Nx",		"Ox",
58 	"Pc",		"Pf",		"Po",		"Pq",
59 	"Qc",		"Ql",		"Qo",		"Qq",
60 	"Re",		"Rs",		"Sc",		"So",
61 	"Sq",		"Sm",		"Sx",		"Sy",
62 	"Tn",		"Ux",		"Xc",		"Xo",
63 	"Fo",		"Fc",		"Oo",		"Oc",
64 	"Bk",		"Ek",		"Bt",		"Hf",
65 	"Fr",		"Ud",		"Lb",		"Lp",
66 	"Lk",		"Mt",		"Brq",		"Bro",
67 	/* LINTED */
68 	"Brc",		"%C",		"Es",		"En",
69 	/* LINTED */
70 	"Dx",		"%Q",		"br",		"sp",
71 	/* LINTED */
72 	"%U",		"Ta"
73 	};
74 
75 const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {
76 	"split",		"nosplit",		"ragged",
77 	"unfilled",		"literal",		"file",
78 	"offset",		"bullet",		"dash",
79 	"hyphen",		"item",			"enum",
80 	"tag",			"diag",			"hang",
81 	"ohang",		"inset",		"column",
82 	"width",		"compact",		"std",
83 	"filled",		"words",		"emphasis",
84 	"symbolic",		"nested",		"centered"
85 	};
86 
87 const	char * const *mdoc_macronames = __mdoc_macronames;
88 const	char * const *mdoc_argnames = __mdoc_argnames;
89 
90 static	void		  mdoc_node_free(struct mdoc_node *);
91 static	void		  mdoc_node_unlink(struct mdoc *,
92 				struct mdoc_node *);
93 static	void		  mdoc_free1(struct mdoc *);
94 static	void		  mdoc_alloc1(struct mdoc *);
95 static	struct mdoc_node *node_alloc(struct mdoc *, int, int,
96 				enum mdoct, enum mdoc_type);
97 static	int		  node_append(struct mdoc *,
98 				struct mdoc_node *);
99 static	int		  mdoc_ptext(struct mdoc *, int, char *, int);
100 static	int		  mdoc_pmacro(struct mdoc *, int, char *, int);
101 static	int		  mdoc_span_alloc(struct mdoc *,
102 				const struct tbl_span *);
103 
104 
105 const struct mdoc_node *
106 mdoc_node(const struct mdoc *m)
107 {
108 
109 	assert( ! (MDOC_HALT & m->flags));
110 	return(m->first);
111 }
112 
113 
114 const struct mdoc_meta *
115 mdoc_meta(const struct mdoc *m)
116 {
117 
118 	assert( ! (MDOC_HALT & m->flags));
119 	return(&m->meta);
120 }
121 
122 
123 /*
124  * Frees volatile resources (parse tree, meta-data, fields).
125  */
126 static void
127 mdoc_free1(struct mdoc *mdoc)
128 {
129 
130 	if (mdoc->first)
131 		mdoc_node_delete(mdoc, mdoc->first);
132 	if (mdoc->meta.title)
133 		free(mdoc->meta.title);
134 	if (mdoc->meta.os)
135 		free(mdoc->meta.os);
136 	if (mdoc->meta.name)
137 		free(mdoc->meta.name);
138 	if (mdoc->meta.arch)
139 		free(mdoc->meta.arch);
140 	if (mdoc->meta.vol)
141 		free(mdoc->meta.vol);
142 	if (mdoc->meta.msec)
143 		free(mdoc->meta.msec);
144 }
145 
146 
147 /*
148  * Allocate all volatile resources (parse tree, meta-data, fields).
149  */
150 static void
151 mdoc_alloc1(struct mdoc *mdoc)
152 {
153 
154 	memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
155 	mdoc->flags = 0;
156 	mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
157 	mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
158 	mdoc->first = mdoc->last;
159 	mdoc->last->type = MDOC_ROOT;
160 	mdoc->next = MDOC_NEXT_CHILD;
161 }
162 
163 
164 /*
165  * Free up volatile resources (see mdoc_free1()) then re-initialises the
166  * data with mdoc_alloc1().  After invocation, parse data has been reset
167  * and the parser is ready for re-invocation on a new tree; however,
168  * cross-parse non-volatile data is kept intact.
169  */
170 void
171 mdoc_reset(struct mdoc *mdoc)
172 {
173 
174 	mdoc_free1(mdoc);
175 	mdoc_alloc1(mdoc);
176 }
177 
178 
179 /*
180  * Completely free up all volatile and non-volatile parse resources.
181  * After invocation, the pointer is no longer usable.
182  */
183 void
184 mdoc_free(struct mdoc *mdoc)
185 {
186 
187 	mdoc_free1(mdoc);
188 	free(mdoc);
189 }
190 
191 
192 /*
193  * Allocate volatile and non-volatile parse resources.
194  */
195 struct mdoc *
196 mdoc_alloc(struct regset *regs, void *data, mandocmsg msg)
197 {
198 	struct mdoc	*p;
199 
200 	p = mandoc_calloc(1, sizeof(struct mdoc));
201 
202 	p->msg = msg;
203 	p->data = data;
204 	p->regs = regs;
205 
206 	mdoc_hash_init();
207 	mdoc_alloc1(p);
208 	return(p);
209 }
210 
211 
212 /*
213  * Climb back up the parse tree, validating open scopes.  Mostly calls
214  * through to macro_end() in macro.c.
215  */
216 int
217 mdoc_endparse(struct mdoc *m)
218 {
219 
220 	assert( ! (MDOC_HALT & m->flags));
221 	if (mdoc_macroend(m))
222 		return(1);
223 	m->flags |= MDOC_HALT;
224 	return(0);
225 }
226 
227 int
228 mdoc_addspan(struct mdoc *m, const struct tbl_span *sp)
229 {
230 
231 	assert( ! (MDOC_HALT & m->flags));
232 
233 	/* No text before an initial macro. */
234 
235 	if (SEC_NONE == m->lastnamed) {
236 		/* FIXME: grab from span. */
237 		mdoc_pmsg(m, 0, 0, MANDOCERR_NOTEXT);
238 		return(1);
239 	}
240 
241 	return(mdoc_span_alloc(m, sp));
242 }
243 
244 
245 /*
246  * Main parse routine.  Parses a single line -- really just hands off to
247  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
248  */
249 int
250 mdoc_parseln(struct mdoc *m, int ln, char *buf, int offs)
251 {
252 
253 	assert( ! (MDOC_HALT & m->flags));
254 
255 	m->flags |= MDOC_NEWLINE;
256 
257 	/*
258 	 * Let the roff nS register switch SYNOPSIS mode early,
259 	 * such that the parser knows at all times
260 	 * whether this mode is on or off.
261 	 * Note that this mode is also switched by the Sh macro.
262 	 */
263 	if (m->regs->regs[(int)REG_nS].set) {
264 		if (m->regs->regs[(int)REG_nS].v.u)
265 			m->flags |= MDOC_SYNOPSIS;
266 		else
267 			m->flags &= ~MDOC_SYNOPSIS;
268 	}
269 
270 	return(('.' == buf[offs] || '\'' == buf[offs]) ?
271 			mdoc_pmacro(m, ln, buf, offs) :
272 			mdoc_ptext(m, ln, buf, offs));
273 }
274 
275 
276 int
277 mdoc_vmsg(struct mdoc *mdoc, enum mandocerr t,
278 		int ln, int pos, const char *fmt, ...)
279 {
280 	char		 buf[256];
281 	va_list		 ap;
282 
283 	va_start(ap, fmt);
284 	vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
285 	va_end(ap);
286 
287 	return((*mdoc->msg)(t, mdoc->data, ln, pos, buf));
288 }
289 
290 
291 int
292 mdoc_macro(MACRO_PROT_ARGS)
293 {
294 	assert(tok < MDOC_MAX);
295 
296 	/* If we're in the body, deny prologue calls. */
297 
298 	if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
299 			MDOC_PBODY & m->flags) {
300 		mdoc_pmsg(m, line, ppos, MANDOCERR_BADBODY);
301 		return(1);
302 	}
303 
304 	/* If we're in the prologue, deny "body" macros.  */
305 
306 	if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
307 			! (MDOC_PBODY & m->flags)) {
308 		mdoc_pmsg(m, line, ppos, MANDOCERR_BADPROLOG);
309 		if (NULL == m->meta.msec)
310 			m->meta.msec = mandoc_strdup("1");
311 		if (NULL == m->meta.title)
312 			m->meta.title = mandoc_strdup("UNKNOWN");
313 		if (NULL == m->meta.vol)
314 			m->meta.vol = mandoc_strdup("LOCAL");
315 		if (NULL == m->meta.os)
316 			m->meta.os = mandoc_strdup("LOCAL");
317 		if (0 == m->meta.date)
318 			m->meta.date = time(NULL);
319 		m->flags |= MDOC_PBODY;
320 	}
321 
322 	return((*mdoc_macros[tok].fp)(m, tok, line, ppos, pos, buf));
323 }
324 
325 
326 static int
327 node_append(struct mdoc *mdoc, struct mdoc_node *p)
328 {
329 
330 	assert(mdoc->last);
331 	assert(mdoc->first);
332 	assert(MDOC_ROOT != p->type);
333 
334 	switch (mdoc->next) {
335 	case (MDOC_NEXT_SIBLING):
336 		mdoc->last->next = p;
337 		p->prev = mdoc->last;
338 		p->parent = mdoc->last->parent;
339 		break;
340 	case (MDOC_NEXT_CHILD):
341 		mdoc->last->child = p;
342 		p->parent = mdoc->last;
343 		break;
344 	default:
345 		abort();
346 		/* NOTREACHED */
347 	}
348 
349 	p->parent->nchild++;
350 
351 	/*
352 	 * Copy over the normalised-data pointer of our parent.  Not
353 	 * everybody has one, but copying a null pointer is fine.
354 	 */
355 
356 	switch (p->type) {
357 	case (MDOC_BODY):
358 		/* FALLTHROUGH */
359 	case (MDOC_TAIL):
360 		/* FALLTHROUGH */
361 	case (MDOC_HEAD):
362 		p->norm = p->parent->norm;
363 		break;
364 	default:
365 		break;
366 	}
367 
368 	if ( ! mdoc_valid_pre(mdoc, p))
369 		return(0);
370 
371 	switch (p->type) {
372 	case (MDOC_HEAD):
373 		assert(MDOC_BLOCK == p->parent->type);
374 		p->parent->head = p;
375 		break;
376 	case (MDOC_TAIL):
377 		assert(MDOC_BLOCK == p->parent->type);
378 		p->parent->tail = p;
379 		break;
380 	case (MDOC_BODY):
381 		if (p->end)
382 			break;
383 		assert(MDOC_BLOCK == p->parent->type);
384 		p->parent->body = p;
385 		break;
386 	default:
387 		break;
388 	}
389 
390 	mdoc->last = p;
391 
392 	switch (p->type) {
393 	case (MDOC_TBL):
394 		/* FALLTHROUGH */
395 	case (MDOC_TEXT):
396 		if ( ! mdoc_valid_post(mdoc))
397 			return(0);
398 		break;
399 	default:
400 		break;
401 	}
402 
403 	return(1);
404 }
405 
406 
407 static struct mdoc_node *
408 node_alloc(struct mdoc *m, int line, int pos,
409 		enum mdoct tok, enum mdoc_type type)
410 {
411 	struct mdoc_node *p;
412 
413 	p = mandoc_calloc(1, sizeof(struct mdoc_node));
414 	p->sec = m->lastsec;
415 	p->line = line;
416 	p->pos = pos;
417 	p->tok = tok;
418 	p->type = type;
419 
420 	/* Flag analysis. */
421 
422 	if (MDOC_SYNOPSIS & m->flags)
423 		p->flags |= MDOC_SYNPRETTY;
424 	else
425 		p->flags &= ~MDOC_SYNPRETTY;
426 	if (MDOC_NEWLINE & m->flags)
427 		p->flags |= MDOC_LINE;
428 	m->flags &= ~MDOC_NEWLINE;
429 
430 	return(p);
431 }
432 
433 
434 int
435 mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
436 {
437 	struct mdoc_node *p;
438 
439 	p = node_alloc(m, line, pos, tok, MDOC_TAIL);
440 	if ( ! node_append(m, p))
441 		return(0);
442 	m->next = MDOC_NEXT_CHILD;
443 	return(1);
444 }
445 
446 
447 int
448 mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
449 {
450 	struct mdoc_node *p;
451 
452 	assert(m->first);
453 	assert(m->last);
454 
455 	p = node_alloc(m, line, pos, tok, MDOC_HEAD);
456 	if ( ! node_append(m, p))
457 		return(0);
458 	m->next = MDOC_NEXT_CHILD;
459 	return(1);
460 }
461 
462 
463 int
464 mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok)
465 {
466 	struct mdoc_node *p;
467 
468 	p = node_alloc(m, line, pos, tok, MDOC_BODY);
469 	if ( ! node_append(m, p))
470 		return(0);
471 	m->next = MDOC_NEXT_CHILD;
472 	return(1);
473 }
474 
475 
476 int
477 mdoc_endbody_alloc(struct mdoc *m, int line, int pos, enum mdoct tok,
478 		struct mdoc_node *body, enum mdoc_endbody end)
479 {
480 	struct mdoc_node *p;
481 
482 	p = node_alloc(m, line, pos, tok, MDOC_BODY);
483 	p->pending = body;
484 	p->end = end;
485 	if ( ! node_append(m, p))
486 		return(0);
487 	m->next = MDOC_NEXT_SIBLING;
488 	return(1);
489 }
490 
491 
492 int
493 mdoc_block_alloc(struct mdoc *m, int line, int pos,
494 		enum mdoct tok, struct mdoc_arg *args)
495 {
496 	struct mdoc_node *p;
497 
498 	p = node_alloc(m, line, pos, tok, MDOC_BLOCK);
499 	p->args = args;
500 	if (p->args)
501 		(args->refcnt)++;
502 
503 	switch (tok) {
504 	case (MDOC_Bd):
505 		/* FALLTHROUGH */
506 	case (MDOC_Bf):
507 		/* FALLTHROUGH */
508 	case (MDOC_Bl):
509 		/* FALLTHROUGH */
510 	case (MDOC_Rs):
511 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
512 		break;
513 	default:
514 		break;
515 	}
516 
517 	if ( ! node_append(m, p))
518 		return(0);
519 	m->next = MDOC_NEXT_CHILD;
520 	return(1);
521 }
522 
523 
524 int
525 mdoc_elem_alloc(struct mdoc *m, int line, int pos,
526 		enum mdoct tok, struct mdoc_arg *args)
527 {
528 	struct mdoc_node *p;
529 
530 	p = node_alloc(m, line, pos, tok, MDOC_ELEM);
531 	p->args = args;
532 	if (p->args)
533 		(args->refcnt)++;
534 
535 	switch (tok) {
536 	case (MDOC_An):
537 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
538 		break;
539 	default:
540 		break;
541 	}
542 
543 	if ( ! node_append(m, p))
544 		return(0);
545 	m->next = MDOC_NEXT_CHILD;
546 	return(1);
547 }
548 
549 static int
550 mdoc_span_alloc(struct mdoc *m, const struct tbl_span *sp)
551 {
552 	struct mdoc_node *n;
553 
554 	/* FIXME: grab from tbl_span. */
555 	n = node_alloc(m, 0, 0, MDOC_MAX, MDOC_TBL);
556 	n->span = sp;
557 
558 	if ( ! node_append(m, n))
559 		return(0);
560 
561 	m->next = MDOC_NEXT_SIBLING;
562 	return(1);
563 }
564 
565 
566 int
567 mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p)
568 {
569 	struct mdoc_node *n;
570 	size_t		  sv, len;
571 
572 	len = strlen(p);
573 
574 	n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT);
575 	n->string = mandoc_malloc(len + 1);
576 	sv = strlcpy(n->string, p, len + 1);
577 
578 	/* Prohibit truncation. */
579 	assert(sv < len + 1);
580 
581 	if ( ! node_append(m, n))
582 		return(0);
583 
584 	m->next = MDOC_NEXT_SIBLING;
585 	return(1);
586 }
587 
588 
589 static void
590 mdoc_node_free(struct mdoc_node *p)
591 {
592 
593 	if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
594 		free(p->norm);
595 	if (p->string)
596 		free(p->string);
597 	if (p->args)
598 		mdoc_argv_free(p->args);
599 	free(p);
600 }
601 
602 
603 static void
604 mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n)
605 {
606 
607 	/* Adjust siblings. */
608 
609 	if (n->prev)
610 		n->prev->next = n->next;
611 	if (n->next)
612 		n->next->prev = n->prev;
613 
614 	/* Adjust parent. */
615 
616 	if (n->parent) {
617 		n->parent->nchild--;
618 		if (n->parent->child == n)
619 			n->parent->child = n->prev ? n->prev : n->next;
620 		if (n->parent->last == n)
621 			n->parent->last = n->prev ? n->prev : NULL;
622 	}
623 
624 	/* Adjust parse point, if applicable. */
625 
626 	if (m && m->last == n) {
627 		if (n->prev) {
628 			m->last = n->prev;
629 			m->next = MDOC_NEXT_SIBLING;
630 		} else {
631 			m->last = n->parent;
632 			m->next = MDOC_NEXT_CHILD;
633 		}
634 	}
635 
636 	if (m && m->first == n)
637 		m->first = NULL;
638 }
639 
640 
641 void
642 mdoc_node_delete(struct mdoc *m, struct mdoc_node *p)
643 {
644 
645 	while (p->child) {
646 		assert(p->nchild);
647 		mdoc_node_delete(m, p->child);
648 	}
649 	assert(0 == p->nchild);
650 
651 	mdoc_node_unlink(m, p);
652 	mdoc_node_free(p);
653 }
654 
655 
656 /*
657  * Parse free-form text, that is, a line that does not begin with the
658  * control character.
659  */
660 static int
661 mdoc_ptext(struct mdoc *m, int line, char *buf, int offs)
662 {
663 	char		 *c, *ws, *end;
664 	struct mdoc_node *n;
665 
666 	/* Ignore bogus comments. */
667 
668 	if ('\\' == buf[offs] &&
669 			'.' == buf[offs + 1] &&
670 			'"' == buf[offs + 2]) {
671 		mdoc_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
672 		return(1);
673 	}
674 
675 	/* No text before an initial macro. */
676 
677 	if (SEC_NONE == m->lastnamed) {
678 		mdoc_pmsg(m, line, offs, MANDOCERR_NOTEXT);
679 		return(1);
680 	}
681 
682 	assert(m->last);
683 	n = m->last;
684 
685 	/*
686 	 * Divert directly to list processing if we're encountering a
687 	 * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
688 	 * (a MDOC_BODY means it's already open, in which case we should
689 	 * process within its context in the normal way).
690 	 */
691 
692 	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
693 			LIST_column == n->norm->Bl.type) {
694 		/* `Bl' is open without any children. */
695 		m->flags |= MDOC_FREECOL;
696 		return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
697 	}
698 
699 	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
700 			NULL != n->parent &&
701 			MDOC_Bl == n->parent->tok &&
702 			LIST_column == n->parent->norm->Bl.type) {
703 		/* `Bl' has block-level `It' children. */
704 		m->flags |= MDOC_FREECOL;
705 		return(mdoc_macro(m, MDOC_It, line, offs, &offs, buf));
706 	}
707 
708 	/*
709 	 * Search for the beginning of unescaped trailing whitespace (ws)
710 	 * and for the first character not to be output (end).
711 	 */
712 
713 	/* FIXME: replace with strcspn(). */
714 	ws = NULL;
715 	for (c = end = buf + offs; *c; c++) {
716 		switch (*c) {
717 		case '-':
718 			if (mandoc_hyph(buf + offs, c))
719 				*c = ASCII_HYPH;
720 			ws = NULL;
721 			break;
722 		case ' ':
723 			if (NULL == ws)
724 				ws = c;
725 			continue;
726 		case '\t':
727 			/*
728 			 * Always warn about trailing tabs,
729 			 * even outside literal context,
730 			 * where they should be put on the next line.
731 			 */
732 			if (NULL == ws)
733 				ws = c;
734 			/*
735 			 * Strip trailing tabs in literal context only;
736 			 * outside, they affect the next line.
737 			 */
738 			if (MDOC_LITERAL & m->flags)
739 				continue;
740 			break;
741 		case '\\':
742 			/* Skip the escaped character, too, if any. */
743 			if (c[1])
744 				c++;
745 			/* FALLTHROUGH */
746 		default:
747 			ws = NULL;
748 			break;
749 		}
750 		end = c + 1;
751 	}
752 	*end = '\0';
753 
754 	if (ws)
755 		mdoc_pmsg(m, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
756 
757 	if ('\0' == buf[offs] && ! (MDOC_LITERAL & m->flags)) {
758 		mdoc_pmsg(m, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
759 
760 		/*
761 		 * Insert a `sp' in the case of a blank line.  Technically,
762 		 * blank lines aren't allowed, but enough manuals assume this
763 		 * behaviour that we want to work around it.
764 		 */
765 		if ( ! mdoc_elem_alloc(m, line, offs, MDOC_sp, NULL))
766 			return(0);
767 
768 		m->next = MDOC_NEXT_SIBLING;
769 		return(1);
770 	}
771 
772 	if ( ! mdoc_word_alloc(m, line, offs, buf+offs))
773 		return(0);
774 
775 	if (MDOC_LITERAL & m->flags)
776 		return(1);
777 
778 	/*
779 	 * End-of-sentence check.  If the last character is an unescaped
780 	 * EOS character, then flag the node as being the end of a
781 	 * sentence.  The front-end will know how to interpret this.
782 	 */
783 
784 	assert(buf < end);
785 
786 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
787 		m->last->flags |= MDOC_EOS;
788 
789 	return(1);
790 }
791 
792 
793 /*
794  * Parse a macro line, that is, a line beginning with the control
795  * character.
796  */
797 static int
798 mdoc_pmacro(struct mdoc *m, int ln, char *buf, int offs)
799 {
800 	enum mdoct	  tok;
801 	int		  i, j, sv;
802 	char		  mac[5];
803 	struct mdoc_node *n;
804 
805 	/* Empty lines are ignored. */
806 
807 	offs++;
808 
809 	if ('\0' == buf[offs])
810 		return(1);
811 
812 	i = offs;
813 
814 	/* Accept tabs/whitespace after the initial control char. */
815 
816 	if (' ' == buf[i] || '\t' == buf[i]) {
817 		i++;
818 		while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
819 			i++;
820 		if ('\0' == buf[i])
821 			return(1);
822 	}
823 
824 	sv = i;
825 
826 	/*
827 	 * Copy the first word into a nil-terminated buffer.
828 	 * Stop copying when a tab, space, or eoln is encountered.
829 	 */
830 
831 	j = 0;
832 	while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
833 		mac[j++] = buf[i++];
834 	mac[j] = '\0';
835 
836 	tok = (j > 1 || j < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
837 	if (MDOC_MAX == tok) {
838 		mdoc_vmsg(m, MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
839 		return(1);
840 	}
841 
842 	/* Disregard the first trailing tab, if applicable. */
843 
844 	if ('\t' == buf[i])
845 		i++;
846 
847 	/* Jump to the next non-whitespace word. */
848 
849 	while (buf[i] && ' ' == buf[i])
850 		i++;
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[i] && ' ' == buf[i - 1])
858 		mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
859 
860 	/*
861 	 * If an initial macro or a list invocation, divert directly
862 	 * into macro processing.
863 	 */
864 
865 	if (NULL == m->last || MDOC_It == tok || MDOC_El == tok) {
866 		if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
867 			goto err;
868 		return(1);
869 	}
870 
871 	n = m->last;
872 	assert(m->last);
873 
874 	/*
875 	 * If the first macro of a `Bl -column', open an `It' block
876 	 * context around the parsed macro.
877 	 */
878 
879 	if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
880 			LIST_column == n->norm->Bl.type) {
881 		m->flags |= MDOC_FREECOL;
882 		if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
883 			goto err;
884 		return(1);
885 	}
886 
887 	/*
888 	 * If we're following a block-level `It' within a `Bl -column'
889 	 * context (perhaps opened in the above block or in ptext()),
890 	 * then open an `It' block context around the parsed macro.
891 	 */
892 
893 	if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
894 			NULL != n->parent &&
895 			MDOC_Bl == n->parent->tok &&
896 			LIST_column == n->parent->norm->Bl.type) {
897 		m->flags |= MDOC_FREECOL;
898 		if ( ! mdoc_macro(m, MDOC_It, ln, sv, &sv, buf))
899 			goto err;
900 		return(1);
901 	}
902 
903 	/* Normal processing of a macro. */
904 
905 	if ( ! mdoc_macro(m, tok, ln, sv, &i, buf))
906 		goto err;
907 
908 	return(1);
909 
910 err:	/* Error out. */
911 
912 	m->flags |= MDOC_HALT;
913 	return(0);
914 }
915 
916 
917