1/*	$Id: mdoc.c,v 1.256 2015/10/30 19:04:16 schwarze Exp $ */
2/*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010, 2012-2015 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 <stdarg.h>
25#include <stdio.h>
26#include <stdlib.h>
27#include <string.h>
28#include <time.h>
29
30#include "mandoc_aux.h"
31#include "mandoc.h"
32#include "roff.h"
33#include "mdoc.h"
34#include "libmandoc.h"
35#include "roff_int.h"
36#include "libmdoc.h"
37
38const	char *const __mdoc_macronames[MDOC_MAX + 1] = {
39	"text",
40	"Ap",		"Dd",		"Dt",		"Os",
41	"Sh",		"Ss",		"Pp",		"D1",
42	"Dl",		"Bd",		"Ed",		"Bl",
43	"El",		"It",		"Ad",		"An",
44	"Ar",		"Cd",		"Cm",		"Dv",
45	"Er",		"Ev",		"Ex",		"Fa",
46	"Fd",		"Fl",		"Fn",		"Ft",
47	"Ic",		"In",		"Li",		"Nd",
48	"Nm",		"Op",		"Ot",		"Pa",
49	"Rv",		"St",		"Va",		"Vt",
50	"Xr",		"%A",		"%B",		"%D",
51	"%I",		"%J",		"%N",		"%O",
52	"%P",		"%R",		"%T",		"%V",
53	"Ac",		"Ao",		"Aq",		"At",
54	"Bc",		"Bf",		"Bo",		"Bq",
55	"Bsx",		"Bx",		"Db",		"Dc",
56	"Do",		"Dq",		"Ec",		"Ef",
57	"Em",		"Eo",		"Fx",		"Ms",
58	"No",		"Ns",		"Nx",		"Ox",
59	"Pc",		"Pf",		"Po",		"Pq",
60	"Qc",		"Ql",		"Qo",		"Qq",
61	"Re",		"Rs",		"Sc",		"So",
62	"Sq",		"Sm",		"Sx",		"Sy",
63	"Tn",		"Ux",		"Xc",		"Xo",
64	"Fo",		"Fc",		"Oo",		"Oc",
65	"Bk",		"Ek",		"Bt",		"Hf",
66	"Fr",		"Ud",		"Lb",		"Lp",
67	"Lk",		"Mt",		"Brq",		"Bro",
68	"Brc",		"%C",		"Es",		"En",
69	"Dx",		"%Q",		"br",		"sp",
70	"%U",		"Ta",		"ll",
71};
72
73const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {
74	"split",		"nosplit",		"ragged",
75	"unfilled",		"literal",		"file",
76	"offset",		"bullet",		"dash",
77	"hyphen",		"item",			"enum",
78	"tag",			"diag",			"hang",
79	"ohang",		"inset",		"column",
80	"width",		"compact",		"std",
81	"filled",		"words",		"emphasis",
82	"symbolic",		"nested",		"centered"
83	};
84
85const	char * const *mdoc_macronames = __mdoc_macronames + 1;
86const	char * const *mdoc_argnames = __mdoc_argnames;
87
88static	int		  mdoc_ptext(struct roff_man *, int, char *, int);
89static	int		  mdoc_pmacro(struct roff_man *, int, char *, int);
90
91
92/*
93 * Main parse routine.  Parses a single line -- really just hands off to
94 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
95 */
96int
97mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
98{
99
100	if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
101		mdoc->flags |= MDOC_NEWLINE;
102
103	/*
104	 * Let the roff nS register switch SYNOPSIS mode early,
105	 * such that the parser knows at all times
106	 * whether this mode is on or off.
107	 * Note that this mode is also switched by the Sh macro.
108	 */
109	if (roff_getreg(mdoc->roff, "nS"))
110		mdoc->flags |= MDOC_SYNOPSIS;
111	else
112		mdoc->flags &= ~MDOC_SYNOPSIS;
113
114	return roff_getcontrol(mdoc->roff, buf, &offs) ?
115	    mdoc_pmacro(mdoc, ln, buf, offs) :
116	    mdoc_ptext(mdoc, ln, buf, offs);
117}
118
119void
120mdoc_macro(MACRO_PROT_ARGS)
121{
122	assert(tok > TOKEN_NONE && tok < MDOC_MAX);
123
124	(*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
125}
126
127void
128mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok)
129{
130	struct roff_node *p;
131
132	p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
133	roff_node_append(mdoc, p);
134	mdoc->next = ROFF_NEXT_CHILD;
135}
136
137struct roff_node *
138mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
139		struct roff_node *body, enum mdoc_endbody end)
140{
141	struct roff_node *p;
142
143	body->flags |= MDOC_ENDED;
144	body->parent->flags |= MDOC_ENDED;
145	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
146	p->body = body;
147	p->norm = body->norm;
148	p->end = end;
149	roff_node_append(mdoc, p);
150	mdoc->next = ROFF_NEXT_SIBLING;
151	return p;
152}
153
154struct roff_node *
155mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
156	int tok, struct mdoc_arg *args)
157{
158	struct roff_node *p;
159
160	p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
161	p->args = args;
162	if (p->args)
163		(args->refcnt)++;
164
165	switch (tok) {
166	case MDOC_Bd:
167	case MDOC_Bf:
168	case MDOC_Bl:
169	case MDOC_En:
170	case MDOC_Rs:
171		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
172		break;
173	default:
174		break;
175	}
176	roff_node_append(mdoc, p);
177	mdoc->next = ROFF_NEXT_CHILD;
178	return p;
179}
180
181void
182mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
183	int tok, struct mdoc_arg *args)
184{
185	struct roff_node *p;
186
187	p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
188	p->args = args;
189	if (p->args)
190		(args->refcnt)++;
191
192	switch (tok) {
193	case MDOC_An:
194		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
195		break;
196	default:
197		break;
198	}
199	roff_node_append(mdoc, p);
200	mdoc->next = ROFF_NEXT_CHILD;
201}
202
203void
204mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
205{
206
207	roff_node_unlink(mdoc, p);
208	p->prev = p->next = NULL;
209	roff_node_append(mdoc, p);
210}
211
212/*
213 * Parse free-form text, that is, a line that does not begin with the
214 * control character.
215 */
216static int
217mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
218{
219	struct roff_node *n;
220	char		 *c, *ws, *end;
221
222	assert(mdoc->last);
223	n = mdoc->last;
224
225	/*
226	 * Divert directly to list processing if we're encountering a
227	 * columnar ROFFT_BLOCK with or without a prior ROFFT_BLOCK entry
228	 * (a ROFFT_BODY means it's already open, in which case we should
229	 * process within its context in the normal way).
230	 */
231
232	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
233	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
234		/* `Bl' is open without any children. */
235		mdoc->flags |= MDOC_FREECOL;
236		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
237		return 1;
238	}
239
240	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
241	    NULL != n->parent &&
242	    MDOC_Bl == n->parent->tok &&
243	    LIST_column == n->parent->norm->Bl.type) {
244		/* `Bl' has block-level `It' children. */
245		mdoc->flags |= MDOC_FREECOL;
246		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
247		return 1;
248	}
249
250	/*
251	 * Search for the beginning of unescaped trailing whitespace (ws)
252	 * and for the first character not to be output (end).
253	 */
254
255	/* FIXME: replace with strcspn(). */
256	ws = NULL;
257	for (c = end = buf + offs; *c; c++) {
258		switch (*c) {
259		case ' ':
260			if (NULL == ws)
261				ws = c;
262			continue;
263		case '\t':
264			/*
265			 * Always warn about trailing tabs,
266			 * even outside literal context,
267			 * where they should be put on the next line.
268			 */
269			if (NULL == ws)
270				ws = c;
271			/*
272			 * Strip trailing tabs in literal context only;
273			 * outside, they affect the next line.
274			 */
275			if (MDOC_LITERAL & mdoc->flags)
276				continue;
277			break;
278		case '\\':
279			/* Skip the escaped character, too, if any. */
280			if (c[1])
281				c++;
282			/* FALLTHROUGH */
283		default:
284			ws = NULL;
285			break;
286		}
287		end = c + 1;
288	}
289	*end = '\0';
290
291	if (ws)
292		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
293		    line, (int)(ws-buf), NULL);
294
295	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
296		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
297		    line, (int)(c - buf), NULL);
298
299		/*
300		 * Insert a `sp' in the case of a blank line.  Technically,
301		 * blank lines aren't allowed, but enough manuals assume this
302		 * behaviour that we want to work around it.
303		 */
304		roff_elem_alloc(mdoc, line, offs, MDOC_sp);
305		mdoc->last->flags |= MDOC_VALID | MDOC_ENDED;
306		mdoc->next = ROFF_NEXT_SIBLING;
307		return 1;
308	}
309
310	roff_word_alloc(mdoc, line, offs, buf+offs);
311
312	if (mdoc->flags & MDOC_LITERAL)
313		return 1;
314
315	/*
316	 * End-of-sentence check.  If the last character is an unescaped
317	 * EOS character, then flag the node as being the end of a
318	 * sentence.  The front-end will know how to interpret this.
319	 */
320
321	assert(buf < end);
322
323	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
324		mdoc->last->flags |= MDOC_EOS;
325	return 1;
326}
327
328/*
329 * Parse a macro line, that is, a line beginning with the control
330 * character.
331 */
332static int
333mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
334{
335	struct roff_node *n;
336	const char	 *cp;
337	int		  tok;
338	int		  i, sv;
339	char		  mac[5];
340
341	sv = offs;
342
343	/*
344	 * Copy the first word into a nil-terminated buffer.
345	 * Stop when a space, tab, escape, or eoln is encountered.
346	 */
347
348	i = 0;
349	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
350		mac[i++] = buf[offs++];
351
352	mac[i] = '\0';
353
354	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
355
356	if (tok == TOKEN_NONE) {
357		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
358		    ln, sv, buf + sv - 1);
359		return 1;
360	}
361
362	/* Skip a leading escape sequence or tab. */
363
364	switch (buf[offs]) {
365	case '\\':
366		cp = buf + offs + 1;
367		mandoc_escape(&cp, NULL, NULL);
368		offs = cp - buf;
369		break;
370	case '\t':
371		offs++;
372		break;
373	default:
374		break;
375	}
376
377	/* Jump to the next non-whitespace word. */
378
379	while (buf[offs] && ' ' == buf[offs])
380		offs++;
381
382	/*
383	 * Trailing whitespace.  Note that tabs are allowed to be passed
384	 * into the parser as "text", so we only warn about spaces here.
385	 */
386
387	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
388		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
389		    ln, offs - 1, NULL);
390
391	/*
392	 * If an initial macro or a list invocation, divert directly
393	 * into macro processing.
394	 */
395
396	if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
397		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
398		return 1;
399	}
400
401	n = mdoc->last;
402	assert(mdoc->last);
403
404	/*
405	 * If the first macro of a `Bl -column', open an `It' block
406	 * context around the parsed macro.
407	 */
408
409	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
410	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
411		mdoc->flags |= MDOC_FREECOL;
412		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
413		return 1;
414	}
415
416	/*
417	 * If we're following a block-level `It' within a `Bl -column'
418	 * context (perhaps opened in the above block or in ptext()),
419	 * then open an `It' block context around the parsed macro.
420	 */
421
422	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
423	    NULL != n->parent &&
424	    MDOC_Bl == n->parent->tok &&
425	    LIST_column == n->parent->norm->Bl.type) {
426		mdoc->flags |= MDOC_FREECOL;
427		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
428		return 1;
429	}
430
431	/* Normal processing of a macro. */
432
433	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
434
435	/* In quick mode (for mandocdb), abort after the NAME section. */
436
437	if (mdoc->quick && MDOC_Sh == tok &&
438	    SEC_NAME != mdoc->last->sec)
439		return 2;
440
441	return 1;
442}
443
444enum mdelim
445mdoc_isdelim(const char *p)
446{
447
448	if ('\0' == p[0])
449		return DELIM_NONE;
450
451	if ('\0' == p[1])
452		switch (p[0]) {
453		case '(':
454		case '[':
455			return DELIM_OPEN;
456		case '|':
457			return DELIM_MIDDLE;
458		case '.':
459		case ',':
460		case ';':
461		case ':':
462		case '?':
463		case '!':
464		case ')':
465		case ']':
466			return DELIM_CLOSE;
467		default:
468			return DELIM_NONE;
469		}
470
471	if ('\\' != p[0])
472		return DELIM_NONE;
473
474	if (0 == strcmp(p + 1, "."))
475		return DELIM_CLOSE;
476	if (0 == strcmp(p + 1, "fR|\\fP"))
477		return DELIM_MIDDLE;
478
479	return DELIM_NONE;
480}
481
482void
483mdoc_validate(struct roff_man *mdoc)
484{
485
486	mdoc->last = mdoc->first;
487	mdoc_node_validate(mdoc);
488	mdoc_state_reset(mdoc);
489}
490