xref: /openbsd-src/usr.bin/mandoc/mdoc.c (revision d59bb9942320b767f2a19aaa7690c8c6e30b724c) 
1 /*	$OpenBSD: mdoc.c,v 1.150 2017/03/03 13:55:06 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010, 2012-2017 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 <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 "mandoc_aux.h"
29 #include "mandoc.h"
30 #include "roff.h"
31 #include "mdoc.h"
32 #include "libmandoc.h"
33 #include "roff_int.h"
34 #include "libmdoc.h"
35 
36 const	char *const __mdoc_macronames[MDOC_MAX + 1] = {
37 	"text",
38 	"Ap",		"Dd",		"Dt",		"Os",
39 	"Sh",		"Ss",		"Pp",		"D1",
40 	"Dl",		"Bd",		"Ed",		"Bl",
41 	"El",		"It",		"Ad",		"An",
42 	"Ar",		"Cd",		"Cm",		"Dv",
43 	"Er",		"Ev",		"Ex",		"Fa",
44 	"Fd",		"Fl",		"Fn",		"Ft",
45 	"Ic",		"In",		"Li",		"Nd",
46 	"Nm",		"Op",		"Ot",		"Pa",
47 	"Rv",		"St",		"Va",		"Vt",
48 	"Xr",		"%A",		"%B",		"%D",
49 	"%I",		"%J",		"%N",		"%O",
50 	"%P",		"%R",		"%T",		"%V",
51 	"Ac",		"Ao",		"Aq",		"At",
52 	"Bc",		"Bf",		"Bo",		"Bq",
53 	"Bsx",		"Bx",		"Db",		"Dc",
54 	"Do",		"Dq",		"Ec",		"Ef",
55 	"Em",		"Eo",		"Fx",		"Ms",
56 	"No",		"Ns",		"Nx",		"Ox",
57 	"Pc",		"Pf",		"Po",		"Pq",
58 	"Qc",		"Ql",		"Qo",		"Qq",
59 	"Re",		"Rs",		"Sc",		"So",
60 	"Sq",		"Sm",		"Sx",		"Sy",
61 	"Tn",		"Ux",		"Xc",		"Xo",
62 	"Fo",		"Fc",		"Oo",		"Oc",
63 	"Bk",		"Ek",		"Bt",		"Hf",
64 	"Fr",		"Ud",		"Lb",		"Lp",
65 	"Lk",		"Mt",		"Brq",		"Bro",
66 	"Brc",		"%C",		"Es",		"En",
67 	"Dx",		"%Q",		"br",		"sp",
68 	"%U",		"Ta",		"ll",
69 };
70 
71 const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {
72 	"split",		"nosplit",		"ragged",
73 	"unfilled",		"literal",		"file",
74 	"offset",		"bullet",		"dash",
75 	"hyphen",		"item",			"enum",
76 	"tag",			"diag",			"hang",
77 	"ohang",		"inset",		"column",
78 	"width",		"compact",		"std",
79 	"filled",		"words",		"emphasis",
80 	"symbolic",		"nested",		"centered"
81 	};
82 
83 const	char * const *mdoc_macronames = __mdoc_macronames + 1;
84 const	char * const *mdoc_argnames = __mdoc_argnames;
85 
86 static	int		  mdoc_ptext(struct roff_man *, int, char *, int);
87 static	int		  mdoc_pmacro(struct roff_man *, int, char *, int);
88 
89 
90 /*
91  * Main parse routine.  Parses a single line -- really just hands off to
92  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
93  */
94 int
95 mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
96 {
97 
98 	if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
99 		mdoc->flags |= MDOC_NEWLINE;
100 
101 	/*
102 	 * Let the roff nS register switch SYNOPSIS mode early,
103 	 * such that the parser knows at all times
104 	 * whether this mode is on or off.
105 	 * Note that this mode is also switched by the Sh macro.
106 	 */
107 	if (roff_getreg(mdoc->roff, "nS"))
108 		mdoc->flags |= MDOC_SYNOPSIS;
109 	else
110 		mdoc->flags &= ~MDOC_SYNOPSIS;
111 
112 	return roff_getcontrol(mdoc->roff, buf, &offs) ?
113 	    mdoc_pmacro(mdoc, ln, buf, offs) :
114 	    mdoc_ptext(mdoc, ln, buf, offs);
115 }
116 
117 void
118 mdoc_macro(MACRO_PROT_ARGS)
119 {
120 	assert(tok > TOKEN_NONE && tok < MDOC_MAX);
121 
122 	(*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
123 }
124 
125 void
126 mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok)
127 {
128 	struct roff_node *p;
129 
130 	p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
131 	roff_node_append(mdoc, p);
132 	mdoc->next = ROFF_NEXT_CHILD;
133 }
134 
135 struct roff_node *
136 mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
137 		struct roff_node *body)
138 {
139 	struct roff_node *p;
140 
141 	body->flags |= NODE_ENDED;
142 	body->parent->flags |= NODE_ENDED;
143 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
144 	p->body = body;
145 	p->norm = body->norm;
146 	p->end = ENDBODY_SPACE;
147 	roff_node_append(mdoc, p);
148 	mdoc->next = ROFF_NEXT_SIBLING;
149 	return p;
150 }
151 
152 struct roff_node *
153 mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
154 	int tok, struct mdoc_arg *args)
155 {
156 	struct roff_node *p;
157 
158 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
159 	p->args = args;
160 	if (p->args)
161 		(args->refcnt)++;
162 
163 	switch (tok) {
164 	case MDOC_Bd:
165 	case MDOC_Bf:
166 	case MDOC_Bl:
167 	case MDOC_En:
168 	case MDOC_Rs:
169 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
170 		break;
171 	default:
172 		break;
173 	}
174 	roff_node_append(mdoc, p);
175 	mdoc->next = ROFF_NEXT_CHILD;
176 	return p;
177 }
178 
179 void
180 mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
181 	int tok, struct mdoc_arg *args)
182 {
183 	struct roff_node *p;
184 
185 	p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
186 	p->args = args;
187 	if (p->args)
188 		(args->refcnt)++;
189 
190 	switch (tok) {
191 	case MDOC_An:
192 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
193 		break;
194 	default:
195 		break;
196 	}
197 	roff_node_append(mdoc, p);
198 	mdoc->next = ROFF_NEXT_CHILD;
199 }
200 
201 void
202 mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
203 {
204 
205 	roff_node_unlink(mdoc, p);
206 	p->prev = p->next = NULL;
207 	roff_node_append(mdoc, p);
208 }
209 
210 /*
211  * Parse free-form text, that is, a line that does not begin with the
212  * control character.
213  */
214 static int
215 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
216 {
217 	struct roff_node *n;
218 	char		 *c, *ws, *end;
219 
220 	n = mdoc->last;
221 
222 	/*
223 	 * If a column list contains plain text, assume an implicit item
224 	 * macro.  This can happen one or more times at the beginning
225 	 * of such a list, intermixed with non-It mdoc macros and with
226 	 * nodes generated on the roff level, for example by tbl.
227 	 */
228 
229 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
230 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
231 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
232 	     n->parent->norm->Bl.type == LIST_column)) {
233 		mdoc->flags |= MDOC_FREECOL;
234 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
235 		return 1;
236 	}
237 
238 	/*
239 	 * Search for the beginning of unescaped trailing whitespace (ws)
240 	 * and for the first character not to be output (end).
241 	 */
242 
243 	/* FIXME: replace with strcspn(). */
244 	ws = NULL;
245 	for (c = end = buf + offs; *c; c++) {
246 		switch (*c) {
247 		case ' ':
248 			if (NULL == ws)
249 				ws = c;
250 			continue;
251 		case '\t':
252 			/*
253 			 * Always warn about trailing tabs,
254 			 * even outside literal context,
255 			 * where they should be put on the next line.
256 			 */
257 			if (NULL == ws)
258 				ws = c;
259 			/*
260 			 * Strip trailing tabs in literal context only;
261 			 * outside, they affect the next line.
262 			 */
263 			if (MDOC_LITERAL & mdoc->flags)
264 				continue;
265 			break;
266 		case '\\':
267 			/* Skip the escaped character, too, if any. */
268 			if (c[1])
269 				c++;
270 			/* FALLTHROUGH */
271 		default:
272 			ws = NULL;
273 			break;
274 		}
275 		end = c + 1;
276 	}
277 	*end = '\0';
278 
279 	if (ws)
280 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
281 		    line, (int)(ws-buf), NULL);
282 
283 	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
284 		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
285 		    line, (int)(c - buf), NULL);
286 
287 		/*
288 		 * Insert a `sp' in the case of a blank line.  Technically,
289 		 * blank lines aren't allowed, but enough manuals assume this
290 		 * behaviour that we want to work around it.
291 		 */
292 		roff_elem_alloc(mdoc, line, offs, MDOC_sp);
293 		mdoc->last->flags |= NODE_VALID | NODE_ENDED;
294 		mdoc->next = ROFF_NEXT_SIBLING;
295 		return 1;
296 	}
297 
298 	roff_word_alloc(mdoc, line, offs, buf+offs);
299 
300 	if (mdoc->flags & MDOC_LITERAL)
301 		return 1;
302 
303 	/*
304 	 * End-of-sentence check.  If the last character is an unescaped
305 	 * EOS character, then flag the node as being the end of a
306 	 * sentence.  The front-end will know how to interpret this.
307 	 */
308 
309 	assert(buf < end);
310 
311 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
312 		mdoc->last->flags |= NODE_EOS;
313 
314 	for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
315 		if (c - buf < offs + 2)
316 			continue;
317 		if (end - c < 4)
318 			break;
319 		if (isalpha((unsigned char)c[-2]) &&
320 		    isalpha((unsigned char)c[-1]) &&
321 		    c[1] == ' ' &&
322 		    isupper((unsigned char)(c[2] == ' ' ? c[3] : c[2])) &&
323 		    (c[-2] != 'n' || c[-1] != 'c') &&
324 		    (c[-2] != 'v' || c[-1] != 's'))
325 			mandoc_msg(MANDOCERR_EOS, mdoc->parse,
326 			    line, (int)(c - buf), NULL);
327 	}
328 
329 	return 1;
330 }
331 
332 /*
333  * Parse a macro line, that is, a line beginning with the control
334  * character.
335  */
336 static int
337 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
338 {
339 	struct roff_node *n;
340 	const char	 *cp;
341 	int		  tok;
342 	int		  i, sv;
343 	char		  mac[5];
344 
345 	sv = offs;
346 
347 	/*
348 	 * Copy the first word into a nil-terminated buffer.
349 	 * Stop when a space, tab, escape, or eoln is encountered.
350 	 */
351 
352 	i = 0;
353 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
354 		mac[i++] = buf[offs++];
355 
356 	mac[i] = '\0';
357 
358 	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
359 
360 	if (tok == TOKEN_NONE) {
361 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
362 		    ln, sv, buf + sv - 1);
363 		return 1;
364 	}
365 
366 	/* Skip a leading escape sequence or tab. */
367 
368 	switch (buf[offs]) {
369 	case '\\':
370 		cp = buf + offs + 1;
371 		mandoc_escape(&cp, NULL, NULL);
372 		offs = cp - buf;
373 		break;
374 	case '\t':
375 		offs++;
376 		break;
377 	default:
378 		break;
379 	}
380 
381 	/* Jump to the next non-whitespace word. */
382 
383 	while (buf[offs] == ' ')
384 		offs++;
385 
386 	/*
387 	 * Trailing whitespace.  Note that tabs are allowed to be passed
388 	 * into the parser as "text", so we only warn about spaces here.
389 	 */
390 
391 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
392 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
393 		    ln, offs - 1, NULL);
394 
395 	/*
396 	 * If an initial macro or a list invocation, divert directly
397 	 * into macro processing.
398 	 */
399 
400 	n = mdoc->last;
401 	if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
402 		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
403 		return 1;
404 	}
405 
406 	/*
407 	 * If a column list contains a non-It macro, assume an implicit
408 	 * item macro.  This can happen one or more times at the
409 	 * beginning of such a list, intermixed with text lines and
410 	 * with nodes generated on the roff level, for example by tbl.
411 	 */
412 
413 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
414 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
415 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
416 	     n->parent->norm->Bl.type == LIST_column)) {
417 		mdoc->flags |= MDOC_FREECOL;
418 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
419 		return 1;
420 	}
421 
422 	/* Normal processing of a macro. */
423 
424 	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
425 
426 	/* In quick mode (for mandocdb), abort after the NAME section. */
427 
428 	if (mdoc->quick && MDOC_Sh == tok &&
429 	    SEC_NAME != mdoc->last->sec)
430 		return 2;
431 
432 	return 1;
433 }
434 
435 enum mdelim
436 mdoc_isdelim(const char *p)
437 {
438 
439 	if ('\0' == p[0])
440 		return DELIM_NONE;
441 
442 	if ('\0' == p[1])
443 		switch (p[0]) {
444 		case '(':
445 		case '[':
446 			return DELIM_OPEN;
447 		case '|':
448 			return DELIM_MIDDLE;
449 		case '.':
450 		case ',':
451 		case ';':
452 		case ':':
453 		case '?':
454 		case '!':
455 		case ')':
456 		case ']':
457 			return DELIM_CLOSE;
458 		default:
459 			return DELIM_NONE;
460 		}
461 
462 	if ('\\' != p[0])
463 		return DELIM_NONE;
464 
465 	if (0 == strcmp(p + 1, "."))
466 		return DELIM_CLOSE;
467 	if (0 == strcmp(p + 1, "fR|\\fP"))
468 		return DELIM_MIDDLE;
469 
470 	return DELIM_NONE;
471 }
472 
473 void
474 mdoc_validate(struct roff_man *mdoc)
475 {
476 
477 	mdoc->last = mdoc->first;
478 	mdoc_node_validate(mdoc);
479 	mdoc_state_reset(mdoc);
480 }
481