xref: /openbsd/usr.bin/mandoc/mdoc.c (revision 998de4a5) 
1 /*	$OpenBSD: mdoc.c,v 1.146 2016/08/20 14:43:39 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010, 2012-2016 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, enum mdoc_endbody end)
138 {
139 	struct roff_node *p;
140 
141 	body->flags |= MDOC_ENDED;
142 	body->parent->flags |= MDOC_ENDED;
143 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
144 	p->body = body;
145 	p->norm = body->norm;
146 	p->end = end;
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 |= MDOC_VALID | MDOC_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 |= MDOC_EOS;
313 	return 1;
314 }
315 
316 /*
317  * Parse a macro line, that is, a line beginning with the control
318  * character.
319  */
320 static int
321 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
322 {
323 	struct roff_node *n;
324 	const char	 *cp;
325 	int		  tok;
326 	int		  i, sv;
327 	char		  mac[5];
328 
329 	sv = offs;
330 
331 	/*
332 	 * Copy the first word into a nil-terminated buffer.
333 	 * Stop when a space, tab, escape, or eoln is encountered.
334 	 */
335 
336 	i = 0;
337 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
338 		mac[i++] = buf[offs++];
339 
340 	mac[i] = '\0';
341 
342 	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
343 
344 	if (tok == TOKEN_NONE) {
345 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
346 		    ln, sv, buf + sv - 1);
347 		return 1;
348 	}
349 
350 	/* Skip a leading escape sequence or tab. */
351 
352 	switch (buf[offs]) {
353 	case '\\':
354 		cp = buf + offs + 1;
355 		mandoc_escape(&cp, NULL, NULL);
356 		offs = cp - buf;
357 		break;
358 	case '\t':
359 		offs++;
360 		break;
361 	default:
362 		break;
363 	}
364 
365 	/* Jump to the next non-whitespace word. */
366 
367 	while (buf[offs] && ' ' == buf[offs])
368 		offs++;
369 
370 	/*
371 	 * Trailing whitespace.  Note that tabs are allowed to be passed
372 	 * into the parser as "text", so we only warn about spaces here.
373 	 */
374 
375 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
376 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
377 		    ln, offs - 1, NULL);
378 
379 	/*
380 	 * If an initial macro or a list invocation, divert directly
381 	 * into macro processing.
382 	 */
383 
384 	n = mdoc->last;
385 	if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
386 		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
387 		return 1;
388 	}
389 
390 	/*
391 	 * If a column list contains a non-It macro, assume an implicit
392 	 * item macro.  This can happen one or more times at the
393 	 * beginning of such a list, intermixed with text lines and
394 	 * with nodes generated on the roff level, for example by tbl.
395 	 */
396 
397 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
398 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
399 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
400 	     n->parent->norm->Bl.type == LIST_column)) {
401 		mdoc->flags |= MDOC_FREECOL;
402 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
403 		return 1;
404 	}
405 
406 	/* Normal processing of a macro. */
407 
408 	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
409 
410 	/* In quick mode (for mandocdb), abort after the NAME section. */
411 
412 	if (mdoc->quick && MDOC_Sh == tok &&
413 	    SEC_NAME != mdoc->last->sec)
414 		return 2;
415 
416 	return 1;
417 }
418 
419 enum mdelim
420 mdoc_isdelim(const char *p)
421 {
422 
423 	if ('\0' == p[0])
424 		return DELIM_NONE;
425 
426 	if ('\0' == p[1])
427 		switch (p[0]) {
428 		case '(':
429 		case '[':
430 			return DELIM_OPEN;
431 		case '|':
432 			return DELIM_MIDDLE;
433 		case '.':
434 		case ',':
435 		case ';':
436 		case ':':
437 		case '?':
438 		case '!':
439 		case ')':
440 		case ']':
441 			return DELIM_CLOSE;
442 		default:
443 			return DELIM_NONE;
444 		}
445 
446 	if ('\\' != p[0])
447 		return DELIM_NONE;
448 
449 	if (0 == strcmp(p + 1, "."))
450 		return DELIM_CLOSE;
451 	if (0 == strcmp(p + 1, "fR|\\fP"))
452 		return DELIM_MIDDLE;
453 
454 	return DELIM_NONE;
455 }
456 
457 void
458 mdoc_validate(struct roff_man *mdoc)
459 {
460 
461 	mdoc->last = mdoc->first;
462 	mdoc_node_validate(mdoc);
463 	mdoc_state_reset(mdoc);
464 }
465