xref: /openbsd-src/usr.bin/mandoc/mdoc.c (revision cb39b41371628601fbe4c618205356d538b9d08a) 
1 /*	$OpenBSD: mdoc.c,v 1.141 2015/04/23 16:17:04 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 <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 	if (mdoc->flags & MDOC_PBODY) {
123 		if (tok == MDOC_Dt) {
124 			mandoc_vmsg(MANDOCERR_DT_LATE,
125 			    mdoc->parse, line, ppos,
126 			    "Dt %s", buf + *pos);
127 			return;
128 		}
129 	} else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
130 		if (mdoc->meta.title == NULL) {
131 			mandoc_vmsg(MANDOCERR_DT_NOTITLE,
132 			    mdoc->parse, line, ppos, "%s %s",
133 			    mdoc_macronames[tok], buf + *pos);
134 			mdoc->meta.title = mandoc_strdup("UNTITLED");
135 		}
136 		if (NULL == mdoc->meta.vol)
137 			mdoc->meta.vol = mandoc_strdup("LOCAL");
138 		mdoc->flags |= MDOC_PBODY;
139 	}
140 	(*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
141 }
142 
143 void
144 mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok)
145 {
146 	struct roff_node *p;
147 
148 	p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
149 	roff_node_append(mdoc, p);
150 	mdoc->next = ROFF_NEXT_CHILD;
151 }
152 
153 struct roff_node *
154 mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
155 		struct roff_node *body, enum mdoc_endbody end)
156 {
157 	struct roff_node *p;
158 
159 	body->flags |= MDOC_ENDED;
160 	body->parent->flags |= MDOC_ENDED;
161 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
162 	p->body = body;
163 	p->norm = body->norm;
164 	p->end = end;
165 	roff_node_append(mdoc, p);
166 	mdoc->next = ROFF_NEXT_SIBLING;
167 	return(p);
168 }
169 
170 struct roff_node *
171 mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
172 	int tok, struct mdoc_arg *args)
173 {
174 	struct roff_node *p;
175 
176 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
177 	p->args = args;
178 	if (p->args)
179 		(args->refcnt)++;
180 
181 	switch (tok) {
182 	case MDOC_Bd:
183 		/* FALLTHROUGH */
184 	case MDOC_Bf:
185 		/* FALLTHROUGH */
186 	case MDOC_Bl:
187 		/* FALLTHROUGH */
188 	case MDOC_En:
189 		/* FALLTHROUGH */
190 	case MDOC_Rs:
191 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
192 		break;
193 	default:
194 		break;
195 	}
196 	roff_node_append(mdoc, p);
197 	mdoc->next = ROFF_NEXT_CHILD;
198 	return(p);
199 }
200 
201 void
202 mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
203 	int tok, struct mdoc_arg *args)
204 {
205 	struct roff_node *p;
206 
207 	p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
208 	p->args = args;
209 	if (p->args)
210 		(args->refcnt)++;
211 
212 	switch (tok) {
213 	case MDOC_An:
214 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
215 		break;
216 	default:
217 		break;
218 	}
219 	roff_node_append(mdoc, p);
220 	mdoc->next = ROFF_NEXT_CHILD;
221 }
222 
223 void
224 mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
225 {
226 
227 	roff_node_unlink(mdoc, p);
228 	roff_node_append(mdoc, p);
229 }
230 
231 /*
232  * Parse free-form text, that is, a line that does not begin with the
233  * control character.
234  */
235 static int
236 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
237 {
238 	struct roff_node *n;
239 	char		 *c, *ws, *end;
240 
241 	assert(mdoc->last);
242 	n = mdoc->last;
243 
244 	/*
245 	 * Divert directly to list processing if we're encountering a
246 	 * columnar ROFFT_BLOCK with or without a prior ROFFT_BLOCK entry
247 	 * (a ROFFT_BODY means it's already open, in which case we should
248 	 * process within its context in the normal way).
249 	 */
250 
251 	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
252 	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
253 		/* `Bl' is open without any children. */
254 		mdoc->flags |= MDOC_FREECOL;
255 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
256 		return(1);
257 	}
258 
259 	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
260 	    NULL != n->parent &&
261 	    MDOC_Bl == n->parent->tok &&
262 	    LIST_column == n->parent->norm->Bl.type) {
263 		/* `Bl' has block-level `It' children. */
264 		mdoc->flags |= MDOC_FREECOL;
265 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
266 		return(1);
267 	}
268 
269 	/*
270 	 * Search for the beginning of unescaped trailing whitespace (ws)
271 	 * and for the first character not to be output (end).
272 	 */
273 
274 	/* FIXME: replace with strcspn(). */
275 	ws = NULL;
276 	for (c = end = buf + offs; *c; c++) {
277 		switch (*c) {
278 		case ' ':
279 			if (NULL == ws)
280 				ws = c;
281 			continue;
282 		case '\t':
283 			/*
284 			 * Always warn about trailing tabs,
285 			 * even outside literal context,
286 			 * where they should be put on the next line.
287 			 */
288 			if (NULL == ws)
289 				ws = c;
290 			/*
291 			 * Strip trailing tabs in literal context only;
292 			 * outside, they affect the next line.
293 			 */
294 			if (MDOC_LITERAL & mdoc->flags)
295 				continue;
296 			break;
297 		case '\\':
298 			/* Skip the escaped character, too, if any. */
299 			if (c[1])
300 				c++;
301 			/* FALLTHROUGH */
302 		default:
303 			ws = NULL;
304 			break;
305 		}
306 		end = c + 1;
307 	}
308 	*end = '\0';
309 
310 	if (ws)
311 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
312 		    line, (int)(ws-buf), NULL);
313 
314 	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
315 		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
316 		    line, (int)(c - buf), NULL);
317 
318 		/*
319 		 * Insert a `sp' in the case of a blank line.  Technically,
320 		 * blank lines aren't allowed, but enough manuals assume this
321 		 * behaviour that we want to work around it.
322 		 */
323 		roff_elem_alloc(mdoc, line, offs, MDOC_sp);
324 		mdoc->next = ROFF_NEXT_SIBLING;
325 		mdoc_valid_post(mdoc);
326 		return(1);
327 	}
328 
329 	roff_word_alloc(mdoc, line, offs, buf+offs);
330 
331 	if (mdoc->flags & MDOC_LITERAL)
332 		return(1);
333 
334 	/*
335 	 * End-of-sentence check.  If the last character is an unescaped
336 	 * EOS character, then flag the node as being the end of a
337 	 * sentence.  The front-end will know how to interpret this.
338 	 */
339 
340 	assert(buf < end);
341 
342 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
343 		mdoc->last->flags |= MDOC_EOS;
344 	return(1);
345 }
346 
347 /*
348  * Parse a macro line, that is, a line beginning with the control
349  * character.
350  */
351 static int
352 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
353 {
354 	struct roff_node *n;
355 	const char	 *cp;
356 	int		  tok;
357 	int		  i, sv;
358 	char		  mac[5];
359 
360 	sv = offs;
361 
362 	/*
363 	 * Copy the first word into a nil-terminated buffer.
364 	 * Stop when a space, tab, escape, or eoln is encountered.
365 	 */
366 
367 	i = 0;
368 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
369 		mac[i++] = buf[offs++];
370 
371 	mac[i] = '\0';
372 
373 	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
374 
375 	if (tok == TOKEN_NONE) {
376 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
377 		    ln, sv, buf + sv - 1);
378 		return(1);
379 	}
380 
381 	/* Skip a leading escape sequence or tab. */
382 
383 	switch (buf[offs]) {
384 	case '\\':
385 		cp = buf + offs + 1;
386 		mandoc_escape(&cp, NULL, NULL);
387 		offs = cp - buf;
388 		break;
389 	case '\t':
390 		offs++;
391 		break;
392 	default:
393 		break;
394 	}
395 
396 	/* Jump to the next non-whitespace word. */
397 
398 	while (buf[offs] && ' ' == buf[offs])
399 		offs++;
400 
401 	/*
402 	 * Trailing whitespace.  Note that tabs are allowed to be passed
403 	 * into the parser as "text", so we only warn about spaces here.
404 	 */
405 
406 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
407 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
408 		    ln, offs - 1, NULL);
409 
410 	/*
411 	 * If an initial macro or a list invocation, divert directly
412 	 * into macro processing.
413 	 */
414 
415 	if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
416 		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
417 		return(1);
418 	}
419 
420 	n = mdoc->last;
421 	assert(mdoc->last);
422 
423 	/*
424 	 * If the first macro of a `Bl -column', open an `It' block
425 	 * context around the parsed macro.
426 	 */
427 
428 	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
429 	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
430 		mdoc->flags |= MDOC_FREECOL;
431 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
432 		return(1);
433 	}
434 
435 	/*
436 	 * If we're following a block-level `It' within a `Bl -column'
437 	 * context (perhaps opened in the above block or in ptext()),
438 	 * then open an `It' block context around the parsed macro.
439 	 */
440 
441 	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
442 	    NULL != n->parent &&
443 	    MDOC_Bl == n->parent->tok &&
444 	    LIST_column == n->parent->norm->Bl.type) {
445 		mdoc->flags |= MDOC_FREECOL;
446 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
447 		return(1);
448 	}
449 
450 	/* Normal processing of a macro. */
451 
452 	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
453 
454 	/* In quick mode (for mandocdb), abort after the NAME section. */
455 
456 	if (mdoc->quick && MDOC_Sh == tok &&
457 	    SEC_NAME != mdoc->last->sec)
458 		return(2);
459 
460 	return(1);
461 }
462 
463 enum mdelim
464 mdoc_isdelim(const char *p)
465 {
466 
467 	if ('\0' == p[0])
468 		return(DELIM_NONE);
469 
470 	if ('\0' == p[1])
471 		switch (p[0]) {
472 		case '(':
473 			/* FALLTHROUGH */
474 		case '[':
475 			return(DELIM_OPEN);
476 		case '|':
477 			return(DELIM_MIDDLE);
478 		case '.':
479 			/* FALLTHROUGH */
480 		case ',':
481 			/* FALLTHROUGH */
482 		case ';':
483 			/* FALLTHROUGH */
484 		case ':':
485 			/* FALLTHROUGH */
486 		case '?':
487 			/* FALLTHROUGH */
488 		case '!':
489 			/* FALLTHROUGH */
490 		case ')':
491 			/* FALLTHROUGH */
492 		case ']':
493 			return(DELIM_CLOSE);
494 		default:
495 			return(DELIM_NONE);
496 		}
497 
498 	if ('\\' != p[0])
499 		return(DELIM_NONE);
500 
501 	if (0 == strcmp(p + 1, "."))
502 		return(DELIM_CLOSE);
503 	if (0 == strcmp(p + 1, "fR|\\fP"))
504 		return(DELIM_MIDDLE);
505 
506 	return(DELIM_NONE);
507 }
508