xref: /openbsd-src/usr.bin/mandoc/mdoc.c (revision 0b7734b3d77bb9b21afec6f4621cae6c805dbd45) 
1 /*	$OpenBSD: mdoc.c,v 1.145 2015/10/30 19:03:36 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 	(*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 	assert(mdoc->last);
221 	n = mdoc->last;
222 
223 	/*
224 	 * Divert directly to list processing if we're encountering a
225 	 * columnar ROFFT_BLOCK with or without a prior ROFFT_BLOCK entry
226 	 * (a ROFFT_BODY means it's already open, in which case we should
227 	 * process within its context in the normal way).
228 	 */
229 
230 	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
231 	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
232 		/* `Bl' is open without any children. */
233 		mdoc->flags |= MDOC_FREECOL;
234 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
235 		return 1;
236 	}
237 
238 	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
239 	    NULL != n->parent &&
240 	    MDOC_Bl == n->parent->tok &&
241 	    LIST_column == n->parent->norm->Bl.type) {
242 		/* `Bl' has block-level `It' children. */
243 		mdoc->flags |= MDOC_FREECOL;
244 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
245 		return 1;
246 	}
247 
248 	/*
249 	 * Search for the beginning of unescaped trailing whitespace (ws)
250 	 * and for the first character not to be output (end).
251 	 */
252 
253 	/* FIXME: replace with strcspn(). */
254 	ws = NULL;
255 	for (c = end = buf + offs; *c; c++) {
256 		switch (*c) {
257 		case ' ':
258 			if (NULL == ws)
259 				ws = c;
260 			continue;
261 		case '\t':
262 			/*
263 			 * Always warn about trailing tabs,
264 			 * even outside literal context,
265 			 * where they should be put on the next line.
266 			 */
267 			if (NULL == ws)
268 				ws = c;
269 			/*
270 			 * Strip trailing tabs in literal context only;
271 			 * outside, they affect the next line.
272 			 */
273 			if (MDOC_LITERAL & mdoc->flags)
274 				continue;
275 			break;
276 		case '\\':
277 			/* Skip the escaped character, too, if any. */
278 			if (c[1])
279 				c++;
280 			/* FALLTHROUGH */
281 		default:
282 			ws = NULL;
283 			break;
284 		}
285 		end = c + 1;
286 	}
287 	*end = '\0';
288 
289 	if (ws)
290 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
291 		    line, (int)(ws-buf), NULL);
292 
293 	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
294 		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
295 		    line, (int)(c - buf), NULL);
296 
297 		/*
298 		 * Insert a `sp' in the case of a blank line.  Technically,
299 		 * blank lines aren't allowed, but enough manuals assume this
300 		 * behaviour that we want to work around it.
301 		 */
302 		roff_elem_alloc(mdoc, line, offs, MDOC_sp);
303 		mdoc->last->flags |= MDOC_VALID | MDOC_ENDED;
304 		mdoc->next = ROFF_NEXT_SIBLING;
305 		return 1;
306 	}
307 
308 	roff_word_alloc(mdoc, line, offs, buf+offs);
309 
310 	if (mdoc->flags & MDOC_LITERAL)
311 		return 1;
312 
313 	/*
314 	 * End-of-sentence check.  If the last character is an unescaped
315 	 * EOS character, then flag the node as being the end of a
316 	 * sentence.  The front-end will know how to interpret this.
317 	 */
318 
319 	assert(buf < end);
320 
321 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
322 		mdoc->last->flags |= MDOC_EOS;
323 	return 1;
324 }
325 
326 /*
327  * Parse a macro line, that is, a line beginning with the control
328  * character.
329  */
330 static int
331 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
332 {
333 	struct roff_node *n;
334 	const char	 *cp;
335 	int		  tok;
336 	int		  i, sv;
337 	char		  mac[5];
338 
339 	sv = offs;
340 
341 	/*
342 	 * Copy the first word into a nil-terminated buffer.
343 	 * Stop when a space, tab, escape, or eoln is encountered.
344 	 */
345 
346 	i = 0;
347 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
348 		mac[i++] = buf[offs++];
349 
350 	mac[i] = '\0';
351 
352 	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
353 
354 	if (tok == TOKEN_NONE) {
355 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
356 		    ln, sv, buf + sv - 1);
357 		return 1;
358 	}
359 
360 	/* Skip a leading escape sequence or tab. */
361 
362 	switch (buf[offs]) {
363 	case '\\':
364 		cp = buf + offs + 1;
365 		mandoc_escape(&cp, NULL, NULL);
366 		offs = cp - buf;
367 		break;
368 	case '\t':
369 		offs++;
370 		break;
371 	default:
372 		break;
373 	}
374 
375 	/* Jump to the next non-whitespace word. */
376 
377 	while (buf[offs] && ' ' == buf[offs])
378 		offs++;
379 
380 	/*
381 	 * Trailing whitespace.  Note that tabs are allowed to be passed
382 	 * into the parser as "text", so we only warn about spaces here.
383 	 */
384 
385 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
386 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
387 		    ln, offs - 1, NULL);
388 
389 	/*
390 	 * If an initial macro or a list invocation, divert directly
391 	 * into macro processing.
392 	 */
393 
394 	if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
395 		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
396 		return 1;
397 	}
398 
399 	n = mdoc->last;
400 	assert(mdoc->last);
401 
402 	/*
403 	 * If the first macro of a `Bl -column', open an `It' block
404 	 * context around the parsed macro.
405 	 */
406 
407 	if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
408 	    n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
409 		mdoc->flags |= MDOC_FREECOL;
410 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
411 		return 1;
412 	}
413 
414 	/*
415 	 * If we're following a block-level `It' within a `Bl -column'
416 	 * context (perhaps opened in the above block or in ptext()),
417 	 * then open an `It' block context around the parsed macro.
418 	 */
419 
420 	if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
421 	    NULL != n->parent &&
422 	    MDOC_Bl == n->parent->tok &&
423 	    LIST_column == n->parent->norm->Bl.type) {
424 		mdoc->flags |= MDOC_FREECOL;
425 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
426 		return 1;
427 	}
428 
429 	/* Normal processing of a macro. */
430 
431 	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
432 
433 	/* In quick mode (for mandocdb), abort after the NAME section. */
434 
435 	if (mdoc->quick && MDOC_Sh == tok &&
436 	    SEC_NAME != mdoc->last->sec)
437 		return 2;
438 
439 	return 1;
440 }
441 
442 enum mdelim
443 mdoc_isdelim(const char *p)
444 {
445 
446 	if ('\0' == p[0])
447 		return DELIM_NONE;
448 
449 	if ('\0' == p[1])
450 		switch (p[0]) {
451 		case '(':
452 		case '[':
453 			return DELIM_OPEN;
454 		case '|':
455 			return DELIM_MIDDLE;
456 		case '.':
457 		case ',':
458 		case ';':
459 		case ':':
460 		case '?':
461 		case '!':
462 		case ')':
463 		case ']':
464 			return DELIM_CLOSE;
465 		default:
466 			return DELIM_NONE;
467 		}
468 
469 	if ('\\' != p[0])
470 		return DELIM_NONE;
471 
472 	if (0 == strcmp(p + 1, "."))
473 		return DELIM_CLOSE;
474 	if (0 == strcmp(p + 1, "fR|\\fP"))
475 		return DELIM_MIDDLE;
476 
477 	return DELIM_NONE;
478 }
479 
480 void
481 mdoc_validate(struct roff_man *mdoc)
482 {
483 
484 	mdoc->last = mdoc->first;
485 	mdoc_node_validate(mdoc);
486 	mdoc_state_reset(mdoc);
487 }
488