xref: /illumos-gate/usr/src/cmd/mandoc/mdoc.c (revision a40ea1a7d80eee1b409e9dcc2e48c730988147ea) 
1 /*	$Id: mdoc.c,v 1.260 2017/02/16 03:00:23 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 "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 
38 const	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 
73 const	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 
85 const	char * const *mdoc_macronames = __mdoc_macronames + 1;
86 const	char * const *mdoc_argnames = __mdoc_argnames;
87 
88 static	int		  mdoc_ptext(struct roff_man *, int, char *, int);
89 static	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  */
96 int
97 mdoc_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 
119 void
120 mdoc_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 
127 void
128 mdoc_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 
137 struct roff_node *
138 mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
139 		struct roff_node *body)
140 {
141 	struct roff_node *p;
142 
143 	body->flags |= NODE_ENDED;
144 	body->parent->flags |= NODE_ENDED;
145 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
146 	p->body = body;
147 	p->norm = body->norm;
148 	p->end = ENDBODY_SPACE;
149 	roff_node_append(mdoc, p);
150 	mdoc->next = ROFF_NEXT_SIBLING;
151 	return p;
152 }
153 
154 struct roff_node *
155 mdoc_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 
181 void
182 mdoc_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 
203 void
204 mdoc_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  */
216 static int
217 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
218 {
219 	struct roff_node *n;
220 	char		 *c, *ws, *end;
221 
222 	n = mdoc->last;
223 
224 	/*
225 	 * If a column list contains plain text, assume an implicit item
226 	 * macro.  This can happen one or more times at the beginning
227 	 * of such a list, intermixed with non-It mdoc macros and with
228 	 * nodes generated on the roff level, for example by tbl.
229 	 */
230 
231 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
232 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
233 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
234 	     n->parent->norm->Bl.type == LIST_column)) {
235 		mdoc->flags |= MDOC_FREECOL;
236 		mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
237 		return 1;
238 	}
239 
240 	/*
241 	 * Search for the beginning of unescaped trailing whitespace (ws)
242 	 * and for the first character not to be output (end).
243 	 */
244 
245 	/* FIXME: replace with strcspn(). */
246 	ws = NULL;
247 	for (c = end = buf + offs; *c; c++) {
248 		switch (*c) {
249 		case ' ':
250 			if (NULL == ws)
251 				ws = c;
252 			continue;
253 		case '\t':
254 			/*
255 			 * Always warn about trailing tabs,
256 			 * even outside literal context,
257 			 * where they should be put on the next line.
258 			 */
259 			if (NULL == ws)
260 				ws = c;
261 			/*
262 			 * Strip trailing tabs in literal context only;
263 			 * outside, they affect the next line.
264 			 */
265 			if (MDOC_LITERAL & mdoc->flags)
266 				continue;
267 			break;
268 		case '\\':
269 			/* Skip the escaped character, too, if any. */
270 			if (c[1])
271 				c++;
272 			/* FALLTHROUGH */
273 		default:
274 			ws = NULL;
275 			break;
276 		}
277 		end = c + 1;
278 	}
279 	*end = '\0';
280 
281 	if (ws)
282 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
283 		    line, (int)(ws-buf), NULL);
284 
285 	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
286 		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
287 		    line, (int)(c - buf), NULL);
288 
289 		/*
290 		 * Insert a `sp' in the case of a blank line.  Technically,
291 		 * blank lines aren't allowed, but enough manuals assume this
292 		 * behaviour that we want to work around it.
293 		 */
294 		roff_elem_alloc(mdoc, line, offs, MDOC_sp);
295 		mdoc->last->flags |= NODE_VALID | NODE_ENDED;
296 		mdoc->next = ROFF_NEXT_SIBLING;
297 		return 1;
298 	}
299 
300 	roff_word_alloc(mdoc, line, offs, buf+offs);
301 
302 	if (mdoc->flags & MDOC_LITERAL)
303 		return 1;
304 
305 	/*
306 	 * End-of-sentence check.  If the last character is an unescaped
307 	 * EOS character, then flag the node as being the end of a
308 	 * sentence.  The front-end will know how to interpret this.
309 	 */
310 
311 	assert(buf < end);
312 
313 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
314 		mdoc->last->flags |= NODE_EOS;
315 
316 	for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
317 		if (c - buf < offs + 2)
318 			continue;
319 		if (end - c < 4)
320 			break;
321 		if (isalpha((unsigned char)c[-2]) &&
322 		    isalpha((unsigned char)c[-1]) &&
323 		    c[1] == ' ' &&
324 		    isupper((unsigned char)(c[2] == ' ' ? c[3] : c[2])) &&
325 		    (c[-2] != 'n' || c[-1] != 'c') &&
326 		    (c[-2] != 'v' || c[-1] != 's'))
327 			mandoc_msg(MANDOCERR_EOS, mdoc->parse,
328 			    line, (int)(c - buf), NULL);
329 	}
330 
331 	return 1;
332 }
333 
334 /*
335  * Parse a macro line, that is, a line beginning with the control
336  * character.
337  */
338 static int
339 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
340 {
341 	struct roff_node *n;
342 	const char	 *cp;
343 	int		  tok;
344 	int		  i, sv;
345 	char		  mac[5];
346 
347 	sv = offs;
348 
349 	/*
350 	 * Copy the first word into a nil-terminated buffer.
351 	 * Stop when a space, tab, escape, or eoln is encountered.
352 	 */
353 
354 	i = 0;
355 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
356 		mac[i++] = buf[offs++];
357 
358 	mac[i] = '\0';
359 
360 	tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
361 
362 	if (tok == TOKEN_NONE) {
363 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
364 		    ln, sv, buf + sv - 1);
365 		return 1;
366 	}
367 
368 	/* Skip a leading escape sequence or tab. */
369 
370 	switch (buf[offs]) {
371 	case '\\':
372 		cp = buf + offs + 1;
373 		mandoc_escape(&cp, NULL, NULL);
374 		offs = cp - buf;
375 		break;
376 	case '\t':
377 		offs++;
378 		break;
379 	default:
380 		break;
381 	}
382 
383 	/* Jump to the next non-whitespace word. */
384 
385 	while (buf[offs] && ' ' == buf[offs])
386 		offs++;
387 
388 	/*
389 	 * Trailing whitespace.  Note that tabs are allowed to be passed
390 	 * into the parser as "text", so we only warn about spaces here.
391 	 */
392 
393 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
394 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
395 		    ln, offs - 1, NULL);
396 
397 	/*
398 	 * If an initial macro or a list invocation, divert directly
399 	 * into macro processing.
400 	 */
401 
402 	n = mdoc->last;
403 	if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
404 		mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
405 		return 1;
406 	}
407 
408 	/*
409 	 * If a column list contains a non-It macro, assume an implicit
410 	 * item macro.  This can happen one or more times at the
411 	 * beginning of such a list, intermixed with text lines and
412 	 * with nodes generated on the roff level, for example by tbl.
413 	 */
414 
415 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
416 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
417 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
418 	     n->parent->norm->Bl.type == LIST_column)) {
419 		mdoc->flags |= MDOC_FREECOL;
420 		mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
421 		return 1;
422 	}
423 
424 	/* Normal processing of a macro. */
425 
426 	mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
427 
428 	/* In quick mode (for mandocdb), abort after the NAME section. */
429 
430 	if (mdoc->quick && MDOC_Sh == tok &&
431 	    SEC_NAME != mdoc->last->sec)
432 		return 2;
433 
434 	return 1;
435 }
436 
437 enum mdelim
438 mdoc_isdelim(const char *p)
439 {
440 
441 	if ('\0' == p[0])
442 		return DELIM_NONE;
443 
444 	if ('\0' == p[1])
445 		switch (p[0]) {
446 		case '(':
447 		case '[':
448 			return DELIM_OPEN;
449 		case '|':
450 			return DELIM_MIDDLE;
451 		case '.':
452 		case ',':
453 		case ';':
454 		case ':':
455 		case '?':
456 		case '!':
457 		case ')':
458 		case ']':
459 			return DELIM_CLOSE;
460 		default:
461 			return DELIM_NONE;
462 		}
463 
464 	if ('\\' != p[0])
465 		return DELIM_NONE;
466 
467 	if (0 == strcmp(p + 1, "."))
468 		return DELIM_CLOSE;
469 	if (0 == strcmp(p + 1, "fR|\\fP"))
470 		return DELIM_MIDDLE;
471 
472 	return DELIM_NONE;
473 }
474 
475 void
476 mdoc_validate(struct roff_man *mdoc)
477 {
478 
479 	mdoc->last = mdoc->first;
480 	mdoc_node_validate(mdoc);
481 	mdoc_state_reset(mdoc);
482 }
483