xref: /openbsd-src/usr.bin/mandoc/mdoc.c (revision d1df930ffab53da22f3324c32bed7ac5709915e6) 
1 /*	$OpenBSD: mdoc.c,v 1.158 2018/08/17 20:31:52 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010, 2012-2018 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_argnames[MDOC_ARG_MAX] = {
37 	"split",		"nosplit",		"ragged",
38 	"unfilled",		"literal",		"file",
39 	"offset",		"bullet",		"dash",
40 	"hyphen",		"item",			"enum",
41 	"tag",			"diag",			"hang",
42 	"ohang",		"inset",		"column",
43 	"width",		"compact",		"std",
44 	"filled",		"words",		"emphasis",
45 	"symbolic",		"nested",		"centered"
46 };
47 const	char * const *mdoc_argnames = __mdoc_argnames;
48 
49 static	int		  mdoc_ptext(struct roff_man *, int, char *, int);
50 static	int		  mdoc_pmacro(struct roff_man *, int, char *, int);
51 
52 
53 /*
54  * Main parse routine.  Parses a single line -- really just hands off to
55  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
56  */
57 int
58 mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
59 {
60 
61 	if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
62 		mdoc->flags |= MDOC_NEWLINE;
63 
64 	/*
65 	 * Let the roff nS register switch SYNOPSIS mode early,
66 	 * such that the parser knows at all times
67 	 * whether this mode is on or off.
68 	 * Note that this mode is also switched by the Sh macro.
69 	 */
70 	if (roff_getreg(mdoc->roff, "nS"))
71 		mdoc->flags |= MDOC_SYNOPSIS;
72 	else
73 		mdoc->flags &= ~MDOC_SYNOPSIS;
74 
75 	return roff_getcontrol(mdoc->roff, buf, &offs) ?
76 	    mdoc_pmacro(mdoc, ln, buf, offs) :
77 	    mdoc_ptext(mdoc, ln, buf, offs);
78 }
79 
80 void
81 mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
82 {
83 	struct roff_node *p;
84 
85 	p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
86 	roff_node_append(mdoc, p);
87 	mdoc->next = ROFF_NEXT_CHILD;
88 }
89 
90 struct roff_node *
91 mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
92     enum roff_tok tok, struct roff_node *body)
93 {
94 	struct roff_node *p;
95 
96 	body->flags |= NODE_ENDED;
97 	body->parent->flags |= NODE_ENDED;
98 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
99 	p->body = body;
100 	p->norm = body->norm;
101 	p->end = ENDBODY_SPACE;
102 	roff_node_append(mdoc, p);
103 	mdoc->next = ROFF_NEXT_SIBLING;
104 	return p;
105 }
106 
107 struct roff_node *
108 mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
109     enum roff_tok tok, struct mdoc_arg *args)
110 {
111 	struct roff_node *p;
112 
113 	p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
114 	p->args = args;
115 	if (p->args)
116 		(args->refcnt)++;
117 
118 	switch (tok) {
119 	case MDOC_Bd:
120 	case MDOC_Bf:
121 	case MDOC_Bl:
122 	case MDOC_En:
123 	case MDOC_Rs:
124 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
125 		break;
126 	default:
127 		break;
128 	}
129 	roff_node_append(mdoc, p);
130 	mdoc->next = ROFF_NEXT_CHILD;
131 	return p;
132 }
133 
134 void
135 mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
136      enum roff_tok tok, struct mdoc_arg *args)
137 {
138 	struct roff_node *p;
139 
140 	p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
141 	p->args = args;
142 	if (p->args)
143 		(args->refcnt)++;
144 
145 	switch (tok) {
146 	case MDOC_An:
147 		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
148 		break;
149 	default:
150 		break;
151 	}
152 	roff_node_append(mdoc, p);
153 	mdoc->next = ROFF_NEXT_CHILD;
154 }
155 
156 void
157 mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
158 {
159 
160 	roff_node_unlink(mdoc, p);
161 	p->prev = p->next = NULL;
162 	roff_node_append(mdoc, p);
163 }
164 
165 /*
166  * Parse free-form text, that is, a line that does not begin with the
167  * control character.
168  */
169 static int
170 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
171 {
172 	struct roff_node *n;
173 	const char	 *cp, *sp;
174 	char		 *c, *ws, *end;
175 
176 	n = mdoc->last;
177 
178 	/*
179 	 * If a column list contains plain text, assume an implicit item
180 	 * macro.  This can happen one or more times at the beginning
181 	 * of such a list, intermixed with non-It mdoc macros and with
182 	 * nodes generated on the roff level, for example by tbl.
183 	 */
184 
185 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
186 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
187 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
188 	     n->parent->norm->Bl.type == LIST_column)) {
189 		mdoc->flags |= MDOC_FREECOL;
190 		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
191 		    line, offs, &offs, buf);
192 		return 1;
193 	}
194 
195 	/*
196 	 * Search for the beginning of unescaped trailing whitespace (ws)
197 	 * and for the first character not to be output (end).
198 	 */
199 
200 	/* FIXME: replace with strcspn(). */
201 	ws = NULL;
202 	for (c = end = buf + offs; *c; c++) {
203 		switch (*c) {
204 		case ' ':
205 			if (NULL == ws)
206 				ws = c;
207 			continue;
208 		case '\t':
209 			/*
210 			 * Always warn about trailing tabs,
211 			 * even outside literal context,
212 			 * where they should be put on the next line.
213 			 */
214 			if (NULL == ws)
215 				ws = c;
216 			/*
217 			 * Strip trailing tabs in literal context only;
218 			 * outside, they affect the next line.
219 			 */
220 			if (MDOC_LITERAL & mdoc->flags)
221 				continue;
222 			break;
223 		case '\\':
224 			/* Skip the escaped character, too, if any. */
225 			if (c[1])
226 				c++;
227 			/* FALLTHROUGH */
228 		default:
229 			ws = NULL;
230 			break;
231 		}
232 		end = c + 1;
233 	}
234 	*end = '\0';
235 
236 	if (ws)
237 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
238 		    line, (int)(ws-buf), NULL);
239 
240 	/*
241 	 * Blank lines are allowed in no-fill mode
242 	 * and cancel preceding \c,
243 	 * but add a single vertical space elsewhere.
244 	 */
245 
246 	if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
247 		switch (mdoc->last->type) {
248 		case ROFFT_TEXT:
249 			sp = mdoc->last->string;
250 			cp = end = strchr(sp, '\0') - 2;
251 			if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
252 				break;
253 			while (cp > sp && cp[-1] == '\\')
254 				cp--;
255 			if ((end - cp) % 2)
256 				break;
257 			*end = '\0';
258 			return 1;
259 		default:
260 			break;
261 		}
262 		mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
263 		    line, (int)(c - buf), NULL);
264 		roff_elem_alloc(mdoc, line, offs, ROFF_sp);
265 		mdoc->last->flags |= NODE_VALID | NODE_ENDED;
266 		mdoc->next = ROFF_NEXT_SIBLING;
267 		return 1;
268 	}
269 
270 	roff_word_alloc(mdoc, line, offs, buf+offs);
271 
272 	if (mdoc->flags & MDOC_LITERAL)
273 		return 1;
274 
275 	/*
276 	 * End-of-sentence check.  If the last character is an unescaped
277 	 * EOS character, then flag the node as being the end of a
278 	 * sentence.  The front-end will know how to interpret this.
279 	 */
280 
281 	assert(buf < end);
282 
283 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
284 		mdoc->last->flags |= NODE_EOS;
285 
286 	for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
287 		if (c - buf < offs + 2)
288 			continue;
289 		if (end - c < 3)
290 			break;
291 		if (c[1] != ' ' ||
292 		    isalnum((unsigned char)c[-2]) == 0 ||
293 		    isalnum((unsigned char)c[-1]) == 0 ||
294 		    (c[-2] == 'n' && c[-1] == 'c') ||
295 		    (c[-2] == 'v' && c[-1] == 's'))
296 			continue;
297 		c += 2;
298 		if (*c == ' ')
299 			c++;
300 		if (*c == ' ')
301 			c++;
302 		if (isupper((unsigned char)(*c)))
303 			mandoc_msg(MANDOCERR_EOS, mdoc->parse,
304 			    line, (int)(c - buf), NULL);
305 	}
306 
307 	return 1;
308 }
309 
310 /*
311  * Parse a macro line, that is, a line beginning with the control
312  * character.
313  */
314 static int
315 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
316 {
317 	struct roff_node *n;
318 	const char	 *cp;
319 	size_t		  sz;
320 	enum roff_tok	  tok;
321 	int		  sv;
322 
323 	/* Determine the line macro. */
324 
325 	sv = offs;
326 	tok = TOKEN_NONE;
327 	for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
328 		offs++;
329 	if (sz == 2 || sz == 3)
330 		tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
331 	if (tok == TOKEN_NONE) {
332 		mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
333 		    ln, sv, buf + sv - 1);
334 		return 1;
335 	}
336 
337 	/* Skip a leading escape sequence or tab. */
338 
339 	switch (buf[offs]) {
340 	case '\\':
341 		cp = buf + offs + 1;
342 		mandoc_escape(&cp, NULL, NULL);
343 		offs = cp - buf;
344 		break;
345 	case '\t':
346 		offs++;
347 		break;
348 	default:
349 		break;
350 	}
351 
352 	/* Jump to the next non-whitespace word. */
353 
354 	while (buf[offs] == ' ')
355 		offs++;
356 
357 	/*
358 	 * Trailing whitespace.  Note that tabs are allowed to be passed
359 	 * into the parser as "text", so we only warn about spaces here.
360 	 */
361 
362 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
363 		mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
364 		    ln, offs - 1, NULL);
365 
366 	/*
367 	 * If an initial macro or a list invocation, divert directly
368 	 * into macro processing.
369 	 */
370 
371 	n = mdoc->last;
372 	if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
373 		(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
374 		return 1;
375 	}
376 
377 	/*
378 	 * If a column list contains a non-It macro, assume an implicit
379 	 * item macro.  This can happen one or more times at the
380 	 * beginning of such a list, intermixed with text lines and
381 	 * with nodes generated on the roff level, for example by tbl.
382 	 */
383 
384 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
385 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
386 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
387 	     n->parent->norm->Bl.type == LIST_column)) {
388 		mdoc->flags |= MDOC_FREECOL;
389 		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
390 		return 1;
391 	}
392 
393 	/* Normal processing of a macro. */
394 
395 	(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
396 
397 	/* In quick mode (for mandocdb), abort after the NAME section. */
398 
399 	if (mdoc->quick && MDOC_Sh == tok &&
400 	    SEC_NAME != mdoc->last->sec)
401 		return 2;
402 
403 	return 1;
404 }
405 
406 enum mdelim
407 mdoc_isdelim(const char *p)
408 {
409 
410 	if ('\0' == p[0])
411 		return DELIM_NONE;
412 
413 	if ('\0' == p[1])
414 		switch (p[0]) {
415 		case '(':
416 		case '[':
417 			return DELIM_OPEN;
418 		case '|':
419 			return DELIM_MIDDLE;
420 		case '.':
421 		case ',':
422 		case ';':
423 		case ':':
424 		case '?':
425 		case '!':
426 		case ')':
427 		case ']':
428 			return DELIM_CLOSE;
429 		default:
430 			return DELIM_NONE;
431 		}
432 
433 	if ('\\' != p[0])
434 		return DELIM_NONE;
435 
436 	if (0 == strcmp(p + 1, "."))
437 		return DELIM_CLOSE;
438 	if (0 == strcmp(p + 1, "fR|\\fP"))
439 		return DELIM_MIDDLE;
440 
441 	return DELIM_NONE;
442 }
443 
444 void
445 mdoc_validate(struct roff_man *mdoc)
446 {
447 
448 	mdoc->last = mdoc->first;
449 	mdoc_node_validate(mdoc);
450 	mdoc_state_reset(mdoc);
451 }
452