xref: /openbsd-src/usr.bin/mandoc/mdoc.c (revision 99fd087599a8791921855f21bd7e36130f39aadc) 
1 /*	$OpenBSD: mdoc.c,v 1.163 2018/12/31 07:45:42 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 /*
157  * Parse free-form text, that is, a line that does not begin with the
158  * control character.
159  */
160 static int
161 mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
162 {
163 	struct roff_node *n;
164 	const char	 *cp, *sp;
165 	char		 *c, *ws, *end;
166 
167 	n = mdoc->last;
168 
169 	/*
170 	 * If a column list contains plain text, assume an implicit item
171 	 * macro.  This can happen one or more times at the beginning
172 	 * of such a list, intermixed with non-It mdoc macros and with
173 	 * nodes generated on the roff level, for example by tbl.
174 	 */
175 
176 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
177 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
178 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
179 	     n->parent->norm->Bl.type == LIST_column)) {
180 		mdoc->flags |= MDOC_FREECOL;
181 		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
182 		    line, offs, &offs, buf);
183 		return 1;
184 	}
185 
186 	/*
187 	 * Search for the beginning of unescaped trailing whitespace (ws)
188 	 * and for the first character not to be output (end).
189 	 */
190 
191 	/* FIXME: replace with strcspn(). */
192 	ws = NULL;
193 	for (c = end = buf + offs; *c; c++) {
194 		switch (*c) {
195 		case ' ':
196 			if (NULL == ws)
197 				ws = c;
198 			continue;
199 		case '\t':
200 			/*
201 			 * Always warn about trailing tabs,
202 			 * even outside literal context,
203 			 * where they should be put on the next line.
204 			 */
205 			if (NULL == ws)
206 				ws = c;
207 			/*
208 			 * Strip trailing tabs in literal context only;
209 			 * outside, they affect the next line.
210 			 */
211 			if (mdoc->flags & ROFF_NOFILL)
212 				continue;
213 			break;
214 		case '\\':
215 			/* Skip the escaped character, too, if any. */
216 			if (c[1])
217 				c++;
218 			/* FALLTHROUGH */
219 		default:
220 			ws = NULL;
221 			break;
222 		}
223 		end = c + 1;
224 	}
225 	*end = '\0';
226 
227 	if (ws)
228 		mandoc_msg(MANDOCERR_SPACE_EOL, line, (int)(ws - buf), NULL);
229 
230 	/*
231 	 * Blank lines are allowed in no-fill mode
232 	 * and cancel preceding \c,
233 	 * but add a single vertical space elsewhere.
234 	 */
235 
236 	if (buf[offs] == '\0' && (mdoc->flags & ROFF_NOFILL) == 0) {
237 		switch (mdoc->last->type) {
238 		case ROFFT_TEXT:
239 			sp = mdoc->last->string;
240 			cp = end = strchr(sp, '\0') - 2;
241 			if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
242 				break;
243 			while (cp > sp && cp[-1] == '\\')
244 				cp--;
245 			if ((end - cp) % 2)
246 				break;
247 			*end = '\0';
248 			return 1;
249 		default:
250 			break;
251 		}
252 		mandoc_msg(MANDOCERR_FI_BLANK, line, (int)(c - buf), NULL);
253 		roff_elem_alloc(mdoc, line, offs, ROFF_sp);
254 		mdoc->last->flags |= NODE_VALID | NODE_ENDED;
255 		mdoc->next = ROFF_NEXT_SIBLING;
256 		return 1;
257 	}
258 
259 	roff_word_alloc(mdoc, line, offs, buf+offs);
260 
261 	if (mdoc->flags & ROFF_NOFILL)
262 		return 1;
263 
264 	/*
265 	 * End-of-sentence check.  If the last character is an unescaped
266 	 * EOS character, then flag the node as being the end of a
267 	 * sentence.  The front-end will know how to interpret this.
268 	 */
269 
270 	assert(buf < end);
271 
272 	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
273 		mdoc->last->flags |= NODE_EOS;
274 
275 	for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
276 		if (c - buf < offs + 2)
277 			continue;
278 		if (end - c < 3)
279 			break;
280 		if (c[1] != ' ' ||
281 		    isalnum((unsigned char)c[-2]) == 0 ||
282 		    isalnum((unsigned char)c[-1]) == 0 ||
283 		    (c[-2] == 'n' && c[-1] == 'c') ||
284 		    (c[-2] == 'v' && c[-1] == 's'))
285 			continue;
286 		c += 2;
287 		if (*c == ' ')
288 			c++;
289 		if (*c == ' ')
290 			c++;
291 		if (isupper((unsigned char)(*c)))
292 			mandoc_msg(MANDOCERR_EOS, line, (int)(c - buf), NULL);
293 	}
294 
295 	return 1;
296 }
297 
298 /*
299  * Parse a macro line, that is, a line beginning with the control
300  * character.
301  */
302 static int
303 mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
304 {
305 	struct roff_node *n;
306 	const char	 *cp;
307 	size_t		  sz;
308 	enum roff_tok	  tok;
309 	int		  sv;
310 
311 	/* Determine the line macro. */
312 
313 	sv = offs;
314 	tok = TOKEN_NONE;
315 	for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
316 		offs++;
317 	if (sz == 2 || sz == 3)
318 		tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
319 	if (tok == TOKEN_NONE) {
320 		mandoc_msg(MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
321 		return 1;
322 	}
323 
324 	/* Skip a leading escape sequence or tab. */
325 
326 	switch (buf[offs]) {
327 	case '\\':
328 		cp = buf + offs + 1;
329 		mandoc_escape(&cp, NULL, NULL);
330 		offs = cp - buf;
331 		break;
332 	case '\t':
333 		offs++;
334 		break;
335 	default:
336 		break;
337 	}
338 
339 	/* Jump to the next non-whitespace word. */
340 
341 	while (buf[offs] == ' ')
342 		offs++;
343 
344 	/*
345 	 * Trailing whitespace.  Note that tabs are allowed to be passed
346 	 * into the parser as "text", so we only warn about spaces here.
347 	 */
348 
349 	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
350 		mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);
351 
352 	/*
353 	 * If an initial macro or a list invocation, divert directly
354 	 * into macro processing.
355 	 */
356 
357 	n = mdoc->last;
358 	if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
359 		(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
360 		return 1;
361 	}
362 
363 	/*
364 	 * If a column list contains a non-It macro, assume an implicit
365 	 * item macro.  This can happen one or more times at the
366 	 * beginning of such a list, intermixed with text lines and
367 	 * with nodes generated on the roff level, for example by tbl.
368 	 */
369 
370 	if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
371 	     n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
372 	    (n->parent != NULL && n->parent->tok == MDOC_Bl &&
373 	     n->parent->norm->Bl.type == LIST_column)) {
374 		mdoc->flags |= MDOC_FREECOL;
375 		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
376 		return 1;
377 	}
378 
379 	/* Normal processing of a macro. */
380 
381 	(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
382 
383 	/* In quick mode (for mandocdb), abort after the NAME section. */
384 
385 	if (mdoc->quick && MDOC_Sh == tok &&
386 	    SEC_NAME != mdoc->last->sec)
387 		return 2;
388 
389 	return 1;
390 }
391 
392 enum mdelim
393 mdoc_isdelim(const char *p)
394 {
395 
396 	if ('\0' == p[0])
397 		return DELIM_NONE;
398 
399 	if ('\0' == p[1])
400 		switch (p[0]) {
401 		case '(':
402 		case '[':
403 			return DELIM_OPEN;
404 		case '|':
405 			return DELIM_MIDDLE;
406 		case '.':
407 		case ',':
408 		case ';':
409 		case ':':
410 		case '?':
411 		case '!':
412 		case ')':
413 		case ']':
414 			return DELIM_CLOSE;
415 		default:
416 			return DELIM_NONE;
417 		}
418 
419 	if ('\\' != p[0])
420 		return DELIM_NONE;
421 
422 	if (0 == strcmp(p + 1, "."))
423 		return DELIM_CLOSE;
424 	if (0 == strcmp(p + 1, "fR|\\fP"))
425 		return DELIM_MIDDLE;
426 
427 	return DELIM_NONE;
428 }
429