xref: /openbsd/usr.bin/mandoc/man_macro.c (revision cecf84d4) 
1 /*	$OpenBSD: man_macro.c,v 1.71 2015/04/23 15:35:39 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012, 2013, 2014, 2015 Ingo Schwarze <schwarze@openbsd.org>
5  * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
6  *
7  * Permission to use, copy, modify, and distribute this software for any
8  * purpose with or without fee is hereby granted, provided that the above
9  * copyright notice and this permission notice appear in all copies.
10  *
11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18  */
19 #include <sys/types.h>
20 
21 #include <assert.h>
22 #include <ctype.h>
23 #include <stdlib.h>
24 #include <string.h>
25 
26 #include "mandoc.h"
27 #include "roff.h"
28 #include "man.h"
29 #include "libmandoc.h"
30 #include "roff_int.h"
31 #include "libman.h"
32 
33 static	void		 blk_close(MACRO_PROT_ARGS);
34 static	void		 blk_exp(MACRO_PROT_ARGS);
35 static	void		 blk_imp(MACRO_PROT_ARGS);
36 static	void		 in_line_eoln(MACRO_PROT_ARGS);
37 static	int		 man_args(struct roff_man *, int,
38 				int *, char *, char **);
39 static	void		 rew_scope(struct roff_man *, int);
40 
41 const	struct man_macro __man_macros[MAN_MAX] = {
42 	{ in_line_eoln, MAN_NSCOPED }, /* br */
43 	{ in_line_eoln, MAN_BSCOPE }, /* TH */
44 	{ blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SH */
45 	{ blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SS */
46 	{ blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* TP */
47 	{ blk_imp, MAN_BSCOPE }, /* LP */
48 	{ blk_imp, MAN_BSCOPE }, /* PP */
49 	{ blk_imp, MAN_BSCOPE }, /* P */
50 	{ blk_imp, MAN_BSCOPE }, /* IP */
51 	{ blk_imp, MAN_BSCOPE }, /* HP */
52 	{ in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SM */
53 	{ in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SB */
54 	{ in_line_eoln, 0 }, /* BI */
55 	{ in_line_eoln, 0 }, /* IB */
56 	{ in_line_eoln, 0 }, /* BR */
57 	{ in_line_eoln, 0 }, /* RB */
58 	{ in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* R */
59 	{ in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* B */
60 	{ in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* I */
61 	{ in_line_eoln, 0 }, /* IR */
62 	{ in_line_eoln, 0 }, /* RI */
63 	{ in_line_eoln, MAN_NSCOPED }, /* sp */
64 	{ in_line_eoln, MAN_BSCOPE }, /* nf */
65 	{ in_line_eoln, MAN_BSCOPE }, /* fi */
66 	{ blk_close, MAN_BSCOPE }, /* RE */
67 	{ blk_exp, MAN_BSCOPE }, /* RS */
68 	{ in_line_eoln, 0 }, /* DT */
69 	{ in_line_eoln, 0 }, /* UC */
70 	{ in_line_eoln, MAN_NSCOPED }, /* PD */
71 	{ in_line_eoln, 0 }, /* AT */
72 	{ in_line_eoln, 0 }, /* in */
73 	{ in_line_eoln, 0 }, /* ft */
74 	{ in_line_eoln, 0 }, /* OP */
75 	{ in_line_eoln, MAN_BSCOPE }, /* EX */
76 	{ in_line_eoln, MAN_BSCOPE }, /* EE */
77 	{ blk_exp, MAN_BSCOPE }, /* UR */
78 	{ blk_close, MAN_BSCOPE }, /* UE */
79 	{ in_line_eoln, 0 }, /* ll */
80 };
81 
82 const	struct man_macro * const man_macros = __man_macros;
83 
84 
85 void
86 man_unscope(struct roff_man *man, const struct roff_node *to)
87 {
88 	struct roff_node *n;
89 
90 	to = to->parent;
91 	n = man->last;
92 	while (n != to) {
93 
94 		/* Reached the end of the document? */
95 
96 		if (to == NULL && ! (n->flags & MAN_VALID)) {
97 			if (man->flags & (MAN_BLINE | MAN_ELINE) &&
98 			    man_macros[n->tok].flags & MAN_SCOPED) {
99 				mandoc_vmsg(MANDOCERR_BLK_LINE,
100 				    man->parse, n->line, n->pos,
101 				    "EOF breaks %s",
102 				    man_macronames[n->tok]);
103 				if (man->flags & MAN_ELINE)
104 					man->flags &= ~MAN_ELINE;
105 				else {
106 					assert(n->type == ROFFT_HEAD);
107 					n = n->parent;
108 					man->flags &= ~MAN_BLINE;
109 				}
110 				man->last = n;
111 				n = n->parent;
112 				roff_node_delete(man, man->last);
113 				continue;
114 			}
115 			if (n->type == ROFFT_BLOCK &&
116 			    man_macros[n->tok].fp == blk_exp)
117 				mandoc_msg(MANDOCERR_BLK_NOEND,
118 				    man->parse, n->line, n->pos,
119 				    man_macronames[n->tok]);
120 		}
121 
122 		/*
123 		 * We might delete the man->last node
124 		 * in the post-validation phase.
125 		 * Save a pointer to the parent such that
126 		 * we know where to continue the iteration.
127 		 */
128 
129 		man->last = n;
130 		n = n->parent;
131 		man_valid_post(man);
132 	}
133 
134 	/*
135 	 * If we ended up at the parent of the node we were
136 	 * supposed to rewind to, that means the target node
137 	 * got deleted, so add the next node we parse as a child
138 	 * of the parent instead of as a sibling of the target.
139 	 */
140 
141 	man->next = (man->last == to) ?
142 	    ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;
143 }
144 
145 /*
146  * Rewinding entails ascending the parse tree until a coherent point,
147  * for example, the `SH' macro will close out any intervening `SS'
148  * scopes.  When a scope is closed, it must be validated and actioned.
149  */
150 static void
151 rew_scope(struct roff_man *man, int tok)
152 {
153 	struct roff_node *n;
154 
155 	/* Preserve empty paragraphs before RS. */
156 
157 	n = man->last;
158 	if (tok == MAN_RS && n->nchild == 0 &&
159 	    (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
160 		return;
161 
162 	for (;;) {
163 		if (n->type == ROFFT_ROOT)
164 			return;
165 		if (n->flags & MAN_VALID) {
166 			n = n->parent;
167 			continue;
168 		}
169 		if (n->type != ROFFT_BLOCK) {
170 			if (n->parent->type == ROFFT_ROOT) {
171 				man_unscope(man, n);
172 				return;
173 			} else {
174 				n = n->parent;
175 				continue;
176 			}
177 		}
178 		if (tok != MAN_SH && (n->tok == MAN_SH ||
179 		    (tok != MAN_SS && (n->tok == MAN_SS ||
180 		     man_macros[n->tok].fp == blk_exp))))
181 			return;
182 		man_unscope(man, n);
183 		n = man->last;
184 	}
185 }
186 
187 
188 /*
189  * Close out a generic explicit macro.
190  */
191 void
192 blk_close(MACRO_PROT_ARGS)
193 {
194 	int			 ntok;
195 	const struct roff_node	*nn;
196 	char			*p;
197 	int			 nrew, target;
198 
199 	nrew = 1;
200 	switch (tok) {
201 	case MAN_RE:
202 		ntok = MAN_RS;
203 		if ( ! man_args(man, line, pos, buf, &p))
204 			break;
205 		for (nn = man->last->parent; nn; nn = nn->parent)
206 			if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
207 				nrew++;
208 		target = strtol(p, &p, 10);
209 		if (*p != '\0')
210 			mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
211 			    line, p - buf, "RE ... %s", p);
212 		if (target == 0)
213 			target = 1;
214 		nrew -= target;
215 		if (nrew < 1) {
216 			mandoc_vmsg(MANDOCERR_RE_NOTOPEN, man->parse,
217 			    line, ppos, "RE %d", target);
218 			return;
219 		}
220 		break;
221 	case MAN_UE:
222 		ntok = MAN_UR;
223 		break;
224 	default:
225 		abort();
226 		/* NOTREACHED */
227 	}
228 
229 	for (nn = man->last->parent; nn; nn = nn->parent)
230 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
231 			break;
232 
233 	if (nn == NULL) {
234 		mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
235 		    line, ppos, man_macronames[tok]);
236 		rew_scope(man, MAN_PP);
237 	} else {
238 		line = man->last->line;
239 		ppos = man->last->pos;
240 		ntok = man->last->tok;
241 		man_unscope(man, nn);
242 
243 		/* Move a trailing paragraph behind the block. */
244 
245 		if (ntok == MAN_LP || ntok == MAN_PP || ntok == MAN_P) {
246 			*pos = strlen(buf);
247 			blk_imp(man, ntok, line, ppos, pos, buf);
248 		}
249 	}
250 }
251 
252 void
253 blk_exp(MACRO_PROT_ARGS)
254 {
255 	struct roff_node *head;
256 	char		*p;
257 	int		 la;
258 
259 	rew_scope(man, tok);
260 	roff_block_alloc(man, line, ppos, tok);
261 	head = roff_head_alloc(man, line, ppos, tok);
262 
263 	la = *pos;
264 	if (man_args(man, line, pos, buf, &p))
265 		roff_word_alloc(man, line, la, p);
266 
267 	if (buf[*pos] != '\0')
268 		mandoc_vmsg(MANDOCERR_ARG_EXCESS,
269 		    man->parse, line, *pos, "%s ... %s",
270 		    man_macronames[tok], buf + *pos);
271 
272 	man_unscope(man, head);
273 	roff_body_alloc(man, line, ppos, tok);
274 }
275 
276 /*
277  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
278  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
279  * scopes, such as `SH' closing out an `SS', are defined in the rew
280  * routines.
281  */
282 void
283 blk_imp(MACRO_PROT_ARGS)
284 {
285 	int		 la;
286 	char		*p;
287 	struct roff_node *n;
288 
289 	rew_scope(man, tok);
290 	n = roff_block_alloc(man, line, ppos, tok);
291 	if (n->tok == MAN_SH || n->tok == MAN_SS)
292 		man->flags &= ~MAN_LITERAL;
293 	n = roff_head_alloc(man, line, ppos, tok);
294 
295 	/* Add line arguments. */
296 
297 	for (;;) {
298 		la = *pos;
299 		if ( ! man_args(man, line, pos, buf, &p))
300 			break;
301 		roff_word_alloc(man, line, la, p);
302 	}
303 
304 	/*
305 	 * For macros having optional next-line scope,
306 	 * keep the head open if there were no arguments.
307 	 * For `TP', always keep the head open.
308 	 */
309 
310 	if (man_macros[tok].flags & MAN_SCOPED &&
311 	    (tok == MAN_TP || n == man->last)) {
312 		man->flags |= MAN_BLINE;
313 		return;
314 	}
315 
316 	/* Close out the head and open the body. */
317 
318 	man_unscope(man, n);
319 	roff_body_alloc(man, line, ppos, tok);
320 }
321 
322 void
323 in_line_eoln(MACRO_PROT_ARGS)
324 {
325 	int		 la;
326 	char		*p;
327 	struct roff_node *n;
328 
329 	roff_elem_alloc(man, line, ppos, tok);
330 	n = man->last;
331 
332 	for (;;) {
333 		if (buf[*pos] != '\0' && (tok == MAN_br ||
334 		    tok == MAN_fi || tok == MAN_nf)) {
335 			mandoc_vmsg(MANDOCERR_ARG_SKIP,
336 			    man->parse, line, *pos, "%s %s",
337 			    man_macronames[tok], buf + *pos);
338 			break;
339 		}
340 		if (buf[*pos] != '\0' && man->last != n &&
341 		    (tok == MAN_PD || tok == MAN_ft || tok == MAN_sp)) {
342 			mandoc_vmsg(MANDOCERR_ARG_EXCESS,
343 			    man->parse, line, *pos, "%s ... %s",
344 			    man_macronames[tok], buf + *pos);
345 			break;
346 		}
347 		la = *pos;
348 		if ( ! man_args(man, line, pos, buf, &p))
349 			break;
350 		if (man_macros[tok].flags & MAN_JOIN &&
351 		    man->last->type == ROFFT_TEXT)
352 			roff_word_append(man, p);
353 		else
354 			roff_word_alloc(man, line, la, p);
355 	}
356 
357 	/*
358 	 * Append MAN_EOS in case the last snipped argument
359 	 * ends with a dot, e.g. `.IR syslog (3).'
360 	 */
361 
362 	if (n != man->last &&
363 	    mandoc_eos(man->last->string, strlen(man->last->string)))
364 		man->last->flags |= MAN_EOS;
365 
366 	/*
367 	 * If no arguments are specified and this is MAN_SCOPED (i.e.,
368 	 * next-line scoped), then set our mode to indicate that we're
369 	 * waiting for terms to load into our context.
370 	 */
371 
372 	if (n == man->last && man_macros[tok].flags & MAN_SCOPED) {
373 		assert( ! (man_macros[tok].flags & MAN_NSCOPED));
374 		man->flags |= MAN_ELINE;
375 		return;
376 	}
377 
378 	assert(man->last->type != ROFFT_ROOT);
379 	man->next = ROFF_NEXT_SIBLING;
380 
381 	/*
382 	 * Rewind our element scope.  Note that when TH is pruned, we'll
383 	 * be back at the root, so make sure that we don't clobber as
384 	 * its sibling.
385 	 */
386 
387 	for ( ; man->last; man->last = man->last->parent) {
388 		if (man->last == n)
389 			break;
390 		if (man->last->type == ROFFT_ROOT)
391 			break;
392 		man_valid_post(man);
393 	}
394 
395 	assert(man->last);
396 
397 	/*
398 	 * Same here regarding whether we're back at the root.
399 	 */
400 
401 	if (man->last->type != ROFFT_ROOT)
402 		man_valid_post(man);
403 }
404 
405 void
406 man_endparse(struct roff_man *man)
407 {
408 
409 	man_unscope(man, man->first);
410 }
411 
412 static int
413 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
414 {
415 	char	 *start;
416 
417 	assert(*pos);
418 	*v = start = buf + *pos;
419 	assert(' ' != *start);
420 
421 	if ('\0' == *start)
422 		return(0);
423 
424 	*v = mandoc_getarg(man->parse, v, line, pos);
425 	return(1);
426 }
427