xref: /openbsd-src/usr.bin/mandoc/man_macro.c (revision 0b7734b3d77bb9b21afec6f4621cae6c805dbd45)
1 /*	$OpenBSD: man_macro.c,v 1.76 2016/01/08 17:48:04 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_NSCOPED }, /* nf */
65 	{ in_line_eoln, MAN_NSCOPED }, /* 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->last->flags |= MAN_VALID;
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->child == NULL &&
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 	}
227 
228 	for (nn = man->last->parent; nn; nn = nn->parent)
229 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
230 			break;
231 
232 	if (nn == NULL) {
233 		mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
234 		    line, ppos, man_macronames[tok]);
235 		rew_scope(man, MAN_PP);
236 	} else {
237 		line = man->last->line;
238 		ppos = man->last->pos;
239 		ntok = man->last->tok;
240 		man_unscope(man, nn);
241 
242 		/* Move a trailing paragraph behind the block. */
243 
244 		if (ntok == MAN_LP || ntok == MAN_PP || ntok == MAN_P) {
245 			*pos = strlen(buf);
246 			blk_imp(man, ntok, line, ppos, pos, buf);
247 		}
248 	}
249 }
250 
251 void
252 blk_exp(MACRO_PROT_ARGS)
253 {
254 	struct roff_node *head;
255 	char		*p;
256 	int		 la;
257 
258 	rew_scope(man, tok);
259 	roff_block_alloc(man, line, ppos, tok);
260 	head = roff_head_alloc(man, line, ppos, tok);
261 
262 	la = *pos;
263 	if (man_args(man, line, pos, buf, &p))
264 		roff_word_alloc(man, line, la, p);
265 
266 	if (buf[*pos] != '\0')
267 		mandoc_vmsg(MANDOCERR_ARG_EXCESS,
268 		    man->parse, line, *pos, "%s ... %s",
269 		    man_macronames[tok], buf + *pos);
270 
271 	man_unscope(man, head);
272 	roff_body_alloc(man, line, ppos, tok);
273 }
274 
275 /*
276  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
277  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
278  * scopes, such as `SH' closing out an `SS', are defined in the rew
279  * routines.
280  */
281 void
282 blk_imp(MACRO_PROT_ARGS)
283 {
284 	int		 la;
285 	char		*p;
286 	struct roff_node *n;
287 
288 	rew_scope(man, tok);
289 	n = roff_block_alloc(man, line, ppos, tok);
290 	if (n->tok == MAN_SH || n->tok == MAN_SS)
291 		man->flags &= ~MAN_LITERAL;
292 	n = roff_head_alloc(man, line, ppos, tok);
293 
294 	/* Add line arguments. */
295 
296 	for (;;) {
297 		la = *pos;
298 		if ( ! man_args(man, line, pos, buf, &p))
299 			break;
300 		roff_word_alloc(man, line, la, p);
301 	}
302 
303 	/*
304 	 * For macros having optional next-line scope,
305 	 * keep the head open if there were no arguments.
306 	 * For `TP', always keep the head open.
307 	 */
308 
309 	if (man_macros[tok].flags & MAN_SCOPED &&
310 	    (tok == MAN_TP || n == man->last)) {
311 		man->flags |= MAN_BLINE;
312 		return;
313 	}
314 
315 	/* Close out the head and open the body. */
316 
317 	man_unscope(man, n);
318 	roff_body_alloc(man, line, ppos, tok);
319 }
320 
321 void
322 in_line_eoln(MACRO_PROT_ARGS)
323 {
324 	int		 la;
325 	char		*p;
326 	struct roff_node *n;
327 
328 	roff_elem_alloc(man, line, ppos, tok);
329 	n = man->last;
330 
331 	for (;;) {
332 		if (buf[*pos] != '\0' && (tok == MAN_br ||
333 		    tok == MAN_fi || tok == MAN_nf)) {
334 			mandoc_vmsg(MANDOCERR_ARG_SKIP,
335 			    man->parse, line, *pos, "%s %s",
336 			    man_macronames[tok], buf + *pos);
337 			break;
338 		}
339 		if (buf[*pos] != '\0' && man->last != n &&
340 		    (tok == MAN_PD || tok == MAN_ft || tok == MAN_sp)) {
341 			mandoc_vmsg(MANDOCERR_ARG_EXCESS,
342 			    man->parse, line, *pos, "%s ... %s",
343 			    man_macronames[tok], buf + *pos);
344 			break;
345 		}
346 		la = *pos;
347 		if ( ! man_args(man, line, pos, buf, &p))
348 			break;
349 		if (man_macros[tok].flags & MAN_JOIN &&
350 		    man->last->type == ROFFT_TEXT)
351 			roff_word_append(man, p);
352 		else
353 			roff_word_alloc(man, line, la, p);
354 	}
355 
356 	/*
357 	 * Append MAN_EOS in case the last snipped argument
358 	 * ends with a dot, e.g. `.IR syslog (3).'
359 	 */
360 
361 	if (n != man->last &&
362 	    mandoc_eos(man->last->string, strlen(man->last->string)))
363 		man->last->flags |= MAN_EOS;
364 
365 	/*
366 	 * If no arguments are specified and this is MAN_SCOPED (i.e.,
367 	 * next-line scoped), then set our mode to indicate that we're
368 	 * waiting for terms to load into our context.
369 	 */
370 
371 	if (n == man->last && man_macros[tok].flags & MAN_SCOPED) {
372 		assert( ! (man_macros[tok].flags & MAN_NSCOPED));
373 		man->flags |= MAN_ELINE;
374 		return;
375 	}
376 
377 	assert(man->last->type != ROFFT_ROOT);
378 	man->next = ROFF_NEXT_SIBLING;
379 
380 	/* Rewind our element scope. */
381 
382 	for ( ; man->last; man->last = man->last->parent) {
383 		man_state(man, man->last);
384 		if (man->last == n)
385 			break;
386 	}
387 }
388 
389 void
390 man_endparse(struct roff_man *man)
391 {
392 
393 	man_unscope(man, man->first);
394 	man->flags &= ~MAN_LITERAL;
395 }
396 
397 static int
398 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
399 {
400 	char	 *start;
401 
402 	assert(*pos);
403 	*v = start = buf + *pos;
404 	assert(' ' != *start);
405 
406 	if ('\0' == *start)
407 		return 0;
408 
409 	*v = mandoc_getarg(man->parse, v, line, pos);
410 	return 1;
411 }
412