xref: /openbsd-src/usr.bin/mandoc/man_macro.c (revision d1df930ffab53da22f3324c32bed7ac5709915e6)
1 /*	$OpenBSD: man_macro.c,v 1.95 2018/08/26 16:18:38 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012-2015, 2017, 2018 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 *, enum roff_tok);
40 
41 static const struct man_macro man_macros[MAN_MAX - MAN_TH] = {
42 	{ in_line_eoln, MAN_XSCOPE }, /* TH */
43 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SH */
44 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SS */
45 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TP */
46 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TQ */
47 	{ blk_imp, MAN_XSCOPE }, /* LP */
48 	{ blk_imp, MAN_XSCOPE }, /* PP */
49 	{ blk_imp, MAN_XSCOPE }, /* P */
50 	{ blk_imp, MAN_XSCOPE }, /* IP */
51 	{ blk_imp, MAN_XSCOPE }, /* HP */
52 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SM */
53 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | 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_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* R */
59 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* B */
60 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* I */
61 	{ in_line_eoln, 0 }, /* IR */
62 	{ in_line_eoln, 0 }, /* RI */
63 	{ in_line_eoln, MAN_NSCOPED }, /* nf */
64 	{ in_line_eoln, MAN_NSCOPED }, /* fi */
65 	{ blk_close, MAN_XSCOPE }, /* RE */
66 	{ blk_exp, MAN_XSCOPE }, /* RS */
67 	{ in_line_eoln, 0 }, /* DT */
68 	{ in_line_eoln, 0 }, /* UC */
69 	{ in_line_eoln, MAN_NSCOPED }, /* PD */
70 	{ in_line_eoln, 0 }, /* AT */
71 	{ in_line_eoln, MAN_NSCOPED }, /* in */
72 	{ blk_imp, MAN_XSCOPE }, /* SY */
73 	{ blk_close, MAN_XSCOPE }, /* YS */
74 	{ in_line_eoln, 0 }, /* OP */
75 	{ in_line_eoln, MAN_XSCOPE }, /* EX */
76 	{ in_line_eoln, MAN_XSCOPE }, /* EE */
77 	{ blk_exp, MAN_XSCOPE }, /* UR */
78 	{ blk_close, MAN_XSCOPE }, /* UE */
79 	{ blk_exp, MAN_XSCOPE }, /* MT */
80 	{ blk_close, MAN_XSCOPE }, /* ME */
81 };
82 
83 
84 const struct man_macro *
85 man_macro(enum roff_tok tok)
86 {
87 	assert(tok >= MAN_TH && tok <= MAN_MAX);
88 	return man_macros + (tok - MAN_TH);
89 }
90 
91 void
92 man_unscope(struct roff_man *man, const struct roff_node *to)
93 {
94 	struct roff_node *n;
95 
96 	to = to->parent;
97 	n = man->last;
98 	while (n != to) {
99 
100 		/* Reached the end of the document? */
101 
102 		if (to == NULL && ! (n->flags & NODE_VALID)) {
103 			if (man->flags & (MAN_BLINE | MAN_ELINE) &&
104 			    man_macro(n->tok)->flags &
105 			     (MAN_BSCOPED | MAN_NSCOPED)) {
106 				mandoc_vmsg(MANDOCERR_BLK_LINE,
107 				    man->parse, n->line, n->pos,
108 				    "EOF breaks %s", roff_name[n->tok]);
109 				if (man->flags & MAN_ELINE)
110 					man->flags &= ~MAN_ELINE;
111 				else {
112 					assert(n->type == ROFFT_HEAD);
113 					n = n->parent;
114 					man->flags &= ~MAN_BLINE;
115 				}
116 				man->last = n;
117 				n = n->parent;
118 				roff_node_delete(man, man->last);
119 				continue;
120 			}
121 			if (n->type == ROFFT_BLOCK &&
122 			    man_macro(n->tok)->fp == blk_exp)
123 				mandoc_msg(MANDOCERR_BLK_NOEND,
124 				    man->parse, n->line, n->pos,
125 				    roff_name[n->tok]);
126 		}
127 
128 		/*
129 		 * We might delete the man->last node
130 		 * in the post-validation phase.
131 		 * Save a pointer to the parent such that
132 		 * we know where to continue the iteration.
133 		 */
134 
135 		man->last = n;
136 		n = n->parent;
137 		man->last->flags |= NODE_VALID;
138 	}
139 
140 	/*
141 	 * If we ended up at the parent of the node we were
142 	 * supposed to rewind to, that means the target node
143 	 * got deleted, so add the next node we parse as a child
144 	 * of the parent instead of as a sibling of the target.
145 	 */
146 
147 	man->next = (man->last == to) ?
148 	    ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;
149 }
150 
151 /*
152  * Rewinding entails ascending the parse tree until a coherent point,
153  * for example, the `SH' macro will close out any intervening `SS'
154  * scopes.  When a scope is closed, it must be validated and actioned.
155  */
156 static void
157 rew_scope(struct roff_man *man, enum roff_tok tok)
158 {
159 	struct roff_node *n;
160 
161 	/* Preserve empty paragraphs before RS. */
162 
163 	n = man->last;
164 	if (tok == MAN_RS && n->child == NULL &&
165 	    (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
166 		return;
167 
168 	for (;;) {
169 		if (n->type == ROFFT_ROOT)
170 			return;
171 		if (n->flags & NODE_VALID) {
172 			n = n->parent;
173 			continue;
174 		}
175 		if (n->type != ROFFT_BLOCK) {
176 			if (n->parent->type == ROFFT_ROOT) {
177 				man_unscope(man, n);
178 				return;
179 			} else {
180 				n = n->parent;
181 				continue;
182 			}
183 		}
184 		if (tok != MAN_SH && (n->tok == MAN_SH ||
185 		    (tok != MAN_SS && (n->tok == MAN_SS ||
186 		     man_macro(n->tok)->fp == blk_exp))))
187 			return;
188 		man_unscope(man, n);
189 		n = man->last;
190 	}
191 }
192 
193 
194 /*
195  * Close out a generic explicit macro.
196  */
197 void
198 blk_close(MACRO_PROT_ARGS)
199 {
200 	enum roff_tok		 ctok, ntok;
201 	const struct roff_node	*nn;
202 	char			*p;
203 	int			 cline, cpos, nrew, target;
204 
205 	nrew = 1;
206 	switch (tok) {
207 	case MAN_RE:
208 		ntok = MAN_RS;
209 		if ( ! man_args(man, line, pos, buf, &p))
210 			break;
211 		for (nn = man->last->parent; nn; nn = nn->parent)
212 			if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
213 				nrew++;
214 		target = strtol(p, &p, 10);
215 		if (*p != '\0')
216 			mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
217 			    line, p - buf, "RE ... %s", p);
218 		if (target == 0)
219 			target = 1;
220 		nrew -= target;
221 		if (nrew < 1) {
222 			mandoc_vmsg(MANDOCERR_RE_NOTOPEN, man->parse,
223 			    line, ppos, "RE %d", target);
224 			return;
225 		}
226 		break;
227 	case MAN_YS:
228 		ntok = MAN_SY;
229 		break;
230 	case MAN_UE:
231 		ntok = MAN_UR;
232 		break;
233 	case MAN_ME:
234 		ntok = MAN_MT;
235 		break;
236 	default:
237 		abort();
238 	}
239 
240 	for (nn = man->last->parent; nn; nn = nn->parent)
241 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
242 			break;
243 
244 	if (nn == NULL) {
245 		mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
246 		    line, ppos, roff_name[tok]);
247 		rew_scope(man, MAN_PP);
248 		if (tok == MAN_RE) {
249 			roff_elem_alloc(man, line, ppos, ROFF_br);
250 			man->last->flags |= NODE_LINE |
251 			    NODE_VALID | NODE_ENDED;
252 			man->next = ROFF_NEXT_SIBLING;
253 		}
254 		return;
255 	}
256 
257 	cline = man->last->line;
258 	cpos = man->last->pos;
259 	ctok = man->last->tok;
260 	man_unscope(man, nn);
261 
262 	if (tok == MAN_RE && nn->head->aux > 0)
263 		roff_setreg(man->roff, "an-margin", nn->head->aux, '-');
264 
265 	/* Trailing text. */
266 
267 	if (buf[*pos] != '\0') {
268 		roff_word_alloc(man, line, ppos, buf + *pos);
269 		man->last->flags |= NODE_DELIMC;
270 		if (mandoc_eos(man->last->string, strlen(man->last->string)))
271 			man->last->flags |= NODE_EOS;
272 	}
273 
274 	/* Move a trailing paragraph behind the block. */
275 
276 	if (ctok == MAN_LP || ctok == MAN_PP || ctok == MAN_P) {
277 		*pos = strlen(buf);
278 		blk_imp(man, ctok, cline, cpos, pos, buf);
279 	}
280 
281 	/* Synopsis blocks need an explicit end marker for spacing. */
282 
283 	if (tok == MAN_YS && man->last == nn) {
284 		roff_elem_alloc(man, line, ppos, tok);
285 		man_unscope(man, man->last);
286 	}
287 }
288 
289 void
290 blk_exp(MACRO_PROT_ARGS)
291 {
292 	struct roff_node *head;
293 	char		*p;
294 	int		 la;
295 
296 	if (tok == MAN_RS)
297 		rew_scope(man, tok);
298 	roff_block_alloc(man, line, ppos, tok);
299 	head = roff_head_alloc(man, line, ppos, tok);
300 
301 	la = *pos;
302 	if (man_args(man, line, pos, buf, &p)) {
303 		roff_word_alloc(man, line, la, p);
304 		if (tok == MAN_RS) {
305 			if (roff_getreg(man->roff, "an-margin") == 0)
306 				roff_setreg(man->roff, "an-margin",
307 				    7 * 24, '=');
308 			if ((head->aux = strtod(p, NULL) * 24.0) > 0)
309 				roff_setreg(man->roff, "an-margin",
310 				    head->aux, '+');
311 		}
312 	}
313 
314 	if (buf[*pos] != '\0')
315 		mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse, line,
316 		    *pos, "%s ... %s", roff_name[tok], buf + *pos);
317 
318 	man_unscope(man, head);
319 	roff_body_alloc(man, line, ppos, tok);
320 }
321 
322 /*
323  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
324  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
325  * scopes, such as `SH' closing out an `SS', are defined in the rew
326  * routines.
327  */
328 void
329 blk_imp(MACRO_PROT_ARGS)
330 {
331 	int		 la;
332 	char		*p;
333 	struct roff_node *n;
334 
335 	rew_scope(man, tok);
336 	n = roff_block_alloc(man, line, ppos, tok);
337 	if (n->tok == MAN_SH || n->tok == MAN_SS)
338 		man->flags &= ~MAN_LITERAL;
339 	n = roff_head_alloc(man, line, ppos, tok);
340 
341 	/* Add line arguments. */
342 
343 	for (;;) {
344 		la = *pos;
345 		if ( ! man_args(man, line, pos, buf, &p))
346 			break;
347 		roff_word_alloc(man, line, la, p);
348 	}
349 
350 	/*
351 	 * For macros having optional next-line scope,
352 	 * keep the head open if there were no arguments.
353 	 * For `TP' and `TQ', always keep the head open.
354 	 */
355 
356 	if (man_macro(tok)->flags & MAN_BSCOPED &&
357 	    (tok == MAN_TP || tok == MAN_TQ || n == man->last)) {
358 		man->flags |= MAN_BLINE;
359 		return;
360 	}
361 
362 	/* Close out the head and open the body. */
363 
364 	man_unscope(man, n);
365 	roff_body_alloc(man, line, ppos, tok);
366 }
367 
368 void
369 in_line_eoln(MACRO_PROT_ARGS)
370 {
371 	int		 la;
372 	char		*p;
373 	struct roff_node *n;
374 
375 	roff_elem_alloc(man, line, ppos, tok);
376 	n = man->last;
377 
378 	for (;;) {
379 		if (buf[*pos] != '\0' && (tok == MAN_fi || tok == MAN_nf)) {
380 			mandoc_vmsg(MANDOCERR_ARG_SKIP,
381 			    man->parse, line, *pos, "%s %s",
382 			    roff_name[tok], buf + *pos);
383 			break;
384 		}
385 		if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {
386 			mandoc_vmsg(MANDOCERR_ARG_EXCESS,
387 			    man->parse, line, *pos, "%s ... %s",
388 			    roff_name[tok], buf + *pos);
389 			break;
390 		}
391 		la = *pos;
392 		if ( ! man_args(man, line, pos, buf, &p))
393 			break;
394 		if (man_macro(tok)->flags & MAN_JOIN &&
395 		    man->last->type == ROFFT_TEXT)
396 			roff_word_append(man, p);
397 		else
398 			roff_word_alloc(man, line, la, p);
399 	}
400 
401 	/*
402 	 * Append NODE_EOS in case the last snipped argument
403 	 * ends with a dot, e.g. `.IR syslog (3).'
404 	 */
405 
406 	if (n != man->last &&
407 	    mandoc_eos(man->last->string, strlen(man->last->string)))
408 		man->last->flags |= NODE_EOS;
409 
410 	/*
411 	 * If no arguments are specified and this is MAN_ESCOPED (i.e.,
412 	 * next-line scoped), then set our mode to indicate that we're
413 	 * waiting for terms to load into our context.
414 	 */
415 
416 	if (n == man->last && man_macro(tok)->flags & MAN_ESCOPED) {
417 		man->flags |= MAN_ELINE;
418 		return;
419 	}
420 
421 	assert(man->last->type != ROFFT_ROOT);
422 	man->next = ROFF_NEXT_SIBLING;
423 
424 	/* Rewind our element scope. */
425 
426 	for ( ; man->last; man->last = man->last->parent) {
427 		man_state(man, man->last);
428 		if (man->last == n)
429 			break;
430 	}
431 
432 	/* Rewind next-line scoped ancestors, if any. */
433 
434 	if (man_macro(tok)->flags & MAN_ESCOPED)
435 		man_descope(man, line, ppos, NULL);
436 }
437 
438 void
439 man_endparse(struct roff_man *man)
440 {
441 
442 	man_unscope(man, man->first);
443 	man->flags &= ~MAN_LITERAL;
444 }
445 
446 static int
447 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
448 {
449 	char	 *start;
450 
451 	assert(*pos);
452 	*v = start = buf + *pos;
453 	assert(' ' != *start);
454 
455 	if ('\0' == *start)
456 		return 0;
457 
458 	*v = mandoc_getarg(man->parse, v, line, pos);
459 	return 1;
460 }
461