xref: /openbsd/usr.bin/mandoc/man_macro.c (revision 3cab2bb3) 
1 /*	$OpenBSD: man_macro.c,v 1.106 2019/01/05 18:59:37 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012-2015, 2017-2019 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 <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 
27 #include "mandoc.h"
28 #include "roff.h"
29 #include "man.h"
30 #include "libmandoc.h"
31 #include "roff_int.h"
32 #include "libman.h"
33 
34 static	void		 blk_close(MACRO_PROT_ARGS);
35 static	void		 blk_exp(MACRO_PROT_ARGS);
36 static	void		 blk_imp(MACRO_PROT_ARGS);
37 static	void		 in_line_eoln(MACRO_PROT_ARGS);
38 static	int		 man_args(struct roff_man *, int,
39 				int *, char *, char **);
40 static	void		 rew_scope(struct roff_man *, enum roff_tok);
41 
42 static const struct man_macro man_macros[MAN_MAX - MAN_TH] = {
43 	{ in_line_eoln, MAN_XSCOPE }, /* TH */
44 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SH */
45 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SS */
46 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TP */
47 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TQ */
48 	{ blk_imp, MAN_XSCOPE }, /* LP */
49 	{ blk_imp, MAN_XSCOPE }, /* PP */
50 	{ blk_imp, MAN_XSCOPE }, /* P */
51 	{ blk_imp, MAN_XSCOPE }, /* IP */
52 	{ blk_imp, MAN_XSCOPE }, /* HP */
53 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SM */
54 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SB */
55 	{ in_line_eoln, 0 }, /* BI */
56 	{ in_line_eoln, 0 }, /* IB */
57 	{ in_line_eoln, 0 }, /* BR */
58 	{ in_line_eoln, 0 }, /* RB */
59 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* R */
60 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* B */
61 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* I */
62 	{ in_line_eoln, 0 }, /* IR */
63 	{ in_line_eoln, 0 }, /* RI */
64 	{ blk_close, MAN_XSCOPE }, /* RE */
65 	{ blk_exp, MAN_XSCOPE }, /* RS */
66 	{ in_line_eoln, 0 }, /* DT */
67 	{ in_line_eoln, 0 }, /* UC */
68 	{ in_line_eoln, MAN_NSCOPED }, /* PD */
69 	{ in_line_eoln, 0 }, /* AT */
70 	{ in_line_eoln, MAN_NSCOPED }, /* in */
71 	{ blk_imp, MAN_XSCOPE }, /* SY */
72 	{ blk_close, MAN_XSCOPE }, /* YS */
73 	{ in_line_eoln, 0 }, /* OP */
74 	{ in_line_eoln, MAN_XSCOPE }, /* EX */
75 	{ in_line_eoln, MAN_XSCOPE }, /* EE */
76 	{ blk_exp, MAN_XSCOPE }, /* UR */
77 	{ blk_close, MAN_XSCOPE }, /* UE */
78 	{ blk_exp, MAN_XSCOPE }, /* MT */
79 	{ blk_close, MAN_XSCOPE }, /* ME */
80 };
81 
82 
83 const struct man_macro *
84 man_macro(enum roff_tok tok)
85 {
86 	assert(tok >= MAN_TH && tok <= MAN_MAX);
87 	return man_macros + (tok - MAN_TH);
88 }
89 
90 void
91 man_unscope(struct roff_man *man, const struct roff_node *to)
92 {
93 	struct roff_node *n;
94 
95 	to = to->parent;
96 	n = man->last;
97 	while (n != to) {
98 
99 		/* Reached the end of the document? */
100 
101 		if (to == NULL && ! (n->flags & NODE_VALID)) {
102 			if (man->flags & (MAN_BLINE | MAN_ELINE) &&
103 			    man_macro(n->tok)->flags &
104 			     (MAN_BSCOPED | MAN_NSCOPED)) {
105 				mandoc_msg(MANDOCERR_BLK_LINE,
106 				    n->line, n->pos,
107 				    "EOF breaks %s", roff_name[n->tok]);
108 				if (man->flags & MAN_ELINE)
109 					man->flags &= ~MAN_ELINE;
110 				else {
111 					assert(n->type == ROFFT_HEAD);
112 					n = n->parent;
113 					man->flags &= ~MAN_BLINE;
114 				}
115 				man->last = n;
116 				n = n->parent;
117 				roff_node_delete(man, man->last);
118 				continue;
119 			}
120 			if (n->type == ROFFT_BLOCK &&
121 			    man_macro(n->tok)->fp == blk_exp)
122 				mandoc_msg(MANDOCERR_BLK_NOEND,
123 				    n->line, n->pos, "%s",
124 				    roff_name[n->tok]);
125 		}
126 
127 		/*
128 		 * We might delete the man->last node
129 		 * in the post-validation phase.
130 		 * Save a pointer to the parent such that
131 		 * we know where to continue the iteration.
132 		 */
133 
134 		man->last = n;
135 		n = n->parent;
136 		man->last->flags |= NODE_VALID;
137 	}
138 
139 	/*
140 	 * If we ended up at the parent of the node we were
141 	 * supposed to rewind to, that means the target node
142 	 * got deleted, so add the next node we parse as a child
143 	 * of the parent instead of as a sibling of the target.
144 	 */
145 
146 	man->next = (man->last == to) ?
147 	    ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;
148 }
149 
150 /*
151  * Rewinding entails ascending the parse tree until a coherent point,
152  * for example, the `SH' macro will close out any intervening `SS'
153  * scopes.  When a scope is closed, it must be validated and actioned.
154  */
155 static void
156 rew_scope(struct roff_man *man, enum roff_tok tok)
157 {
158 	struct roff_node *n;
159 
160 	/* Preserve empty paragraphs before RS. */
161 
162 	n = man->last;
163 	if (tok == MAN_RS && n->child == NULL &&
164 	    (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
165 		return;
166 
167 	for (;;) {
168 		if (n->type == ROFFT_ROOT)
169 			return;
170 		if (n->flags & NODE_VALID) {
171 			n = n->parent;
172 			continue;
173 		}
174 		if (n->type != ROFFT_BLOCK) {
175 			if (n->parent->type == ROFFT_ROOT) {
176 				man_unscope(man, n);
177 				return;
178 			} else {
179 				n = n->parent;
180 				continue;
181 			}
182 		}
183 		if (tok != MAN_SH && (n->tok == MAN_SH ||
184 		    (tok != MAN_SS && (n->tok == MAN_SS ||
185 		     man_macro(n->tok)->fp == blk_exp))))
186 			return;
187 		man_unscope(man, n);
188 		n = man->last;
189 	}
190 }
191 
192 
193 /*
194  * Close out a generic explicit macro.
195  */
196 void
197 blk_close(MACRO_PROT_ARGS)
198 {
199 	enum roff_tok		 ctok, ntok;
200 	const struct roff_node	*nn;
201 	char			*p, *ep;
202 	int			 cline, cpos, la, nrew, target;
203 
204 	nrew = 1;
205 	switch (tok) {
206 	case MAN_RE:
207 		ntok = MAN_RS;
208 		la = *pos;
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, &ep, 10);
215 		if (*ep != '\0')
216 			mandoc_msg(MANDOCERR_ARG_EXCESS, line,
217 			    la + (buf[la] == '"') + (int)(ep - p),
218 			    "RE ... %s", ep);
219 		free(p);
220 		if (target == 0)
221 			target = 1;
222 		nrew -= target;
223 		if (nrew < 1) {
224 			mandoc_msg(MANDOCERR_RE_NOTOPEN,
225 			    line, ppos, "RE %d", target);
226 			return;
227 		}
228 		break;
229 	case MAN_YS:
230 		ntok = MAN_SY;
231 		break;
232 	case MAN_UE:
233 		ntok = MAN_UR;
234 		break;
235 	case MAN_ME:
236 		ntok = MAN_MT;
237 		break;
238 	default:
239 		abort();
240 	}
241 
242 	for (nn = man->last->parent; nn; nn = nn->parent)
243 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
244 			break;
245 
246 	if (nn == NULL) {
247 		mandoc_msg(MANDOCERR_BLK_NOTOPEN,
248 		    line, ppos, "%s", roff_name[tok]);
249 		rew_scope(man, MAN_PP);
250 		if (tok == MAN_RE) {
251 			roff_elem_alloc(man, line, ppos, ROFF_br);
252 			man->last->flags |= NODE_LINE |
253 			    NODE_VALID | NODE_ENDED;
254 			man->next = ROFF_NEXT_SIBLING;
255 		}
256 		return;
257 	}
258 
259 	cline = man->last->line;
260 	cpos = man->last->pos;
261 	ctok = man->last->tok;
262 	man_unscope(man, nn);
263 
264 	if (tok == MAN_RE && nn->head->aux > 0)
265 		roff_setreg(man->roff, "an-margin", nn->head->aux, '-');
266 
267 	/* Trailing text. */
268 
269 	if (buf[*pos] != '\0') {
270 		roff_word_alloc(man, line, ppos, buf + *pos);
271 		man->last->flags |= NODE_DELIMC;
272 		if (mandoc_eos(man->last->string, strlen(man->last->string)))
273 			man->last->flags |= NODE_EOS;
274 	}
275 
276 	/* Move a trailing paragraph behind the block. */
277 
278 	if (ctok == MAN_LP || ctok == MAN_PP || ctok == MAN_P) {
279 		*pos = strlen(buf);
280 		blk_imp(man, ctok, cline, cpos, pos, buf);
281 	}
282 
283 	/* Synopsis blocks need an explicit end marker for spacing. */
284 
285 	if (tok == MAN_YS && man->last == nn) {
286 		roff_elem_alloc(man, line, ppos, tok);
287 		man_unscope(man, man->last);
288 	}
289 }
290 
291 void
292 blk_exp(MACRO_PROT_ARGS)
293 {
294 	struct roff_node *head;
295 	char		*p;
296 	int		 la;
297 
298 	if (tok == MAN_RS) {
299 		rew_scope(man, tok);
300 		man->flags |= ROFF_NONOFILL;
301 	}
302 	roff_block_alloc(man, line, ppos, tok);
303 	head = roff_head_alloc(man, line, ppos, tok);
304 
305 	la = *pos;
306 	if (man_args(man, line, pos, buf, &p)) {
307 		roff_word_alloc(man, line, la, p);
308 		if (tok == MAN_RS) {
309 			if (roff_getreg(man->roff, "an-margin") == 0)
310 				roff_setreg(man->roff, "an-margin",
311 				    7 * 24, '=');
312 			if ((head->aux = strtod(p, NULL) * 24.0) > 0)
313 				roff_setreg(man->roff, "an-margin",
314 				    head->aux, '+');
315 		}
316 		free(p);
317 	}
318 
319 	if (buf[*pos] != '\0')
320 		mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
321 		    "%s ... %s", roff_name[tok], buf + *pos);
322 
323 	man_unscope(man, head);
324 	roff_body_alloc(man, line, ppos, tok);
325 	man->flags &= ~ROFF_NONOFILL;
326 }
327 
328 /*
329  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
330  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
331  * scopes, such as `SH' closing out an `SS', are defined in the rew
332  * routines.
333  */
334 void
335 blk_imp(MACRO_PROT_ARGS)
336 {
337 	int		 la;
338 	char		*p;
339 	struct roff_node *n;
340 
341 	rew_scope(man, tok);
342 	man->flags |= ROFF_NONOFILL;
343 	if (tok == MAN_SH || tok == MAN_SS)
344 		man->flags &= ~ROFF_NOFILL;
345 	roff_block_alloc(man, line, ppos, tok);
346 	n = roff_head_alloc(man, line, ppos, tok);
347 
348 	/* Add line arguments. */
349 
350 	for (;;) {
351 		la = *pos;
352 		if ( ! man_args(man, line, pos, buf, &p))
353 			break;
354 		roff_word_alloc(man, line, la, p);
355 		free(p);
356 	}
357 
358 	/*
359 	 * For macros having optional next-line scope,
360 	 * keep the head open if there were no arguments.
361 	 * For `TP' and `TQ', always keep the head open.
362 	 */
363 
364 	if (man_macro(tok)->flags & MAN_BSCOPED &&
365 	    (tok == MAN_TP || tok == MAN_TQ || n == man->last)) {
366 		man->flags |= MAN_BLINE;
367 		return;
368 	}
369 
370 	/* Close out the head and open the body. */
371 
372 	man_unscope(man, n);
373 	roff_body_alloc(man, line, ppos, tok);
374 	man->flags &= ~ROFF_NONOFILL;
375 }
376 
377 void
378 in_line_eoln(MACRO_PROT_ARGS)
379 {
380 	int		 la;
381 	char		*p;
382 	struct roff_node *n;
383 
384 	roff_elem_alloc(man, line, ppos, tok);
385 	n = man->last;
386 
387 	if (tok == MAN_EX)
388 		man->flags |= ROFF_NOFILL;
389 	else if (tok == MAN_EE)
390 		man->flags &= ~ROFF_NOFILL;
391 
392 	for (;;) {
393 		if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {
394 			mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
395 			    "%s ... %s", roff_name[tok], buf + *pos);
396 			break;
397 		}
398 		la = *pos;
399 		if ( ! man_args(man, line, pos, buf, &p))
400 			break;
401 		if (man_macro(tok)->flags & MAN_JOIN &&
402 		    man->last->type == ROFFT_TEXT)
403 			roff_word_append(man, p);
404 		else
405 			roff_word_alloc(man, line, la, p);
406 		free(p);
407 	}
408 
409 	/*
410 	 * Append NODE_EOS in case the last snipped argument
411 	 * ends with a dot, e.g. `.IR syslog (3).'
412 	 */
413 
414 	if (n != man->last &&
415 	    mandoc_eos(man->last->string, strlen(man->last->string)))
416 		man->last->flags |= NODE_EOS;
417 
418 	/*
419 	 * If no arguments are specified and this is MAN_ESCOPED (i.e.,
420 	 * next-line scoped), then set our mode to indicate that we're
421 	 * waiting for terms to load into our context.
422 	 */
423 
424 	if (n == man->last && man_macro(tok)->flags & MAN_ESCOPED) {
425 		man->flags |= MAN_ELINE;
426 		return;
427 	}
428 
429 	assert(man->last->type != ROFFT_ROOT);
430 	man->next = ROFF_NEXT_SIBLING;
431 
432 	/* Rewind our element scope. */
433 
434 	for ( ; man->last; man->last = man->last->parent) {
435 		man->last->flags |= NODE_VALID;
436 		if (man->last == n)
437 			break;
438 	}
439 
440 	/* Rewind next-line scoped ancestors, if any. */
441 
442 	if (man_macro(tok)->flags & MAN_ESCOPED)
443 		man_descope(man, line, ppos, NULL);
444 }
445 
446 void
447 man_endparse(struct roff_man *man)
448 {
449 	man_unscope(man, man->meta.first);
450 }
451 
452 static int
453 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
454 {
455 	char	 *start;
456 
457 	assert(*pos);
458 	*v = start = buf + *pos;
459 	assert(' ' != *start);
460 
461 	if ('\0' == *start)
462 		return 0;
463 
464 	*v = roff_getarg(man->roff, v, line, pos);
465 	return 1;
466 }
467