xref: /freebsd/contrib/mandoc/man_macro.c (revision a0b9e2e854027e6ff61fb075a1309dbc71c42b54)
1 /*	$Id: man_macro.c,v 1.144 2019/01/05 18:59:46 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 "config.h"
20 
21 #include <sys/types.h>
22 
23 #include <assert.h>
24 #include <ctype.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 
29 #include "mandoc.h"
30 #include "roff.h"
31 #include "man.h"
32 #include "libmandoc.h"
33 #include "roff_int.h"
34 #include "libman.h"
35 
36 static	void		 blk_close(MACRO_PROT_ARGS);
37 static	void		 blk_exp(MACRO_PROT_ARGS);
38 static	void		 blk_imp(MACRO_PROT_ARGS);
39 static	void		 in_line_eoln(MACRO_PROT_ARGS);
40 static	int		 man_args(struct roff_man *, int,
41 				int *, char *, char **);
42 static	void		 rew_scope(struct roff_man *, enum roff_tok);
43 
44 static const struct man_macro man_macros[MAN_MAX - MAN_TH] = {
45 	{ in_line_eoln, MAN_XSCOPE }, /* TH */
46 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SH */
47 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SS */
48 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TP */
49 	{ blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TQ */
50 	{ blk_imp, MAN_XSCOPE }, /* LP */
51 	{ blk_imp, MAN_XSCOPE }, /* PP */
52 	{ blk_imp, MAN_XSCOPE }, /* P */
53 	{ blk_imp, MAN_XSCOPE }, /* IP */
54 	{ blk_imp, MAN_XSCOPE }, /* HP */
55 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SM */
56 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SB */
57 	{ in_line_eoln, 0 }, /* BI */
58 	{ in_line_eoln, 0 }, /* IB */
59 	{ in_line_eoln, 0 }, /* BR */
60 	{ in_line_eoln, 0 }, /* RB */
61 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* R */
62 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* B */
63 	{ in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* I */
64 	{ in_line_eoln, 0 }, /* IR */
65 	{ in_line_eoln, 0 }, /* RI */
66 	{ blk_close, MAN_XSCOPE }, /* RE */
67 	{ blk_exp, MAN_XSCOPE }, /* 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, MAN_NSCOPED }, /* in */
73 	{ blk_imp, MAN_XSCOPE }, /* SY */
74 	{ blk_close, MAN_XSCOPE }, /* YS */
75 	{ in_line_eoln, 0 }, /* OP */
76 	{ in_line_eoln, MAN_XSCOPE }, /* EX */
77 	{ in_line_eoln, MAN_XSCOPE }, /* EE */
78 	{ blk_exp, MAN_XSCOPE }, /* UR */
79 	{ blk_close, MAN_XSCOPE }, /* UE */
80 	{ blk_exp, MAN_XSCOPE }, /* MT */
81 	{ blk_close, MAN_XSCOPE }, /* ME */
82 };
83 
84 
85 const struct man_macro *
86 man_macro(enum roff_tok tok)
87 {
88 	assert(tok >= MAN_TH && tok <= MAN_MAX);
89 	return man_macros + (tok - MAN_TH);
90 }
91 
92 void
93 man_unscope(struct roff_man *man, const struct roff_node *to)
94 {
95 	struct roff_node *n;
96 
97 	to = to->parent;
98 	n = man->last;
99 	while (n != to) {
100 
101 		/* Reached the end of the document? */
102 
103 		if (to == NULL && ! (n->flags & NODE_VALID)) {
104 			if (man->flags & (MAN_BLINE | MAN_ELINE) &&
105 			    man_macro(n->tok)->flags &
106 			     (MAN_BSCOPED | MAN_NSCOPED)) {
107 				mandoc_msg(MANDOCERR_BLK_LINE,
108 				    n->line, n->pos,
109 				    "EOF breaks %s", roff_name[n->tok]);
110 				if (man->flags & MAN_ELINE)
111 					man->flags &= ~MAN_ELINE;
112 				else {
113 					assert(n->type == ROFFT_HEAD);
114 					n = n->parent;
115 					man->flags &= ~MAN_BLINE;
116 				}
117 				man->last = n;
118 				n = n->parent;
119 				roff_node_delete(man, man->last);
120 				continue;
121 			}
122 			if (n->type == ROFFT_BLOCK &&
123 			    man_macro(n->tok)->fp == blk_exp)
124 				mandoc_msg(MANDOCERR_BLK_NOEND,
125 				    n->line, n->pos, "%s",
126 				    roff_name[n->tok]);
127 		}
128 
129 		/*
130 		 * We might delete the man->last node
131 		 * in the post-validation phase.
132 		 * Save a pointer to the parent such that
133 		 * we know where to continue the iteration.
134 		 */
135 
136 		man->last = n;
137 		n = n->parent;
138 		man->last->flags |= NODE_VALID;
139 	}
140 
141 	/*
142 	 * If we ended up at the parent of the node we were
143 	 * supposed to rewind to, that means the target node
144 	 * got deleted, so add the next node we parse as a child
145 	 * of the parent instead of as a sibling of the target.
146 	 */
147 
148 	man->next = (man->last == to) ?
149 	    ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;
150 }
151 
152 /*
153  * Rewinding entails ascending the parse tree until a coherent point,
154  * for example, the `SH' macro will close out any intervening `SS'
155  * scopes.  When a scope is closed, it must be validated and actioned.
156  */
157 static void
158 rew_scope(struct roff_man *man, enum roff_tok tok)
159 {
160 	struct roff_node *n;
161 
162 	/* Preserve empty paragraphs before RS. */
163 
164 	n = man->last;
165 	if (tok == MAN_RS && n->child == NULL &&
166 	    (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
167 		return;
168 
169 	for (;;) {
170 		if (n->type == ROFFT_ROOT)
171 			return;
172 		if (n->flags & NODE_VALID) {
173 			n = n->parent;
174 			continue;
175 		}
176 		if (n->type != ROFFT_BLOCK) {
177 			if (n->parent->type == ROFFT_ROOT) {
178 				man_unscope(man, n);
179 				return;
180 			} else {
181 				n = n->parent;
182 				continue;
183 			}
184 		}
185 		if (tok != MAN_SH && (n->tok == MAN_SH ||
186 		    (tok != MAN_SS && (n->tok == MAN_SS ||
187 		     man_macro(n->tok)->fp == blk_exp))))
188 			return;
189 		man_unscope(man, n);
190 		n = man->last;
191 	}
192 }
193 
194 
195 /*
196  * Close out a generic explicit macro.
197  */
198 void
199 blk_close(MACRO_PROT_ARGS)
200 {
201 	enum roff_tok		 ctok, ntok;
202 	const struct roff_node	*nn;
203 	char			*p, *ep;
204 	int			 cline, cpos, la, nrew, target;
205 
206 	nrew = 1;
207 	switch (tok) {
208 	case MAN_RE:
209 		ntok = MAN_RS;
210 		la = *pos;
211 		if ( ! man_args(man, line, pos, buf, &p))
212 			break;
213 		for (nn = man->last->parent; nn; nn = nn->parent)
214 			if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
215 				nrew++;
216 		target = strtol(p, &ep, 10);
217 		if (*ep != '\0')
218 			mandoc_msg(MANDOCERR_ARG_EXCESS, line,
219 			    la + (buf[la] == '"') + (int)(ep - p),
220 			    "RE ... %s", ep);
221 		free(p);
222 		if (target == 0)
223 			target = 1;
224 		nrew -= target;
225 		if (nrew < 1) {
226 			mandoc_msg(MANDOCERR_RE_NOTOPEN,
227 			    line, ppos, "RE %d", target);
228 			return;
229 		}
230 		break;
231 	case MAN_YS:
232 		ntok = MAN_SY;
233 		break;
234 	case MAN_UE:
235 		ntok = MAN_UR;
236 		break;
237 	case MAN_ME:
238 		ntok = MAN_MT;
239 		break;
240 	default:
241 		abort();
242 	}
243 
244 	for (nn = man->last->parent; nn; nn = nn->parent)
245 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
246 			break;
247 
248 	if (nn == NULL) {
249 		mandoc_msg(MANDOCERR_BLK_NOTOPEN,
250 		    line, ppos, "%s", roff_name[tok]);
251 		rew_scope(man, MAN_PP);
252 		if (tok == MAN_RE) {
253 			roff_elem_alloc(man, line, ppos, ROFF_br);
254 			man->last->flags |= NODE_LINE |
255 			    NODE_VALID | NODE_ENDED;
256 			man->next = ROFF_NEXT_SIBLING;
257 		}
258 		return;
259 	}
260 
261 	cline = man->last->line;
262 	cpos = man->last->pos;
263 	ctok = man->last->tok;
264 	man_unscope(man, nn);
265 
266 	if (tok == MAN_RE && nn->head->aux > 0)
267 		roff_setreg(man->roff, "an-margin", nn->head->aux, '-');
268 
269 	/* Trailing text. */
270 
271 	if (buf[*pos] != '\0') {
272 		roff_word_alloc(man, line, ppos, buf + *pos);
273 		man->last->flags |= NODE_DELIMC;
274 		if (mandoc_eos(man->last->string, strlen(man->last->string)))
275 			man->last->flags |= NODE_EOS;
276 	}
277 
278 	/* Move a trailing paragraph behind the block. */
279 
280 	if (ctok == MAN_LP || ctok == MAN_PP || ctok == MAN_P) {
281 		*pos = strlen(buf);
282 		blk_imp(man, ctok, cline, cpos, pos, buf);
283 	}
284 
285 	/* Synopsis blocks need an explicit end marker for spacing. */
286 
287 	if (tok == MAN_YS && man->last == nn) {
288 		roff_elem_alloc(man, line, ppos, tok);
289 		man_unscope(man, man->last);
290 	}
291 }
292 
293 void
294 blk_exp(MACRO_PROT_ARGS)
295 {
296 	struct roff_node *head;
297 	char		*p;
298 	int		 la;
299 
300 	if (tok == MAN_RS) {
301 		rew_scope(man, tok);
302 		man->flags |= ROFF_NONOFILL;
303 	}
304 	roff_block_alloc(man, line, ppos, tok);
305 	head = roff_head_alloc(man, line, ppos, tok);
306 
307 	la = *pos;
308 	if (man_args(man, line, pos, buf, &p)) {
309 		roff_word_alloc(man, line, la, p);
310 		if (tok == MAN_RS) {
311 			if (roff_getreg(man->roff, "an-margin") == 0)
312 				roff_setreg(man->roff, "an-margin",
313 				    7 * 24, '=');
314 			if ((head->aux = strtod(p, NULL) * 24.0) > 0)
315 				roff_setreg(man->roff, "an-margin",
316 				    head->aux, '+');
317 		}
318 		free(p);
319 	}
320 
321 	if (buf[*pos] != '\0')
322 		mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
323 		    "%s ... %s", roff_name[tok], buf + *pos);
324 
325 	man_unscope(man, head);
326 	roff_body_alloc(man, line, ppos, tok);
327 	man->flags &= ~ROFF_NONOFILL;
328 }
329 
330 /*
331  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
332  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
333  * scopes, such as `SH' closing out an `SS', are defined in the rew
334  * routines.
335  */
336 void
337 blk_imp(MACRO_PROT_ARGS)
338 {
339 	int		 la;
340 	char		*p;
341 	struct roff_node *n;
342 
343 	rew_scope(man, tok);
344 	man->flags |= ROFF_NONOFILL;
345 	if (tok == MAN_SH || tok == MAN_SS)
346 		man->flags &= ~ROFF_NOFILL;
347 	roff_block_alloc(man, line, ppos, tok);
348 	n = roff_head_alloc(man, line, ppos, tok);
349 
350 	/* Add line arguments. */
351 
352 	for (;;) {
353 		la = *pos;
354 		if ( ! man_args(man, line, pos, buf, &p))
355 			break;
356 		roff_word_alloc(man, line, la, p);
357 		free(p);
358 	}
359 
360 	/*
361 	 * For macros having optional next-line scope,
362 	 * keep the head open if there were no arguments.
363 	 * For `TP' and `TQ', always keep the head open.
364 	 */
365 
366 	if (man_macro(tok)->flags & MAN_BSCOPED &&
367 	    (tok == MAN_TP || tok == MAN_TQ || n == man->last)) {
368 		man->flags |= MAN_BLINE;
369 		return;
370 	}
371 
372 	/* Close out the head and open the body. */
373 
374 	man_unscope(man, n);
375 	roff_body_alloc(man, line, ppos, tok);
376 	man->flags &= ~ROFF_NONOFILL;
377 }
378 
379 void
380 in_line_eoln(MACRO_PROT_ARGS)
381 {
382 	int		 la;
383 	char		*p;
384 	struct roff_node *n;
385 
386 	roff_elem_alloc(man, line, ppos, tok);
387 	n = man->last;
388 
389 	if (tok == MAN_EX)
390 		man->flags |= ROFF_NOFILL;
391 	else if (tok == MAN_EE)
392 		man->flags &= ~ROFF_NOFILL;
393 
394 	for (;;) {
395 		if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {
396 			mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
397 			    "%s ... %s", roff_name[tok], buf + *pos);
398 			break;
399 		}
400 		la = *pos;
401 		if ( ! man_args(man, line, pos, buf, &p))
402 			break;
403 		if (man_macro(tok)->flags & MAN_JOIN &&
404 		    man->last->type == ROFFT_TEXT)
405 			roff_word_append(man, p);
406 		else
407 			roff_word_alloc(man, line, la, p);
408 		free(p);
409 	}
410 
411 	/*
412 	 * Append NODE_EOS in case the last snipped argument
413 	 * ends with a dot, e.g. `.IR syslog (3).'
414 	 */
415 
416 	if (n != man->last &&
417 	    mandoc_eos(man->last->string, strlen(man->last->string)))
418 		man->last->flags |= NODE_EOS;
419 
420 	/*
421 	 * If no arguments are specified and this is MAN_ESCOPED (i.e.,
422 	 * next-line scoped), then set our mode to indicate that we're
423 	 * waiting for terms to load into our context.
424 	 */
425 
426 	if (n == man->last && man_macro(tok)->flags & MAN_ESCOPED) {
427 		man->flags |= MAN_ELINE;
428 		return;
429 	}
430 
431 	assert(man->last->type != ROFFT_ROOT);
432 	man->next = ROFF_NEXT_SIBLING;
433 
434 	/* Rewind our element scope. */
435 
436 	for ( ; man->last; man->last = man->last->parent) {
437 		man->last->flags |= NODE_VALID;
438 		if (man->last == n)
439 			break;
440 	}
441 
442 	/* Rewind next-line scoped ancestors, if any. */
443 
444 	if (man_macro(tok)->flags & MAN_ESCOPED)
445 		man_descope(man, line, ppos, NULL);
446 }
447 
448 void
449 man_endparse(struct roff_man *man)
450 {
451 	man_unscope(man, man->meta.first);
452 }
453 
454 static int
455 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
456 {
457 	char	 *start;
458 
459 	assert(*pos);
460 	*v = start = buf + *pos;
461 	assert(' ' != *start);
462 
463 	if ('\0' == *start)
464 		return 0;
465 
466 	*v = roff_getarg(man->roff, v, line, pos);
467 	return 1;
468 }
469