xref: /illumos-gate/usr/src/cmd/mandoc/man_macro.c (revision 4d131170e62381276a07ffc0aeb1b62e527d940c)
1 /* $Id: man_macro.c,v 1.145 2020/09/09 17:01:10 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012-2015, 2017-2020 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 *
man_macro(enum roff_tok tok)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
man_unscope(struct roff_man * man,const struct roff_node * to)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 					if ((man_macro(n->parent->tok)->flags &
112 					    MAN_ESCOPED) == 0)
113 						man->flags &= ~MAN_ELINE;
114 				} else {
115 					assert(n->type == ROFFT_HEAD);
116 					n = n->parent;
117 					man->flags &= ~MAN_BLINE;
118 				}
119 				man->last = n;
120 				n = n->parent;
121 				roff_node_delete(man, man->last);
122 				continue;
123 			}
124 			if (n->type == ROFFT_BLOCK &&
125 			    man_macro(n->tok)->fp == blk_exp)
126 				mandoc_msg(MANDOCERR_BLK_NOEND,
127 				    n->line, n->pos, "%s",
128 				    roff_name[n->tok]);
129 		}
130 
131 		/*
132 		 * We might delete the man->last node
133 		 * in the post-validation phase.
134 		 * Save a pointer to the parent such that
135 		 * we know where to continue the iteration.
136 		 */
137 
138 		man->last = n;
139 		n = n->parent;
140 		man->last->flags |= NODE_VALID;
141 	}
142 
143 	/*
144 	 * If we ended up at the parent of the node we were
145 	 * supposed to rewind to, that means the target node
146 	 * got deleted, so add the next node we parse as a child
147 	 * of the parent instead of as a sibling of the target.
148 	 */
149 
150 	man->next = (man->last == to) ?
151 	    ROFF_NEXT_CHILD : ROFF_NEXT_SIBLING;
152 }
153 
154 /*
155  * Rewinding entails ascending the parse tree until a coherent point,
156  * for example, the `SH' macro will close out any intervening `SS'
157  * scopes.  When a scope is closed, it must be validated and actioned.
158  */
159 static void
rew_scope(struct roff_man * man,enum roff_tok tok)160 rew_scope(struct roff_man *man, enum roff_tok tok)
161 {
162 	struct roff_node *n;
163 
164 	/* Preserve empty paragraphs before RS. */
165 
166 	n = man->last;
167 	if (tok == MAN_RS && n->child == NULL &&
168 	    (n->tok == MAN_P || n->tok == MAN_PP || n->tok == MAN_LP))
169 		return;
170 
171 	for (;;) {
172 		if (n->type == ROFFT_ROOT)
173 			return;
174 		if (n->flags & NODE_VALID) {
175 			n = n->parent;
176 			continue;
177 		}
178 		if (n->type != ROFFT_BLOCK) {
179 			if (n->parent->type == ROFFT_ROOT) {
180 				man_unscope(man, n);
181 				return;
182 			} else {
183 				n = n->parent;
184 				continue;
185 			}
186 		}
187 		if (tok != MAN_SH && (n->tok == MAN_SH ||
188 		    (tok != MAN_SS && (n->tok == MAN_SS ||
189 		     man_macro(n->tok)->fp == blk_exp))))
190 			return;
191 		man_unscope(man, n);
192 		n = man->last;
193 	}
194 }
195 
196 
197 /*
198  * Close out a generic explicit macro.
199  */
200 void
blk_close(MACRO_PROT_ARGS)201 blk_close(MACRO_PROT_ARGS)
202 {
203 	enum roff_tok		 ctok, ntok;
204 	const struct roff_node	*nn;
205 	char			*p, *ep;
206 	int			 cline, cpos, la, nrew, target;
207 
208 	nrew = 1;
209 	switch (tok) {
210 	case MAN_RE:
211 		ntok = MAN_RS;
212 		la = *pos;
213 		if ( ! man_args(man, line, pos, buf, &p))
214 			break;
215 		for (nn = man->last->parent; nn; nn = nn->parent)
216 			if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
217 				nrew++;
218 		target = strtol(p, &ep, 10);
219 		if (*ep != '\0')
220 			mandoc_msg(MANDOCERR_ARG_EXCESS, line,
221 			    la + (buf[la] == '"') + (int)(ep - p),
222 			    "RE ... %s", ep);
223 		free(p);
224 		if (target == 0)
225 			target = 1;
226 		nrew -= target;
227 		if (nrew < 1) {
228 			mandoc_msg(MANDOCERR_RE_NOTOPEN,
229 			    line, ppos, "RE %d", target);
230 			return;
231 		}
232 		break;
233 	case MAN_YS:
234 		ntok = MAN_SY;
235 		break;
236 	case MAN_UE:
237 		ntok = MAN_UR;
238 		break;
239 	case MAN_ME:
240 		ntok = MAN_MT;
241 		break;
242 	default:
243 		abort();
244 	}
245 
246 	for (nn = man->last->parent; nn; nn = nn->parent)
247 		if (nn->tok == ntok && nn->type == ROFFT_BLOCK && ! --nrew)
248 			break;
249 
250 	if (nn == NULL) {
251 		mandoc_msg(MANDOCERR_BLK_NOTOPEN,
252 		    line, ppos, "%s", roff_name[tok]);
253 		rew_scope(man, MAN_PP);
254 		if (tok == MAN_RE) {
255 			roff_elem_alloc(man, line, ppos, ROFF_br);
256 			man->last->flags |= NODE_LINE |
257 			    NODE_VALID | NODE_ENDED;
258 			man->next = ROFF_NEXT_SIBLING;
259 		}
260 		return;
261 	}
262 
263 	cline = man->last->line;
264 	cpos = man->last->pos;
265 	ctok = man->last->tok;
266 	man_unscope(man, nn);
267 
268 	if (tok == MAN_RE && nn->head->aux > 0)
269 		roff_setreg(man->roff, "an-margin", nn->head->aux, '-');
270 
271 	/* Trailing text. */
272 
273 	if (buf[*pos] != '\0') {
274 		roff_word_alloc(man, line, ppos, buf + *pos);
275 		man->last->flags |= NODE_DELIMC;
276 		if (mandoc_eos(man->last->string, strlen(man->last->string)))
277 			man->last->flags |= NODE_EOS;
278 	}
279 
280 	/* Move a trailing paragraph behind the block. */
281 
282 	if (ctok == MAN_LP || ctok == MAN_PP || ctok == MAN_P) {
283 		*pos = strlen(buf);
284 		blk_imp(man, ctok, cline, cpos, pos, buf);
285 	}
286 
287 	/* Synopsis blocks need an explicit end marker for spacing. */
288 
289 	if (tok == MAN_YS && man->last == nn) {
290 		roff_elem_alloc(man, line, ppos, tok);
291 		man_unscope(man, man->last);
292 	}
293 }
294 
295 void
blk_exp(MACRO_PROT_ARGS)296 blk_exp(MACRO_PROT_ARGS)
297 {
298 	struct roff_node *head;
299 	char		*p;
300 	int		 la;
301 
302 	if (tok == MAN_RS) {
303 		rew_scope(man, tok);
304 		man->flags |= ROFF_NONOFILL;
305 	}
306 	roff_block_alloc(man, line, ppos, tok);
307 	head = roff_head_alloc(man, line, ppos, tok);
308 
309 	la = *pos;
310 	if (man_args(man, line, pos, buf, &p)) {
311 		roff_word_alloc(man, line, la, p);
312 		if (tok == MAN_RS) {
313 			if (roff_getreg(man->roff, "an-margin") == 0)
314 				roff_setreg(man->roff, "an-margin",
315 				    7 * 24, '=');
316 			if ((head->aux = strtod(p, NULL) * 24.0) > 0)
317 				roff_setreg(man->roff, "an-margin",
318 				    head->aux, '+');
319 		}
320 		free(p);
321 	}
322 
323 	if (buf[*pos] != '\0')
324 		mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
325 		    "%s ... %s", roff_name[tok], buf + *pos);
326 
327 	man_unscope(man, head);
328 	roff_body_alloc(man, line, ppos, tok);
329 	man->flags &= ~ROFF_NONOFILL;
330 }
331 
332 /*
333  * Parse an implicit-block macro.  These contain a ROFFT_HEAD and a
334  * ROFFT_BODY contained within a ROFFT_BLOCK.  Rules for closing out other
335  * scopes, such as `SH' closing out an `SS', are defined in the rew
336  * routines.
337  */
338 void
blk_imp(MACRO_PROT_ARGS)339 blk_imp(MACRO_PROT_ARGS)
340 {
341 	int		 la;
342 	char		*p;
343 	struct roff_node *n;
344 
345 	rew_scope(man, tok);
346 	man->flags |= ROFF_NONOFILL;
347 	if (tok == MAN_SH || tok == MAN_SS)
348 		man->flags &= ~ROFF_NOFILL;
349 	roff_block_alloc(man, line, ppos, tok);
350 	n = roff_head_alloc(man, line, ppos, tok);
351 
352 	/* Add line arguments. */
353 
354 	for (;;) {
355 		la = *pos;
356 		if ( ! man_args(man, line, pos, buf, &p))
357 			break;
358 		roff_word_alloc(man, line, la, p);
359 		free(p);
360 	}
361 
362 	/*
363 	 * For macros having optional next-line scope,
364 	 * keep the head open if there were no arguments.
365 	 * For `TP' and `TQ', always keep the head open.
366 	 */
367 
368 	if (man_macro(tok)->flags & MAN_BSCOPED &&
369 	    (tok == MAN_TP || tok == MAN_TQ || n == man->last)) {
370 		man->flags |= MAN_BLINE;
371 		return;
372 	}
373 
374 	/* Close out the head and open the body. */
375 
376 	man_unscope(man, n);
377 	roff_body_alloc(man, line, ppos, tok);
378 	man->flags &= ~ROFF_NONOFILL;
379 }
380 
381 void
in_line_eoln(MACRO_PROT_ARGS)382 in_line_eoln(MACRO_PROT_ARGS)
383 {
384 	int		 la;
385 	char		*p;
386 	struct roff_node *n;
387 
388 	roff_elem_alloc(man, line, ppos, tok);
389 	n = man->last;
390 
391 	if (tok == MAN_EX)
392 		man->flags |= ROFF_NOFILL;
393 	else if (tok == MAN_EE)
394 		man->flags &= ~ROFF_NOFILL;
395 
396 	for (;;) {
397 		if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {
398 			mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
399 			    "%s ... %s", roff_name[tok], buf + *pos);
400 			break;
401 		}
402 		la = *pos;
403 		if ( ! man_args(man, line, pos, buf, &p))
404 			break;
405 		if (man_macro(tok)->flags & MAN_JOIN &&
406 		    man->last->type == ROFFT_TEXT)
407 			roff_word_append(man, p);
408 		else
409 			roff_word_alloc(man, line, la, p);
410 		free(p);
411 	}
412 
413 	/*
414 	 * Append NODE_EOS in case the last snipped argument
415 	 * ends with a dot, e.g. `.IR syslog (3).'
416 	 */
417 
418 	if (n != man->last &&
419 	    mandoc_eos(man->last->string, strlen(man->last->string)))
420 		man->last->flags |= NODE_EOS;
421 
422 	/*
423 	 * If no arguments are specified and this is MAN_ESCOPED (i.e.,
424 	 * next-line scoped), then set our mode to indicate that we're
425 	 * waiting for terms to load into our context.
426 	 */
427 
428 	if (n == man->last && man_macro(tok)->flags & MAN_ESCOPED) {
429 		man->flags |= MAN_ELINE;
430 		return;
431 	}
432 
433 	assert(man->last->type != ROFFT_ROOT);
434 	man->next = ROFF_NEXT_SIBLING;
435 
436 	/* Rewind our element scope. */
437 
438 	for ( ; man->last; man->last = man->last->parent) {
439 		man->last->flags |= NODE_VALID;
440 		if (man->last == n)
441 			break;
442 	}
443 
444 	/* Rewind next-line scoped ancestors, if any. */
445 
446 	if (man_macro(tok)->flags & MAN_ESCOPED)
447 		man_descope(man, line, ppos, NULL);
448 }
449 
450 void
man_endparse(struct roff_man * man)451 man_endparse(struct roff_man *man)
452 {
453 	man_unscope(man, man->meta.first);
454 }
455 
456 static int
man_args(struct roff_man * man,int line,int * pos,char * buf,char ** v)457 man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
458 {
459 	char	 *start;
460 
461 	assert(*pos);
462 	*v = start = buf + *pos;
463 	assert(' ' != *start);
464 
465 	if ('\0' == *start)
466 		return 0;
467 
468 	*v = roff_getarg(man->roff, v, line, pos);
469 	return 1;
470 }
471