xref: /freebsd/contrib/mandoc/mdoc_argv.c (revision a63915c2d7ff177ce364488f86eff99949402051)
1 /*	$Id: mdoc_argv.c,v 1.120 2019/07/11 17:06:17 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012, 2014-2019 Ingo Schwarze <schwarze@openbsd.org>
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17  */
18 #include "config.h"
19 
20 #include <sys/types.h>
21 
22 #include <assert.h>
23 #include <stdlib.h>
24 #include <stdio.h>
25 #include <string.h>
26 
27 #include "mandoc_aux.h"
28 #include "mandoc.h"
29 #include "roff.h"
30 #include "mdoc.h"
31 #include "libmandoc.h"
32 #include "roff_int.h"
33 #include "libmdoc.h"
34 
35 #define	MULTI_STEP	 5 /* pre-allocate argument values */
36 #define	DELIMSZ		 6 /* max possible size of a delimiter */
37 
38 enum	argsflag {
39 	ARGSFL_NONE = 0,
40 	ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
41 	ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
42 };
43 
44 enum	argvflag {
45 	ARGV_NONE, /* no args to flag (e.g., -split) */
46 	ARGV_SINGLE, /* one arg to flag (e.g., -file xxx)  */
47 	ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
48 };
49 
50 struct	mdocarg {
51 	enum argsflag	 flags;
52 	const enum mdocargt *argvs;
53 };
54 
55 static	void		 argn_free(struct mdoc_arg *, int);
56 static	enum margserr	 args(struct roff_man *, int, int *,
57 				char *, enum argsflag, char **);
58 static	int		 args_checkpunct(const char *, int);
59 static	void		 argv_multi(struct roff_man *, int,
60 				struct mdoc_argv *, int *, char *);
61 static	void		 argv_single(struct roff_man *, int,
62 				struct mdoc_argv *, int *, char *);
63 
64 static	const enum argvflag argvflags[MDOC_ARG_MAX] = {
65 	ARGV_NONE,	/* MDOC_Split */
66 	ARGV_NONE,	/* MDOC_Nosplit */
67 	ARGV_NONE,	/* MDOC_Ragged */
68 	ARGV_NONE,	/* MDOC_Unfilled */
69 	ARGV_NONE,	/* MDOC_Literal */
70 	ARGV_SINGLE,	/* MDOC_File */
71 	ARGV_SINGLE,	/* MDOC_Offset */
72 	ARGV_NONE,	/* MDOC_Bullet */
73 	ARGV_NONE,	/* MDOC_Dash */
74 	ARGV_NONE,	/* MDOC_Hyphen */
75 	ARGV_NONE,	/* MDOC_Item */
76 	ARGV_NONE,	/* MDOC_Enum */
77 	ARGV_NONE,	/* MDOC_Tag */
78 	ARGV_NONE,	/* MDOC_Diag */
79 	ARGV_NONE,	/* MDOC_Hang */
80 	ARGV_NONE,	/* MDOC_Ohang */
81 	ARGV_NONE,	/* MDOC_Inset */
82 	ARGV_MULTI,	/* MDOC_Column */
83 	ARGV_SINGLE,	/* MDOC_Width */
84 	ARGV_NONE,	/* MDOC_Compact */
85 	ARGV_NONE,	/* MDOC_Std */
86 	ARGV_NONE,	/* MDOC_Filled */
87 	ARGV_NONE,	/* MDOC_Words */
88 	ARGV_NONE,	/* MDOC_Emphasis */
89 	ARGV_NONE,	/* MDOC_Symbolic */
90 	ARGV_NONE	/* MDOC_Symbolic */
91 };
92 
93 static	const enum mdocargt args_Ex[] = {
94 	MDOC_Std,
95 	MDOC_ARG_MAX
96 };
97 
98 static	const enum mdocargt args_An[] = {
99 	MDOC_Split,
100 	MDOC_Nosplit,
101 	MDOC_ARG_MAX
102 };
103 
104 static	const enum mdocargt args_Bd[] = {
105 	MDOC_Ragged,
106 	MDOC_Unfilled,
107 	MDOC_Filled,
108 	MDOC_Literal,
109 	MDOC_File,
110 	MDOC_Offset,
111 	MDOC_Compact,
112 	MDOC_Centred,
113 	MDOC_ARG_MAX
114 };
115 
116 static	const enum mdocargt args_Bf[] = {
117 	MDOC_Emphasis,
118 	MDOC_Literal,
119 	MDOC_Symbolic,
120 	MDOC_ARG_MAX
121 };
122 
123 static	const enum mdocargt args_Bk[] = {
124 	MDOC_Words,
125 	MDOC_ARG_MAX
126 };
127 
128 static	const enum mdocargt args_Bl[] = {
129 	MDOC_Bullet,
130 	MDOC_Dash,
131 	MDOC_Hyphen,
132 	MDOC_Item,
133 	MDOC_Enum,
134 	MDOC_Tag,
135 	MDOC_Diag,
136 	MDOC_Hang,
137 	MDOC_Ohang,
138 	MDOC_Inset,
139 	MDOC_Column,
140 	MDOC_Width,
141 	MDOC_Offset,
142 	MDOC_Compact,
143 	MDOC_Nested,
144 	MDOC_ARG_MAX
145 };
146 
147 static	const struct mdocarg mdocargs[MDOC_MAX - MDOC_Dd] = {
148 	{ ARGSFL_NONE, NULL }, /* Dd */
149 	{ ARGSFL_NONE, NULL }, /* Dt */
150 	{ ARGSFL_NONE, NULL }, /* Os */
151 	{ ARGSFL_NONE, NULL }, /* Sh */
152 	{ ARGSFL_NONE, NULL }, /* Ss */
153 	{ ARGSFL_NONE, NULL }, /* Pp */
154 	{ ARGSFL_DELIM, NULL }, /* D1 */
155 	{ ARGSFL_DELIM, NULL }, /* Dl */
156 	{ ARGSFL_NONE, args_Bd }, /* Bd */
157 	{ ARGSFL_NONE, NULL }, /* Ed */
158 	{ ARGSFL_NONE, args_Bl }, /* Bl */
159 	{ ARGSFL_NONE, NULL }, /* El */
160 	{ ARGSFL_NONE, NULL }, /* It */
161 	{ ARGSFL_DELIM, NULL }, /* Ad */
162 	{ ARGSFL_DELIM, args_An }, /* An */
163 	{ ARGSFL_DELIM, NULL }, /* Ap */
164 	{ ARGSFL_DELIM, NULL }, /* Ar */
165 	{ ARGSFL_DELIM, NULL }, /* Cd */
166 	{ ARGSFL_DELIM, NULL }, /* Cm */
167 	{ ARGSFL_DELIM, NULL }, /* Dv */
168 	{ ARGSFL_DELIM, NULL }, /* Er */
169 	{ ARGSFL_DELIM, NULL }, /* Ev */
170 	{ ARGSFL_NONE, args_Ex }, /* Ex */
171 	{ ARGSFL_DELIM, NULL }, /* Fa */
172 	{ ARGSFL_NONE, NULL }, /* Fd */
173 	{ ARGSFL_DELIM, NULL }, /* Fl */
174 	{ ARGSFL_DELIM, NULL }, /* Fn */
175 	{ ARGSFL_DELIM, NULL }, /* Ft */
176 	{ ARGSFL_DELIM, NULL }, /* Ic */
177 	{ ARGSFL_DELIM, NULL }, /* In */
178 	{ ARGSFL_DELIM, NULL }, /* Li */
179 	{ ARGSFL_NONE, NULL }, /* Nd */
180 	{ ARGSFL_DELIM, NULL }, /* Nm */
181 	{ ARGSFL_DELIM, NULL }, /* Op */
182 	{ ARGSFL_DELIM, NULL }, /* Ot */
183 	{ ARGSFL_DELIM, NULL }, /* Pa */
184 	{ ARGSFL_NONE, args_Ex }, /* Rv */
185 	{ ARGSFL_DELIM, NULL }, /* St */
186 	{ ARGSFL_DELIM, NULL }, /* Va */
187 	{ ARGSFL_DELIM, NULL }, /* Vt */
188 	{ ARGSFL_DELIM, NULL }, /* Xr */
189 	{ ARGSFL_NONE, NULL }, /* %A */
190 	{ ARGSFL_NONE, NULL }, /* %B */
191 	{ ARGSFL_NONE, NULL }, /* %D */
192 	{ ARGSFL_NONE, NULL }, /* %I */
193 	{ ARGSFL_NONE, NULL }, /* %J */
194 	{ ARGSFL_NONE, NULL }, /* %N */
195 	{ ARGSFL_NONE, NULL }, /* %O */
196 	{ ARGSFL_NONE, NULL }, /* %P */
197 	{ ARGSFL_NONE, NULL }, /* %R */
198 	{ ARGSFL_NONE, NULL }, /* %T */
199 	{ ARGSFL_NONE, NULL }, /* %V */
200 	{ ARGSFL_DELIM, NULL }, /* Ac */
201 	{ ARGSFL_NONE, NULL }, /* Ao */
202 	{ ARGSFL_DELIM, NULL }, /* Aq */
203 	{ ARGSFL_DELIM, NULL }, /* At */
204 	{ ARGSFL_DELIM, NULL }, /* Bc */
205 	{ ARGSFL_NONE, args_Bf }, /* Bf */
206 	{ ARGSFL_NONE, NULL }, /* Bo */
207 	{ ARGSFL_DELIM, NULL }, /* Bq */
208 	{ ARGSFL_DELIM, NULL }, /* Bsx */
209 	{ ARGSFL_DELIM, NULL }, /* Bx */
210 	{ ARGSFL_NONE, NULL }, /* Db */
211 	{ ARGSFL_DELIM, NULL }, /* Dc */
212 	{ ARGSFL_NONE, NULL }, /* Do */
213 	{ ARGSFL_DELIM, NULL }, /* Dq */
214 	{ ARGSFL_DELIM, NULL }, /* Ec */
215 	{ ARGSFL_NONE, NULL }, /* Ef */
216 	{ ARGSFL_DELIM, NULL }, /* Em */
217 	{ ARGSFL_NONE, NULL }, /* Eo */
218 	{ ARGSFL_DELIM, NULL }, /* Fx */
219 	{ ARGSFL_DELIM, NULL }, /* Ms */
220 	{ ARGSFL_DELIM, NULL }, /* No */
221 	{ ARGSFL_DELIM, NULL }, /* Ns */
222 	{ ARGSFL_DELIM, NULL }, /* Nx */
223 	{ ARGSFL_DELIM, NULL }, /* Ox */
224 	{ ARGSFL_DELIM, NULL }, /* Pc */
225 	{ ARGSFL_DELIM, NULL }, /* Pf */
226 	{ ARGSFL_NONE, NULL }, /* Po */
227 	{ ARGSFL_DELIM, NULL }, /* Pq */
228 	{ ARGSFL_DELIM, NULL }, /* Qc */
229 	{ ARGSFL_DELIM, NULL }, /* Ql */
230 	{ ARGSFL_NONE, NULL }, /* Qo */
231 	{ ARGSFL_DELIM, NULL }, /* Qq */
232 	{ ARGSFL_NONE, NULL }, /* Re */
233 	{ ARGSFL_NONE, NULL }, /* Rs */
234 	{ ARGSFL_DELIM, NULL }, /* Sc */
235 	{ ARGSFL_NONE, NULL }, /* So */
236 	{ ARGSFL_DELIM, NULL }, /* Sq */
237 	{ ARGSFL_NONE, NULL }, /* Sm */
238 	{ ARGSFL_DELIM, NULL }, /* Sx */
239 	{ ARGSFL_DELIM, NULL }, /* Sy */
240 	{ ARGSFL_DELIM, NULL }, /* Tn */
241 	{ ARGSFL_DELIM, NULL }, /* Ux */
242 	{ ARGSFL_DELIM, NULL }, /* Xc */
243 	{ ARGSFL_NONE, NULL }, /* Xo */
244 	{ ARGSFL_NONE, NULL }, /* Fo */
245 	{ ARGSFL_DELIM, NULL }, /* Fc */
246 	{ ARGSFL_NONE, NULL }, /* Oo */
247 	{ ARGSFL_DELIM, NULL }, /* Oc */
248 	{ ARGSFL_NONE, args_Bk }, /* Bk */
249 	{ ARGSFL_NONE, NULL }, /* Ek */
250 	{ ARGSFL_NONE, NULL }, /* Bt */
251 	{ ARGSFL_NONE, NULL }, /* Hf */
252 	{ ARGSFL_DELIM, NULL }, /* Fr */
253 	{ ARGSFL_NONE, NULL }, /* Ud */
254 	{ ARGSFL_DELIM, NULL }, /* Lb */
255 	{ ARGSFL_NONE, NULL }, /* Lp */
256 	{ ARGSFL_DELIM, NULL }, /* Lk */
257 	{ ARGSFL_DELIM, NULL }, /* Mt */
258 	{ ARGSFL_DELIM, NULL }, /* Brq */
259 	{ ARGSFL_NONE, NULL }, /* Bro */
260 	{ ARGSFL_DELIM, NULL }, /* Brc */
261 	{ ARGSFL_NONE, NULL }, /* %C */
262 	{ ARGSFL_NONE, NULL }, /* Es */
263 	{ ARGSFL_DELIM, NULL }, /* En */
264 	{ ARGSFL_DELIM, NULL }, /* Dx */
265 	{ ARGSFL_NONE, NULL }, /* %Q */
266 	{ ARGSFL_NONE, NULL }, /* %U */
267 	{ ARGSFL_NONE, NULL }, /* Ta */
268 };
269 
270 
271 /*
272  * Parse flags and their arguments from the input line.
273  * These come in the form -flag [argument ...].
274  * Some flags take no argument, some one, some multiple.
275  */
276 void
mdoc_argv(struct roff_man * mdoc,int line,enum roff_tok tok,struct mdoc_arg ** reta,int * pos,char * buf)277 mdoc_argv(struct roff_man *mdoc, int line, enum roff_tok tok,
278 	struct mdoc_arg **reta, int *pos, char *buf)
279 {
280 	struct mdoc_argv	  tmpv;
281 	struct mdoc_argv	**retv;
282 	const enum mdocargt	 *argtable;
283 	char			 *argname;
284 	int			  ipos, retc;
285 	char			  savechar;
286 
287 	*reta = NULL;
288 
289 	/* Which flags does this macro support? */
290 
291 	assert(tok >= MDOC_Dd && tok < MDOC_MAX);
292 	argtable = mdocargs[tok - MDOC_Dd].argvs;
293 	if (argtable == NULL)
294 		return;
295 
296 	/* Loop over the flags on the input line. */
297 
298 	ipos = *pos;
299 	while (buf[ipos] == '-') {
300 
301 		/* Seek to the first unescaped space. */
302 
303 		for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
304 			if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
305 				break;
306 
307 		/*
308 		 * We want to nil-terminate the word to look it up.
309 		 * But we may not have a flag, in which case we need
310 		 * to restore the line as-is.  So keep around the
311 		 * stray byte, which we'll reset upon exiting.
312 		 */
313 
314 		if ((savechar = buf[ipos]) != '\0')
315 			buf[ipos++] = '\0';
316 
317 		/*
318 		 * Now look up the word as a flag.  Use temporary
319 		 * storage that we'll copy into the node's flags.
320 		 */
321 
322 		while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
323 			if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
324 				break;
325 
326 		/* If it isn't a flag, restore the saved byte. */
327 
328 		if (tmpv.arg == MDOC_ARG_MAX) {
329 			if (savechar != '\0')
330 				buf[ipos - 1] = savechar;
331 			break;
332 		}
333 
334 		/* Read to the next word (the first argument). */
335 
336 		while (buf[ipos] == ' ')
337 			ipos++;
338 
339 		/* Parse the arguments of the flag. */
340 
341 		tmpv.line  = line;
342 		tmpv.pos   = *pos;
343 		tmpv.sz    = 0;
344 		tmpv.value = NULL;
345 
346 		switch (argvflags[tmpv.arg]) {
347 		case ARGV_SINGLE:
348 			argv_single(mdoc, line, &tmpv, &ipos, buf);
349 			break;
350 		case ARGV_MULTI:
351 			argv_multi(mdoc, line, &tmpv, &ipos, buf);
352 			break;
353 		case ARGV_NONE:
354 			break;
355 		}
356 
357 		/* Append to the return values. */
358 
359 		if (*reta == NULL)
360 			*reta = mandoc_calloc(1, sizeof(**reta));
361 
362 		retc = ++(*reta)->argc;
363 		retv = &(*reta)->argv;
364 		*retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
365 		memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
366 
367 		/* Prepare for parsing the next flag. */
368 
369 		*pos = ipos;
370 		argtable = mdocargs[tok - MDOC_Dd].argvs;
371 	}
372 }
373 
374 void
mdoc_argv_free(struct mdoc_arg * p)375 mdoc_argv_free(struct mdoc_arg *p)
376 {
377 	int		 i;
378 
379 	if (NULL == p)
380 		return;
381 
382 	if (p->refcnt) {
383 		--(p->refcnt);
384 		if (p->refcnt)
385 			return;
386 	}
387 	assert(p->argc);
388 
389 	for (i = (int)p->argc - 1; i >= 0; i--)
390 		argn_free(p, i);
391 
392 	free(p->argv);
393 	free(p);
394 }
395 
396 static void
argn_free(struct mdoc_arg * p,int iarg)397 argn_free(struct mdoc_arg *p, int iarg)
398 {
399 	struct mdoc_argv *arg;
400 	int		  j;
401 
402 	arg = &p->argv[iarg];
403 
404 	if (arg->sz && arg->value) {
405 		for (j = (int)arg->sz - 1; j >= 0; j--)
406 			free(arg->value[j]);
407 		free(arg->value);
408 	}
409 
410 	for (--p->argc; iarg < (int)p->argc; iarg++)
411 		p->argv[iarg] = p->argv[iarg+1];
412 }
413 
414 enum margserr
mdoc_args(struct roff_man * mdoc,int line,int * pos,char * buf,enum roff_tok tok,char ** v)415 mdoc_args(struct roff_man *mdoc, int line, int *pos,
416 	char *buf, enum roff_tok tok, char **v)
417 {
418 	struct roff_node *n;
419 	enum argsflag	  fl;
420 
421 	fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok - MDOC_Dd].flags;
422 
423 	/*
424 	 * We know that we're in an `It', so it's reasonable to expect
425 	 * us to be sitting in a `Bl'.  Someday this may not be the case
426 	 * (if we allow random `It's sitting out there), so provide a
427 	 * safe fall-back into the default behaviour.
428 	 */
429 
430 	if (tok == MDOC_It) {
431 		for (n = mdoc->last; n != NULL; n = n->parent) {
432 			if (n->tok != MDOC_Bl)
433 				continue;
434 			if (n->norm->Bl.type == LIST_column)
435 				fl = ARGSFL_TABSEP;
436 			break;
437 		}
438 	}
439 
440 	return args(mdoc, line, pos, buf, fl, v);
441 }
442 
443 static enum margserr
args(struct roff_man * mdoc,int line,int * pos,char * buf,enum argsflag fl,char ** v)444 args(struct roff_man *mdoc, int line, int *pos,
445 		char *buf, enum argsflag fl, char **v)
446 {
447 	char		*p;
448 	char		*v_local;
449 	int		 pairs;
450 
451 	if (buf[*pos] == '\0') {
452 		if (mdoc->flags & MDOC_PHRASELIT &&
453 		    ! (mdoc->flags & MDOC_PHRASE)) {
454 			mandoc_msg(MANDOCERR_ARG_QUOTE, line, *pos, NULL);
455 			mdoc->flags &= ~MDOC_PHRASELIT;
456 		}
457 		mdoc->flags &= ~MDOC_PHRASEQL;
458 		return ARGS_EOLN;
459 	}
460 
461 	if (v == NULL)
462 		v = &v_local;
463 	*v = buf + *pos;
464 
465 	if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos))
466 		return ARGS_PUNCT;
467 
468 	/*
469 	 * Tabs in `It' lines in `Bl -column' can't be escaped.
470 	 * Phrases are reparsed for `Ta' and other macros later.
471 	 */
472 
473 	if (fl == ARGSFL_TABSEP) {
474 		if ((p = strchr(*v, '\t')) != NULL) {
475 
476 			/*
477 			 * Words right before and right after
478 			 * tab characters are not parsed,
479 			 * unless there is a blank in between.
480 			 */
481 
482 			if (p > buf && p[-1] != ' ')
483 				mdoc->flags |= MDOC_PHRASEQL;
484 			if (p[1] != ' ')
485 				mdoc->flags |= MDOC_PHRASEQN;
486 
487 			/*
488 			 * One or more blanks after a tab cause
489 			 * one leading blank in the next column.
490 			 * So skip all but one of them.
491 			 */
492 
493 			*pos += (int)(p - *v) + 1;
494 			while (buf[*pos] == ' ' && buf[*pos + 1] == ' ')
495 				(*pos)++;
496 
497 			/*
498 			 * A tab at the end of an input line
499 			 * switches to the next column.
500 			 */
501 
502 			if (buf[*pos] == '\0' || buf[*pos + 1] == '\0')
503 				mdoc->flags |= MDOC_PHRASEQN;
504 		} else {
505 			p = strchr(*v, '\0');
506 			if (p[-1] == ' ')
507 				mandoc_msg(MANDOCERR_SPACE_EOL,
508 				    line, *pos, NULL);
509 			*pos += (int)(p - *v);
510 		}
511 
512 		/* Skip any trailing blank characters. */
513 		while (p > *v && p[-1] == ' ' &&
514 		    (p - 1 == *v || p[-2] != '\\'))
515 			p--;
516 		*p = '\0';
517 
518 		return ARGS_PHRASE;
519 	}
520 
521 	/*
522 	 * Process a quoted literal.  A quote begins with a double-quote
523 	 * and ends with a double-quote NOT preceded by a double-quote.
524 	 * NUL-terminate the literal in place.
525 	 * Collapse pairs of quotes inside quoted literals.
526 	 * Whitespace is NOT involved in literal termination.
527 	 */
528 
529 	if (mdoc->flags & MDOC_PHRASELIT ||
530 	    (mdoc->flags & MDOC_PHRASE && buf[*pos] == '\"')) {
531 		if ((mdoc->flags & MDOC_PHRASELIT) == 0) {
532 			*v = &buf[++(*pos)];
533 			mdoc->flags |= MDOC_PHRASELIT;
534 		}
535 		pairs = 0;
536 		for ( ; buf[*pos]; (*pos)++) {
537 			/* Move following text left after quoted quotes. */
538 			if (pairs)
539 				buf[*pos - pairs] = buf[*pos];
540 			if ('\"' != buf[*pos])
541 				continue;
542 			/* Unquoted quotes end quoted args. */
543 			if ('\"' != buf[*pos + 1])
544 				break;
545 			/* Quoted quotes collapse. */
546 			pairs++;
547 			(*pos)++;
548 		}
549 		if (pairs)
550 			buf[*pos - pairs] = '\0';
551 
552 		if (buf[*pos] == '\0') {
553 			if ( ! (mdoc->flags & MDOC_PHRASE))
554 				mandoc_msg(MANDOCERR_ARG_QUOTE,
555 				    line, *pos, NULL);
556 			return ARGS_WORD;
557 		}
558 
559 		mdoc->flags &= ~MDOC_PHRASELIT;
560 		buf[(*pos)++] = '\0';
561 
562 		if ('\0' == buf[*pos])
563 			return ARGS_WORD;
564 
565 		while (' ' == buf[*pos])
566 			(*pos)++;
567 
568 		if ('\0' == buf[*pos])
569 			mandoc_msg(MANDOCERR_SPACE_EOL, line, *pos, NULL);
570 
571 		return ARGS_WORD;
572 	}
573 
574 	p = &buf[*pos];
575 	*v = roff_getarg(mdoc->roff, &p, line, pos);
576 	if (v == &v_local)
577 		free(*v);
578 
579 	/*
580 	 * After parsing the last word in this phrase,
581 	 * tell lookup() whether or not to interpret it.
582 	 */
583 
584 	if (*p == '\0' && mdoc->flags & MDOC_PHRASEQL) {
585 		mdoc->flags &= ~MDOC_PHRASEQL;
586 		mdoc->flags |= MDOC_PHRASEQF;
587 	}
588 	return ARGS_ALLOC;
589 }
590 
591 /*
592  * Check if the string consists only of space-separated closing
593  * delimiters.  This is a bit of a dance: the first must be a close
594  * delimiter, but it may be followed by middle delimiters.  Arbitrary
595  * whitespace may separate these tokens.
596  */
597 static int
args_checkpunct(const char * buf,int i)598 args_checkpunct(const char *buf, int i)
599 {
600 	int		 j;
601 	char		 dbuf[DELIMSZ];
602 	enum mdelim	 d;
603 
604 	/* First token must be a close-delimiter. */
605 
606 	for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
607 		dbuf[j] = buf[i];
608 
609 	if (DELIMSZ == j)
610 		return 0;
611 
612 	dbuf[j] = '\0';
613 	if (DELIM_CLOSE != mdoc_isdelim(dbuf))
614 		return 0;
615 
616 	while (' ' == buf[i])
617 		i++;
618 
619 	/* Remaining must NOT be open/none. */
620 
621 	while (buf[i]) {
622 		j = 0;
623 		while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
624 			dbuf[j++] = buf[i++];
625 
626 		if (DELIMSZ == j)
627 			return 0;
628 
629 		dbuf[j] = '\0';
630 		d = mdoc_isdelim(dbuf);
631 		if (DELIM_NONE == d || DELIM_OPEN == d)
632 			return 0;
633 
634 		while (' ' == buf[i])
635 			i++;
636 	}
637 
638 	return '\0' == buf[i];
639 }
640 
641 static void
argv_multi(struct roff_man * mdoc,int line,struct mdoc_argv * v,int * pos,char * buf)642 argv_multi(struct roff_man *mdoc, int line,
643 		struct mdoc_argv *v, int *pos, char *buf)
644 {
645 	enum margserr	 ac;
646 	char		*p;
647 
648 	for (v->sz = 0; ; v->sz++) {
649 		if (buf[*pos] == '-')
650 			break;
651 		ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
652 		if (ac == ARGS_EOLN)
653 			break;
654 
655 		if (v->sz % MULTI_STEP == 0)
656 			v->value = mandoc_reallocarray(v->value,
657 			    v->sz + MULTI_STEP, sizeof(char *));
658 
659 		if (ac != ARGS_ALLOC)
660 			p = mandoc_strdup(p);
661 		v->value[(int)v->sz] = p;
662 	}
663 }
664 
665 static void
argv_single(struct roff_man * mdoc,int line,struct mdoc_argv * v,int * pos,char * buf)666 argv_single(struct roff_man *mdoc, int line,
667 		struct mdoc_argv *v, int *pos, char *buf)
668 {
669 	enum margserr	 ac;
670 	char		*p;
671 
672 	ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
673 	if (ac == ARGS_EOLN)
674 		return;
675 
676 	if (ac != ARGS_ALLOC)
677 		p = mandoc_strdup(p);
678 
679 	v->sz = 1;
680 	v->value = mandoc_malloc(sizeof(char *));
681 	v->value[0] = p;
682 }
683