1 /* $Id: mdoc_argv.c,v 1.82 2012/03/23 05:50:24 kristaps Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
21 #include <sys/types.h>
31 #include "libmandoc.h"
33 #define MULTI_STEP 5 /* pre-allocate argument values */
34 #define DELIMSZ 6 /* max possible size of a delimiter */
38 ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
39 ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
43 ARGV_NONE, /* no args to flag (e.g., -split) */
44 ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
45 ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */
46 ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */
51 const enum mdocargt *argvs;
54 static void argn_free(struct mdoc_arg *, int);
55 static enum margserr args(struct mdoc *, int, int *,
56 char *, enum argsflag, char **);
57 static int args_checkpunct(const char *, int);
58 static int argv_multi(struct mdoc *, int,
59 struct mdoc_argv *, int *, char *);
60 static int argv_opt_single(struct mdoc *, int,
61 struct mdoc_argv *, int *, char *);
62 static int argv_single(struct mdoc *, int,
63 struct mdoc_argv *, int *, char *);
65 static const enum argvflag argvflags[MDOC_ARG_MAX] = {
66 ARGV_NONE, /* MDOC_Split */
67 ARGV_NONE, /* MDOC_Nosplit */
68 ARGV_NONE, /* MDOC_Ragged */
69 ARGV_NONE, /* MDOC_Unfilled */
70 ARGV_NONE, /* MDOC_Literal */
71 ARGV_SINGLE, /* MDOC_File */
72 ARGV_OPT_SINGLE, /* MDOC_Offset */
73 ARGV_NONE, /* MDOC_Bullet */
74 ARGV_NONE, /* MDOC_Dash */
75 ARGV_NONE, /* MDOC_Hyphen */
76 ARGV_NONE, /* MDOC_Item */
77 ARGV_NONE, /* MDOC_Enum */
78 ARGV_NONE, /* MDOC_Tag */
79 ARGV_NONE, /* MDOC_Diag */
80 ARGV_NONE, /* MDOC_Hang */
81 ARGV_NONE, /* MDOC_Ohang */
82 ARGV_NONE, /* MDOC_Inset */
83 ARGV_MULTI, /* MDOC_Column */
84 ARGV_OPT_SINGLE, /* MDOC_Width */
85 ARGV_NONE, /* MDOC_Compact */
86 ARGV_NONE, /* MDOC_Std */
87 ARGV_NONE, /* MDOC_Filled */
88 ARGV_NONE, /* MDOC_Words */
89 ARGV_NONE, /* MDOC_Emphasis */
90 ARGV_NONE, /* MDOC_Symbolic */
91 ARGV_NONE /* MDOC_Symbolic */
94 static const enum mdocargt args_Ex[] = {
99 static const enum mdocargt args_An[] = {
105 static const enum mdocargt args_Bd[] = {
117 static const enum mdocargt args_Bf[] = {
124 static const enum mdocargt args_Bk[] = {
129 static const enum mdocargt args_Bl[] = {
148 static const struct mdocarg mdocargs[MDOC_MAX] = {
149 { ARGSFL_NONE, NULL }, /* Ap */
150 { ARGSFL_NONE, NULL }, /* Dd */
151 { ARGSFL_NONE, NULL }, /* Dt */
152 { ARGSFL_NONE, NULL }, /* Os */
153 { ARGSFL_NONE, NULL }, /* Sh */
154 { ARGSFL_NONE, NULL }, /* Ss */
155 { ARGSFL_NONE, NULL }, /* Pp */
156 { ARGSFL_DELIM, NULL }, /* D1 */
157 { ARGSFL_DELIM, NULL }, /* Dl */
158 { ARGSFL_NONE, args_Bd }, /* Bd */
159 { ARGSFL_NONE, NULL }, /* Ed */
160 { ARGSFL_NONE, args_Bl }, /* Bl */
161 { ARGSFL_NONE, NULL }, /* El */
162 { ARGSFL_NONE, NULL }, /* It */
163 { ARGSFL_DELIM, NULL }, /* Ad */
164 { ARGSFL_DELIM, args_An }, /* An */
165 { ARGSFL_DELIM, NULL }, /* Ar */
166 { ARGSFL_NONE, NULL }, /* Cd */
167 { ARGSFL_DELIM, NULL }, /* Cm */
168 { ARGSFL_DELIM, NULL }, /* Dv */
169 { ARGSFL_DELIM, NULL }, /* Er */
170 { ARGSFL_DELIM, NULL }, /* Ev */
171 { ARGSFL_NONE, args_Ex }, /* Ex */
172 { ARGSFL_DELIM, NULL }, /* Fa */
173 { ARGSFL_NONE, NULL }, /* Fd */
174 { ARGSFL_DELIM, NULL }, /* Fl */
175 { ARGSFL_DELIM, NULL }, /* Fn */
176 { ARGSFL_DELIM, NULL }, /* Ft */
177 { ARGSFL_DELIM, NULL }, /* Ic */
178 { ARGSFL_NONE, NULL }, /* In */
179 { ARGSFL_DELIM, NULL }, /* Li */
180 { ARGSFL_NONE, NULL }, /* Nd */
181 { ARGSFL_DELIM, NULL }, /* Nm */
182 { ARGSFL_DELIM, NULL }, /* Op */
183 { ARGSFL_NONE, NULL }, /* Ot */
184 { ARGSFL_DELIM, NULL }, /* Pa */
185 { ARGSFL_NONE, args_Ex }, /* Rv */
186 { ARGSFL_DELIM, NULL }, /* St */
187 { ARGSFL_DELIM, NULL }, /* Va */
188 { ARGSFL_DELIM, NULL }, /* Vt */
189 { ARGSFL_DELIM, NULL }, /* Xr */
190 { ARGSFL_NONE, NULL }, /* %A */
191 { ARGSFL_NONE, NULL }, /* %B */
192 { ARGSFL_NONE, NULL }, /* %D */
193 { ARGSFL_NONE, NULL }, /* %I */
194 { ARGSFL_NONE, NULL }, /* %J */
195 { ARGSFL_NONE, NULL }, /* %N */
196 { ARGSFL_NONE, NULL }, /* %O */
197 { ARGSFL_NONE, NULL }, /* %P */
198 { ARGSFL_NONE, NULL }, /* %R */
199 { ARGSFL_NONE, NULL }, /* %T */
200 { ARGSFL_NONE, NULL }, /* %V */
201 { ARGSFL_DELIM, NULL }, /* Ac */
202 { ARGSFL_NONE, NULL }, /* Ao */
203 { ARGSFL_DELIM, NULL }, /* Aq */
204 { ARGSFL_DELIM, NULL }, /* At */
205 { ARGSFL_DELIM, NULL }, /* Bc */
206 { ARGSFL_NONE, args_Bf }, /* Bf */
207 { ARGSFL_NONE, NULL }, /* Bo */
208 { ARGSFL_DELIM, NULL }, /* Bq */
209 { ARGSFL_DELIM, NULL }, /* Bsx */
210 { ARGSFL_DELIM, NULL }, /* Bx */
211 { ARGSFL_NONE, NULL }, /* Db */
212 { ARGSFL_DELIM, NULL }, /* Dc */
213 { ARGSFL_NONE, NULL }, /* Do */
214 { ARGSFL_DELIM, NULL }, /* Dq */
215 { ARGSFL_DELIM, NULL }, /* Ec */
216 { ARGSFL_NONE, NULL }, /* Ef */
217 { ARGSFL_DELIM, NULL }, /* Em */
218 { ARGSFL_NONE, NULL }, /* Eo */
219 { ARGSFL_DELIM, NULL }, /* Fx */
220 { ARGSFL_DELIM, NULL }, /* Ms */
221 { ARGSFL_DELIM, NULL }, /* No */
222 { ARGSFL_DELIM, NULL }, /* Ns */
223 { ARGSFL_DELIM, NULL }, /* Nx */
224 { ARGSFL_DELIM, NULL }, /* Ox */
225 { ARGSFL_DELIM, NULL }, /* Pc */
226 { ARGSFL_DELIM, NULL }, /* Pf */
227 { ARGSFL_NONE, NULL }, /* Po */
228 { ARGSFL_DELIM, NULL }, /* Pq */
229 { ARGSFL_DELIM, NULL }, /* Qc */
230 { ARGSFL_DELIM, NULL }, /* Ql */
231 { ARGSFL_NONE, NULL }, /* Qo */
232 { ARGSFL_DELIM, NULL }, /* Qq */
233 { ARGSFL_NONE, NULL }, /* Re */
234 { ARGSFL_NONE, NULL }, /* Rs */
235 { ARGSFL_DELIM, NULL }, /* Sc */
236 { ARGSFL_NONE, NULL }, /* So */
237 { ARGSFL_DELIM, NULL }, /* Sq */
238 { ARGSFL_NONE, NULL }, /* Sm */
239 { ARGSFL_DELIM, NULL }, /* Sx */
240 { ARGSFL_DELIM, NULL }, /* Sy */
241 { ARGSFL_DELIM, NULL }, /* Tn */
242 { ARGSFL_DELIM, NULL }, /* Ux */
243 { ARGSFL_DELIM, NULL }, /* Xc */
244 { ARGSFL_NONE, NULL }, /* Xo */
245 { ARGSFL_NONE, NULL }, /* Fo */
246 { ARGSFL_NONE, NULL }, /* Fc */
247 { ARGSFL_NONE, NULL }, /* Oo */
248 { ARGSFL_DELIM, NULL }, /* Oc */
249 { ARGSFL_NONE, args_Bk }, /* Bk */
250 { ARGSFL_NONE, NULL }, /* Ek */
251 { ARGSFL_NONE, NULL }, /* Bt */
252 { ARGSFL_NONE, NULL }, /* Hf */
253 { ARGSFL_NONE, NULL }, /* Fr */
254 { ARGSFL_NONE, NULL }, /* Ud */
255 { ARGSFL_NONE, NULL }, /* Lb */
256 { ARGSFL_NONE, NULL }, /* Lp */
257 { ARGSFL_DELIM, NULL }, /* Lk */
258 { ARGSFL_DELIM, NULL }, /* Mt */
259 { ARGSFL_DELIM, NULL }, /* Brq */
260 { ARGSFL_NONE, NULL }, /* Bro */
261 { ARGSFL_DELIM, NULL }, /* Brc */
262 { ARGSFL_NONE, NULL }, /* %C */
263 { ARGSFL_NONE, NULL }, /* Es */
264 { ARGSFL_NONE, NULL }, /* En */
265 { ARGSFL_NONE, NULL }, /* Dx */
266 { ARGSFL_NONE, NULL }, /* %Q */
267 { ARGSFL_NONE, NULL }, /* br */
268 { ARGSFL_NONE, NULL }, /* sp */
269 { ARGSFL_NONE, NULL }, /* %U */
270 { ARGSFL_NONE, NULL }, /* Ta */
275 * Parse an argument from line text. This comes in the form of -key
276 * [value0...], which may either have a single mandatory value, at least
277 * one mandatory value, an optional single value, or no value.
280 mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
281 struct mdoc_arg **v, int *pos, char *buf)
284 struct mdoc_argv tmp;
285 struct mdoc_arg *arg;
286 const enum mdocargt *ap;
288 if ('\0' == buf[*pos])
290 else if (NULL == (ap = mdocargs[tok].argvs))
292 else if ('-' != buf[*pos])
295 /* Seek to the first unescaped space. */
301 for ( ; buf[*pos] ; (*pos)++)
302 if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
306 * We want to nil-terminate the word to look it up (it's easier
307 * that way). But we may not have a flag, in which case we need
308 * to restore the line as-is. So keep around the stray byte,
309 * which we'll reset upon exiting (if necessary).
312 if ('\0' != (sv = buf[*pos]))
313 buf[(*pos)++] = '\0';
316 * Now look up the word as a flag. Use temporary storage that
317 * we'll copy into the node's flags, if necessary.
320 memset(&tmp, 0, sizeof(struct mdoc_argv));
324 tmp.arg = MDOC_ARG_MAX;
326 while (MDOC_ARG_MAX != (tmp.arg = *ap++))
327 if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
330 if (MDOC_ARG_MAX == tmp.arg) {
332 * The flag was not found.
333 * Restore saved zeroed byte and return as a word.
340 /* Read to the next word (the argument). */
342 while (buf[*pos] && ' ' == buf[*pos])
345 switch (argvflags[tmp.arg]) {
347 if ( ! argv_single(m, line, &tmp, pos, buf))
351 if ( ! argv_multi(m, line, &tmp, pos, buf))
354 case (ARGV_OPT_SINGLE):
355 if ( ! argv_opt_single(m, line, &tmp, pos, buf))
362 if (NULL == (arg = *v))
363 arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
366 arg->argv = mandoc_realloc
367 (arg->argv, arg->argc * sizeof(struct mdoc_argv));
369 memcpy(&arg->argv[(int)arg->argc - 1],
370 &tmp, sizeof(struct mdoc_argv));
376 mdoc_argv_free(struct mdoc_arg *p)
390 for (i = (int)p->argc - 1; i >= 0; i--)
398 argn_free(struct mdoc_arg *p, int iarg)
400 struct mdoc_argv *arg;
403 arg = &p->argv[iarg];
405 if (arg->sz && arg->value) {
406 for (j = (int)arg->sz - 1; j >= 0; j--)
411 for (--p->argc; iarg < (int)p->argc; iarg++)
412 p->argv[iarg] = p->argv[iarg+1];
416 mdoc_zargs(struct mdoc *m, int line, int *pos, char *buf, char **v)
419 return(args(m, line, pos, buf, ARGSFL_NONE, v));
423 mdoc_args(struct mdoc *m, int line, int *pos,
424 char *buf, enum mdoct tok, char **v)
429 fl = mdocargs[tok].flags;
432 return(args(m, line, pos, buf, fl, v));
435 * We know that we're in an `It', so it's reasonable to expect
436 * us to be sitting in a `Bl'. Someday this may not be the case
437 * (if we allow random `It's sitting out there), so provide a
438 * safe fall-back into the default behaviour.
441 for (n = m->last; n; n = n->parent)
442 if (MDOC_Bl == n->tok)
443 if (LIST_column == n->norm->Bl.type) {
448 return(args(m, line, pos, buf, fl, v));
452 args(struct mdoc *m, int line, int *pos,
453 char *buf, enum argsflag fl, char **v)
458 if ('\0' == buf[*pos]) {
459 if (MDOC_PPHRASE & m->flags)
462 * If we're not in a partial phrase and the flag for
463 * being a phrase literal is still set, the punctuation
466 if (MDOC_PHRASELIT & m->flags)
467 mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
469 m->flags &= ~MDOC_PHRASELIT;
475 if (ARGSFL_DELIM == fl)
476 if (args_checkpunct(buf, *pos))
480 * First handle TABSEP items, restricted to `Bl -column'. This
481 * ignores conventional token parsing and instead uses tabs or
482 * `Ta' macros to separate phrases. Phrases are parsed again
483 * for arguments at a later phase.
486 if (ARGSFL_TABSEP == fl) {
487 /* Scan ahead to tab (can't be escaped). */
488 p = strchr(*v, '\t');
491 /* Scan ahead to unescaped `Ta'. */
492 if ( ! (MDOC_PHRASELIT & m->flags))
493 for (pp = *v; ; pp++) {
494 if (NULL == (pp = strstr(pp, "Ta")))
496 if (pp > *v && ' ' != *(pp - 1))
498 if (' ' == *(pp + 2) || '\0' == *(pp + 2))
502 /* By default, assume a phrase. */
506 * Adjust new-buffer position to be beyond delimiter
507 * mark (e.g., Ta -> end + 2).
510 *pos += pp < p ? 2 : 1;
511 rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
513 } else if (p && ! pp) {
516 } else if (pp && ! p) {
524 /* Whitespace check for eoln case... */
525 if ('\0' == *p && ' ' == *(p - 1))
526 mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
528 *pos += (int)(p - *v);
530 /* Strip delimiter's preceding whitespace. */
532 while (pp > *v && ' ' == *pp) {
533 if (pp > *v && '\\' == *(pp - 1))
539 /* Strip delimiter's proceeding whitespace. */
540 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
547 * Process a quoted literal. A quote begins with a double-quote
548 * and ends with a double-quote NOT preceded by a double-quote.
549 * Whitespace is NOT involved in literal termination.
552 if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
553 if ( ! (MDOC_PHRASELIT & m->flags))
556 if (MDOC_PPHRASE & m->flags)
557 m->flags |= MDOC_PHRASELIT;
559 for ( ; buf[*pos]; (*pos)++) {
560 if ('\"' != buf[*pos])
562 if ('\"' != buf[*pos + 1])
567 if ('\0' == buf[*pos]) {
568 if (MDOC_PPHRASE & m->flags)
570 mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
574 m->flags &= ~MDOC_PHRASELIT;
575 buf[(*pos)++] = '\0';
577 if ('\0' == buf[*pos])
580 while (' ' == buf[*pos])
583 if ('\0' == buf[*pos])
584 mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
590 *v = mandoc_getarg(m->parse, &p, line, pos);
596 * Check if the string consists only of space-separated closing
597 * delimiters. This is a bit of a dance: the first must be a close
598 * delimiter, but it may be followed by middle delimiters. Arbitrary
599 * whitespace may separate these tokens.
602 args_checkpunct(const char *buf, int i)
608 /* First token must be a close-delimiter. */
610 for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
617 if (DELIM_CLOSE != mdoc_isdelim(dbuf))
620 while (' ' == buf[i])
623 /* Remaining must NOT be open/none. */
627 while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
628 dbuf[j++] = buf[i++];
634 d = mdoc_isdelim(dbuf);
635 if (DELIM_NONE == d || DELIM_OPEN == d)
638 while (' ' == buf[i])
642 return('\0' == buf[i]);
646 argv_multi(struct mdoc *m, int line,
647 struct mdoc_argv *v, int *pos, char *buf)
652 for (v->sz = 0; ; v->sz++) {
653 if ('-' == buf[*pos])
655 ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
656 if (ARGS_ERROR == ac)
658 else if (ARGS_EOLN == ac)
661 if (0 == v->sz % MULTI_STEP)
662 v->value = mandoc_realloc(v->value,
663 (v->sz + MULTI_STEP) * sizeof(char *));
665 v->value[(int)v->sz] = mandoc_strdup(p);
672 argv_opt_single(struct mdoc *m, int line,
673 struct mdoc_argv *v, int *pos, char *buf)
678 if ('-' == buf[*pos])
681 ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
682 if (ARGS_ERROR == ac)
688 v->value = mandoc_malloc(sizeof(char *));
689 v->value[0] = mandoc_strdup(p);
695 argv_single(struct mdoc *m, int line,
696 struct mdoc_argv *v, int *pos, char *buf)
704 ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
705 if (ARGS_EOLN == ac) {
706 mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
708 } else if (ARGS_ERROR == ac)
712 v->value = mandoc_malloc(sizeof(char *));
713 v->value[0] = mandoc_strdup(p);