]>
git.cameronkatri.com Git - mandoc.git/blob - mdoc_argv.c
1 /* $Id: mdoc_argv.c,v 1.83 2012/04/15 15:26:49 schwarze 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) */
50 const enum mdocargt
*argvs
;
53 static void argn_free(struct mdoc_arg
*, int);
54 static enum margserr
args(struct mdoc
*, int, int *,
55 char *, enum argsflag
, char **);
56 static int args_checkpunct(const char *, int);
57 static int argv_multi(struct mdoc
*, int,
58 struct mdoc_argv
*, int *, char *);
59 static int argv_single(struct mdoc
*, int,
60 struct mdoc_argv
*, int *, char *);
62 static const enum argvflag argvflags
[MDOC_ARG_MAX
] = {
63 ARGV_NONE
, /* MDOC_Split */
64 ARGV_NONE
, /* MDOC_Nosplit */
65 ARGV_NONE
, /* MDOC_Ragged */
66 ARGV_NONE
, /* MDOC_Unfilled */
67 ARGV_NONE
, /* MDOC_Literal */
68 ARGV_SINGLE
, /* MDOC_File */
69 ARGV_SINGLE
, /* MDOC_Offset */
70 ARGV_NONE
, /* MDOC_Bullet */
71 ARGV_NONE
, /* MDOC_Dash */
72 ARGV_NONE
, /* MDOC_Hyphen */
73 ARGV_NONE
, /* MDOC_Item */
74 ARGV_NONE
, /* MDOC_Enum */
75 ARGV_NONE
, /* MDOC_Tag */
76 ARGV_NONE
, /* MDOC_Diag */
77 ARGV_NONE
, /* MDOC_Hang */
78 ARGV_NONE
, /* MDOC_Ohang */
79 ARGV_NONE
, /* MDOC_Inset */
80 ARGV_MULTI
, /* MDOC_Column */
81 ARGV_SINGLE
, /* MDOC_Width */
82 ARGV_NONE
, /* MDOC_Compact */
83 ARGV_NONE
, /* MDOC_Std */
84 ARGV_NONE
, /* MDOC_Filled */
85 ARGV_NONE
, /* MDOC_Words */
86 ARGV_NONE
, /* MDOC_Emphasis */
87 ARGV_NONE
, /* MDOC_Symbolic */
88 ARGV_NONE
/* MDOC_Symbolic */
91 static const enum mdocargt args_Ex
[] = {
96 static const enum mdocargt args_An
[] = {
102 static const enum mdocargt args_Bd
[] = {
114 static const enum mdocargt args_Bf
[] = {
121 static const enum mdocargt args_Bk
[] = {
126 static const enum mdocargt args_Bl
[] = {
145 static const struct mdocarg mdocargs
[MDOC_MAX
] = {
146 { ARGSFL_NONE
, NULL
}, /* Ap */
147 { ARGSFL_NONE
, NULL
}, /* Dd */
148 { ARGSFL_NONE
, NULL
}, /* Dt */
149 { ARGSFL_NONE
, NULL
}, /* Os */
150 { ARGSFL_NONE
, NULL
}, /* Sh */
151 { ARGSFL_NONE
, NULL
}, /* Ss */
152 { ARGSFL_NONE
, NULL
}, /* Pp */
153 { ARGSFL_DELIM
, NULL
}, /* D1 */
154 { ARGSFL_DELIM
, NULL
}, /* Dl */
155 { ARGSFL_NONE
, args_Bd
}, /* Bd */
156 { ARGSFL_NONE
, NULL
}, /* Ed */
157 { ARGSFL_NONE
, args_Bl
}, /* Bl */
158 { ARGSFL_NONE
, NULL
}, /* El */
159 { ARGSFL_NONE
, NULL
}, /* It */
160 { ARGSFL_DELIM
, NULL
}, /* Ad */
161 { ARGSFL_DELIM
, args_An
}, /* An */
162 { ARGSFL_DELIM
, NULL
}, /* Ar */
163 { ARGSFL_NONE
, NULL
}, /* Cd */
164 { ARGSFL_DELIM
, NULL
}, /* Cm */
165 { ARGSFL_DELIM
, NULL
}, /* Dv */
166 { ARGSFL_DELIM
, NULL
}, /* Er */
167 { ARGSFL_DELIM
, NULL
}, /* Ev */
168 { ARGSFL_NONE
, args_Ex
}, /* Ex */
169 { ARGSFL_DELIM
, NULL
}, /* Fa */
170 { ARGSFL_NONE
, NULL
}, /* Fd */
171 { ARGSFL_DELIM
, NULL
}, /* Fl */
172 { ARGSFL_DELIM
, NULL
}, /* Fn */
173 { ARGSFL_DELIM
, NULL
}, /* Ft */
174 { ARGSFL_DELIM
, NULL
}, /* Ic */
175 { ARGSFL_NONE
, NULL
}, /* In */
176 { ARGSFL_DELIM
, NULL
}, /* Li */
177 { ARGSFL_NONE
, NULL
}, /* Nd */
178 { ARGSFL_DELIM
, NULL
}, /* Nm */
179 { ARGSFL_DELIM
, NULL
}, /* Op */
180 { ARGSFL_NONE
, NULL
}, /* Ot */
181 { ARGSFL_DELIM
, NULL
}, /* Pa */
182 { ARGSFL_NONE
, args_Ex
}, /* Rv */
183 { ARGSFL_DELIM
, NULL
}, /* St */
184 { ARGSFL_DELIM
, NULL
}, /* Va */
185 { ARGSFL_DELIM
, NULL
}, /* Vt */
186 { ARGSFL_DELIM
, NULL
}, /* Xr */
187 { ARGSFL_NONE
, NULL
}, /* %A */
188 { ARGSFL_NONE
, NULL
}, /* %B */
189 { ARGSFL_NONE
, NULL
}, /* %D */
190 { ARGSFL_NONE
, NULL
}, /* %I */
191 { ARGSFL_NONE
, NULL
}, /* %J */
192 { ARGSFL_NONE
, NULL
}, /* %N */
193 { ARGSFL_NONE
, NULL
}, /* %O */
194 { ARGSFL_NONE
, NULL
}, /* %P */
195 { ARGSFL_NONE
, NULL
}, /* %R */
196 { ARGSFL_NONE
, NULL
}, /* %T */
197 { ARGSFL_NONE
, NULL
}, /* %V */
198 { ARGSFL_DELIM
, NULL
}, /* Ac */
199 { ARGSFL_NONE
, NULL
}, /* Ao */
200 { ARGSFL_DELIM
, NULL
}, /* Aq */
201 { ARGSFL_DELIM
, NULL
}, /* At */
202 { ARGSFL_DELIM
, NULL
}, /* Bc */
203 { ARGSFL_NONE
, args_Bf
}, /* Bf */
204 { ARGSFL_NONE
, NULL
}, /* Bo */
205 { ARGSFL_DELIM
, NULL
}, /* Bq */
206 { ARGSFL_DELIM
, NULL
}, /* Bsx */
207 { ARGSFL_DELIM
, NULL
}, /* Bx */
208 { ARGSFL_NONE
, NULL
}, /* Db */
209 { ARGSFL_DELIM
, NULL
}, /* Dc */
210 { ARGSFL_NONE
, NULL
}, /* Do */
211 { ARGSFL_DELIM
, NULL
}, /* Dq */
212 { ARGSFL_DELIM
, NULL
}, /* Ec */
213 { ARGSFL_NONE
, NULL
}, /* Ef */
214 { ARGSFL_DELIM
, NULL
}, /* Em */
215 { ARGSFL_NONE
, NULL
}, /* Eo */
216 { ARGSFL_DELIM
, NULL
}, /* Fx */
217 { ARGSFL_DELIM
, NULL
}, /* Ms */
218 { ARGSFL_DELIM
, NULL
}, /* No */
219 { ARGSFL_DELIM
, NULL
}, /* Ns */
220 { ARGSFL_DELIM
, NULL
}, /* Nx */
221 { ARGSFL_DELIM
, NULL
}, /* Ox */
222 { ARGSFL_DELIM
, NULL
}, /* Pc */
223 { ARGSFL_DELIM
, NULL
}, /* Pf */
224 { ARGSFL_NONE
, NULL
}, /* Po */
225 { ARGSFL_DELIM
, NULL
}, /* Pq */
226 { ARGSFL_DELIM
, NULL
}, /* Qc */
227 { ARGSFL_DELIM
, NULL
}, /* Ql */
228 { ARGSFL_NONE
, NULL
}, /* Qo */
229 { ARGSFL_DELIM
, NULL
}, /* Qq */
230 { ARGSFL_NONE
, NULL
}, /* Re */
231 { ARGSFL_NONE
, NULL
}, /* Rs */
232 { ARGSFL_DELIM
, NULL
}, /* Sc */
233 { ARGSFL_NONE
, NULL
}, /* So */
234 { ARGSFL_DELIM
, NULL
}, /* Sq */
235 { ARGSFL_NONE
, NULL
}, /* Sm */
236 { ARGSFL_DELIM
, NULL
}, /* Sx */
237 { ARGSFL_DELIM
, NULL
}, /* Sy */
238 { ARGSFL_DELIM
, NULL
}, /* Tn */
239 { ARGSFL_DELIM
, NULL
}, /* Ux */
240 { ARGSFL_DELIM
, NULL
}, /* Xc */
241 { ARGSFL_NONE
, NULL
}, /* Xo */
242 { ARGSFL_NONE
, NULL
}, /* Fo */
243 { ARGSFL_NONE
, NULL
}, /* Fc */
244 { ARGSFL_NONE
, NULL
}, /* Oo */
245 { ARGSFL_DELIM
, NULL
}, /* Oc */
246 { ARGSFL_NONE
, args_Bk
}, /* Bk */
247 { ARGSFL_NONE
, NULL
}, /* Ek */
248 { ARGSFL_NONE
, NULL
}, /* Bt */
249 { ARGSFL_NONE
, NULL
}, /* Hf */
250 { ARGSFL_NONE
, NULL
}, /* Fr */
251 { ARGSFL_NONE
, NULL
}, /* Ud */
252 { ARGSFL_NONE
, NULL
}, /* Lb */
253 { ARGSFL_NONE
, NULL
}, /* Lp */
254 { ARGSFL_DELIM
, NULL
}, /* Lk */
255 { ARGSFL_DELIM
, NULL
}, /* Mt */
256 { ARGSFL_DELIM
, NULL
}, /* Brq */
257 { ARGSFL_NONE
, NULL
}, /* Bro */
258 { ARGSFL_DELIM
, NULL
}, /* Brc */
259 { ARGSFL_NONE
, NULL
}, /* %C */
260 { ARGSFL_NONE
, NULL
}, /* Es */
261 { ARGSFL_NONE
, NULL
}, /* En */
262 { ARGSFL_NONE
, NULL
}, /* Dx */
263 { ARGSFL_NONE
, NULL
}, /* %Q */
264 { ARGSFL_NONE
, NULL
}, /* br */
265 { ARGSFL_NONE
, NULL
}, /* sp */
266 { ARGSFL_NONE
, NULL
}, /* %U */
267 { ARGSFL_NONE
, NULL
}, /* Ta */
272 * Parse an argument from line text. This comes in the form of -key
273 * [value0...], which may either have a single mandatory value, at least
274 * one mandatory value, an optional single value, or no value.
277 mdoc_argv(struct mdoc
*m
, int line
, enum mdoct tok
,
278 struct mdoc_arg
**v
, int *pos
, char *buf
)
281 struct mdoc_argv tmp
;
282 struct mdoc_arg
*arg
;
283 const enum mdocargt
*ap
;
285 if ('\0' == buf
[*pos
])
287 else if (NULL
== (ap
= mdocargs
[tok
].argvs
))
289 else if ('-' != buf
[*pos
])
292 /* Seek to the first unescaped space. */
298 for ( ; buf
[*pos
] ; (*pos
)++)
299 if (' ' == buf
[*pos
] && '\\' != buf
[*pos
- 1])
303 * We want to nil-terminate the word to look it up (it's easier
304 * that way). But we may not have a flag, in which case we need
305 * to restore the line as-is. So keep around the stray byte,
306 * which we'll reset upon exiting (if necessary).
309 if ('\0' != (sv
= buf
[*pos
]))
310 buf
[(*pos
)++] = '\0';
313 * Now look up the word as a flag. Use temporary storage that
314 * we'll copy into the node's flags, if necessary.
317 memset(&tmp
, 0, sizeof(struct mdoc_argv
));
321 tmp
.arg
= MDOC_ARG_MAX
;
323 while (MDOC_ARG_MAX
!= (tmp
.arg
= *ap
++))
324 if (0 == strcmp(p
, mdoc_argnames
[tmp
.arg
]))
327 if (MDOC_ARG_MAX
== tmp
.arg
) {
329 * The flag was not found.
330 * Restore saved zeroed byte and return as a word.
337 /* Read to the next word (the argument). */
339 while (buf
[*pos
] && ' ' == buf
[*pos
])
342 switch (argvflags
[tmp
.arg
]) {
344 if ( ! argv_single(m
, line
, &tmp
, pos
, buf
))
348 if ( ! argv_multi(m
, line
, &tmp
, pos
, buf
))
355 if (NULL
== (arg
= *v
))
356 arg
= *v
= mandoc_calloc(1, sizeof(struct mdoc_arg
));
359 arg
->argv
= mandoc_realloc
360 (arg
->argv
, arg
->argc
* sizeof(struct mdoc_argv
));
362 memcpy(&arg
->argv
[(int)arg
->argc
- 1],
363 &tmp
, sizeof(struct mdoc_argv
));
369 mdoc_argv_free(struct mdoc_arg
*p
)
383 for (i
= (int)p
->argc
- 1; i
>= 0; i
--)
391 argn_free(struct mdoc_arg
*p
, int iarg
)
393 struct mdoc_argv
*arg
;
396 arg
= &p
->argv
[iarg
];
398 if (arg
->sz
&& arg
->value
) {
399 for (j
= (int)arg
->sz
- 1; j
>= 0; j
--)
404 for (--p
->argc
; iarg
< (int)p
->argc
; iarg
++)
405 p
->argv
[iarg
] = p
->argv
[iarg
+1];
409 mdoc_zargs(struct mdoc
*m
, int line
, int *pos
, char *buf
, char **v
)
412 return(args(m
, line
, pos
, buf
, ARGSFL_NONE
, v
));
416 mdoc_args(struct mdoc
*m
, int line
, int *pos
,
417 char *buf
, enum mdoct tok
, char **v
)
422 fl
= mdocargs
[tok
].flags
;
425 return(args(m
, line
, pos
, buf
, fl
, v
));
428 * We know that we're in an `It', so it's reasonable to expect
429 * us to be sitting in a `Bl'. Someday this may not be the case
430 * (if we allow random `It's sitting out there), so provide a
431 * safe fall-back into the default behaviour.
434 for (n
= m
->last
; n
; n
= n
->parent
)
435 if (MDOC_Bl
== n
->tok
)
436 if (LIST_column
== n
->norm
->Bl
.type
) {
441 return(args(m
, line
, pos
, buf
, fl
, v
));
445 args(struct mdoc
*m
, int line
, int *pos
,
446 char *buf
, enum argsflag fl
, char **v
)
451 if ('\0' == buf
[*pos
]) {
452 if (MDOC_PPHRASE
& m
->flags
)
455 * If we're not in a partial phrase and the flag for
456 * being a phrase literal is still set, the punctuation
459 if (MDOC_PHRASELIT
& m
->flags
)
460 mdoc_pmsg(m
, line
, *pos
, MANDOCERR_BADQUOTE
);
462 m
->flags
&= ~MDOC_PHRASELIT
;
468 if (ARGSFL_DELIM
== fl
)
469 if (args_checkpunct(buf
, *pos
))
473 * First handle TABSEP items, restricted to `Bl -column'. This
474 * ignores conventional token parsing and instead uses tabs or
475 * `Ta' macros to separate phrases. Phrases are parsed again
476 * for arguments at a later phase.
479 if (ARGSFL_TABSEP
== fl
) {
480 /* Scan ahead to tab (can't be escaped). */
481 p
= strchr(*v
, '\t');
484 /* Scan ahead to unescaped `Ta'. */
485 if ( ! (MDOC_PHRASELIT
& m
->flags
))
486 for (pp
= *v
; ; pp
++) {
487 if (NULL
== (pp
= strstr(pp
, "Ta")))
489 if (pp
> *v
&& ' ' != *(pp
- 1))
491 if (' ' == *(pp
+ 2) || '\0' == *(pp
+ 2))
495 /* By default, assume a phrase. */
499 * Adjust new-buffer position to be beyond delimiter
500 * mark (e.g., Ta -> end + 2).
503 *pos
+= pp
< p
? 2 : 1;
504 rc
= pp
< p
? ARGS_PHRASE
: ARGS_PPHRASE
;
506 } else if (p
&& ! pp
) {
509 } else if (pp
&& ! p
) {
517 /* Whitespace check for eoln case... */
518 if ('\0' == *p
&& ' ' == *(p
- 1))
519 mdoc_pmsg(m
, line
, *pos
, MANDOCERR_EOLNSPACE
);
521 *pos
+= (int)(p
- *v
);
523 /* Strip delimiter's preceding whitespace. */
525 while (pp
> *v
&& ' ' == *pp
) {
526 if (pp
> *v
&& '\\' == *(pp
- 1))
532 /* Strip delimiter's proceeding whitespace. */
533 for (pp
= &buf
[*pos
]; ' ' == *pp
; pp
++, (*pos
)++)
540 * Process a quoted literal. A quote begins with a double-quote
541 * and ends with a double-quote NOT preceded by a double-quote.
542 * Whitespace is NOT involved in literal termination.
545 if (MDOC_PHRASELIT
& m
->flags
|| '\"' == buf
[*pos
]) {
546 if ( ! (MDOC_PHRASELIT
& m
->flags
))
549 if (MDOC_PPHRASE
& m
->flags
)
550 m
->flags
|= MDOC_PHRASELIT
;
552 for ( ; buf
[*pos
]; (*pos
)++) {
553 if ('\"' != buf
[*pos
])
555 if ('\"' != buf
[*pos
+ 1])
560 if ('\0' == buf
[*pos
]) {
561 if (MDOC_PPHRASE
& m
->flags
)
563 mdoc_pmsg(m
, line
, *pos
, MANDOCERR_BADQUOTE
);
567 m
->flags
&= ~MDOC_PHRASELIT
;
568 buf
[(*pos
)++] = '\0';
570 if ('\0' == buf
[*pos
])
573 while (' ' == buf
[*pos
])
576 if ('\0' == buf
[*pos
])
577 mdoc_pmsg(m
, line
, *pos
, MANDOCERR_EOLNSPACE
);
583 *v
= mandoc_getarg(m
->parse
, &p
, line
, pos
);
589 * Check if the string consists only of space-separated closing
590 * delimiters. This is a bit of a dance: the first must be a close
591 * delimiter, but it may be followed by middle delimiters. Arbitrary
592 * whitespace may separate these tokens.
595 args_checkpunct(const char *buf
, int i
)
601 /* First token must be a close-delimiter. */
603 for (j
= 0; buf
[i
] && ' ' != buf
[i
] && j
< DELIMSZ
; j
++, i
++)
610 if (DELIM_CLOSE
!= mdoc_isdelim(dbuf
))
613 while (' ' == buf
[i
])
616 /* Remaining must NOT be open/none. */
620 while (buf
[i
] && ' ' != buf
[i
] && j
< DELIMSZ
)
621 dbuf
[j
++] = buf
[i
++];
627 d
= mdoc_isdelim(dbuf
);
628 if (DELIM_NONE
== d
|| DELIM_OPEN
== d
)
631 while (' ' == buf
[i
])
635 return('\0' == buf
[i
]);
639 argv_multi(struct mdoc
*m
, int line
,
640 struct mdoc_argv
*v
, int *pos
, char *buf
)
645 for (v
->sz
= 0; ; v
->sz
++) {
646 if ('-' == buf
[*pos
])
648 ac
= args(m
, line
, pos
, buf
, ARGSFL_NONE
, &p
);
649 if (ARGS_ERROR
== ac
)
651 else if (ARGS_EOLN
== ac
)
654 if (0 == v
->sz
% MULTI_STEP
)
655 v
->value
= mandoc_realloc(v
->value
,
656 (v
->sz
+ MULTI_STEP
) * sizeof(char *));
658 v
->value
[(int)v
->sz
] = mandoc_strdup(p
);
665 argv_single(struct mdoc
*m
, int line
,
666 struct mdoc_argv
*v
, int *pos
, char *buf
)
671 ac
= args(m
, line
, pos
, buf
, ARGSFL_NONE
, &p
);
672 if (ARGS_ERROR
== ac
)
678 v
->value
= mandoc_malloc(sizeof(char *));
679 v
->value
[0] = mandoc_strdup(p
);