3 * $Id: mdwopt.c,v 1.5 1999/05/19 20:23:59 mdw Exp $
5 * Options parsing, similar to GNU @getopt_long@
7 * (c) 1996 Straylight/Edgeware
10 /*----- Licensing notice --------------------------------------------------*
12 * This file is part of many programs.
14 * `mdwopt' is free software; you can redistribute it and/or modify
15 * it under the terms of the GNU Library General Public License as
16 * published by the Free Software Foundation; either version 2 of the
17 * License, or (at your option) any later version.
19 * `mdwopt' is distributed in the hope that it will be useful,
20 * but WITHOUT ANY WARRANTY; without even the implied warranty of
21 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
22 * GNU Library General Public License for more details.
24 * You should have received a copy of the GNU Library General Public
25 * License along with `mdwopt'; if not, write to the Free
26 * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
30 /*----- Revision history --------------------------------------------------*
33 * Revision 1.5 1999/05/19 20:23:59 mdw
34 * Change naming to match newer mLib conventions.
36 * Revision 1.4 1999/05/15 10:25:38 mdw
37 * Fix copyright information.
39 * Revision 1.3 1999/05/14 18:51:42 mdw
40 * Reformat the LGPL notice slightly.
42 * Revision 1.2 1999/05/13 22:57:23 mdw
43 * Change `-ise' to `-ize' throughout.
45 * Revision 1.1.1.1 1999/05/05 19:23:47 mdw
46 * New import. The old CVS repository was lost in a disk disaster.
48 * --- Previous lives ---
51 * Revision 1.7 1997/09/11 09:19:11 mdw
52 * (mo__nextWord): Arrrgh. Don't free the environment variable buffer!
53 * People are still using it!
55 * Revision 1.6 1997/09/11 09:05:54 mdw
56 * (mo__nextWord): Fix bug which returns too many words from environment
59 * Revision 1.5 1997/08/09 20:27:59 mdw
60 * Fix spelling of `Licensing'.
62 * Revision 1.4 1997/07/29 21:11:35 mdw
63 * Reformatted. Fixed buffer overflow when dealing with environment
64 * variables. Included NT in list of daft operating systems with `\' as a
65 * path separator. Fixed address of the FSF.
67 * Revision 1.3 1997/02/26 00:41:10 mdw
68 * Added GPL notice to the top. Slight formatting changes.
70 * Revision 1.2 1996/10/28 13:12:13 mdw
71 * Fixed calls to ctype.h routines. Arguments are cast to unsigned char
72 * to avoid invoking undefined behaviour caused by signedness of chars.
74 * Revision 1.1 1996/09/24 18:01:28 mdw
79 /*----- External dependencies ---------------------------------------------*/
88 /*----- Configuration things ----------------------------------------------*/
92 #elif defined(__OS2__) || defined(__MSDOS__) || defined(__WINNT__)
94 #else /* Assume a sane filing system */
98 /*----- Global variables --------------------------------------------------*/
100 mdwopt_data mdwopt_global
= {0, 0, 0, 1, 0, 0, 0, 0, 0};
103 ORD_PERMUTE
= 0, /* Permute the options (default) */
104 ORD_RETURN
= 1, /* Return non-option things */
105 ORD_POSIX
= 2, /* Do POSIX-type hacking */
106 ORD_NEGATE
= 4 /* Magic negate-next-thing flag */
109 /*----- Main code ---------------------------------------------------------*/
111 /* --- @nextWord@ --- *
113 * Arguments: @int argc@ = number of command line options
114 * @char *argv[]@ = pointer to command line options
115 * @mdwopt_data *data@ = pointer to persistent state
117 * Returns: Pointer to the next word to handle, or 0
119 * Use: Extracts the next word from the command line or environment
123 static char *nextWord(int argc
, char *const *argv
, mdwopt_data
*data
)
125 if (data
->ind
== -1) {
128 while (isspace((unsigned char)*p
))
131 while (*p
&& !isspace((unsigned char)*p
))
142 if (data
->next
== argc
)
144 return (argv
[data
->next
++]);
147 /* --- @permute@ --- *
149 * Arguments: @char *argv[]@ = pointer to command line arguments
150 * @mdwopt_data *data@ = pointer to persistent data
154 * Use: Moves a command line option into the right place.
157 static void permute(char *const *argv
, mdwopt_data
*data
)
159 char **v
= (char **)argv
;
160 if (data
->ind
!= -1) {
161 int i
= data
->next
- 1;
163 while (i
> data
->ind
) {
172 /* --- @findOpt@ --- *
174 * Arguments: @int o@ = which option to search for
175 * @const char *shortopt@ = short options string to search
176 * @mdwopt_data *data@ = pointer to persistant state
178 * Returns: Pointer to rest of short options string (including magic
181 * Use: Looks up a short option in the given string.
184 static const char *findOpt(int o
, const char *shortopt
,
187 const char *p
= shortopt
; /* Point to short opts table */
189 if (!*p
) /* No more options left */
192 if (o
!= *p
|| (p
[1] != '+' && data
->order
& ORD_NEGATE
)) {
193 p
++; /* Skip this option entry */
194 while (*p
== '+') /* Jump a `%|+|%' sign */
196 while (*p
== ':') /* And jump any `%|:|%' characters */
197 p
++; /* Just in case there are any */
204 /* --- @mdwopt@ --- *
206 * Arguments: @int argc@ = number of command line arguments
207 * @char * const *argv@ = pointer to command line arguments
208 * @const char *shortopt@ = pointer to short options information
209 * @const struct option *longopts@ = pointer to long opts info
210 * @int *longind@ = where to store matched longopt
211 * @mdwopt_data *data@ = persistent state for the parser
212 * @int flags@ = various useful flags
214 * Returns: Value of option found next, or an error character, or
215 * @EOF@ for the last thing.
217 * Use: Reads options. The routine should be more-or-less compatible
218 * with standard getopts, although it provides many more
219 * features even than the standard GNU implementation.
221 * The precise manner of options parsing is determined by
222 * various flag settings, which are described below. By setting
223 * flag values appropriately, you can achieve behaviour very
224 * similar to most other getopt routines.
227 * How options parsing appears to users
229 * A command line consists of a number of `words' (which may
230 * contain spaces, according to various shell quoting
231 * conventions). A word may be an option, an argument to an
232 * option, or a non-option. An option begins with a special
233 * character, usually `%|-|%', although `%|+|%' is also used
234 * sometimes. As special exceptions, the word containing only a
235 * `%|-|%' is considered to be a non-option, since it usually
236 * represents standard input or output as a filename, and the
237 * word containing a double-dash `%|--|%' is used to mark all
238 * following words as being non-options regardless of their
241 * Traditionally, all words after the first non-option have been
242 * considered to be non-options automatically, so that options
243 * must be specified before filenames. However, this
244 * implementation can extract all the options from the command
245 * line regardless of their position. This can usually be
246 * disabled by setting one of the environment variables
247 * `%|POSIXLY_CORRECT|%' or `%|_POSIX_OPTION_ORDER|%'.
249 * There are two different styles of options: `short' and
252 * Short options are the sort which Unix has known for ages: an
253 * option is a single letter, preceded by a `%|-|%'. Short
254 * options can be joined together to save space (and possibly to
255 * make silly words): e.g., instead of giving options
256 * `%|-x -y|%', a user could write `%|-xy|%'. Some short
257 * options can have arguments, which appear after the option
258 * letter, either immediately following, or in the next `word'
259 * (so an option with an argument could be written as
260 * `%|-o foo|%' or as `%|-ofoo|%'). Note that options with
261 * optional arguments must be written in the second style.
263 * When a short option controls a flag setting, it is sometimes
264 * possible to explicitly turn the flag off, as well as turning
265 * it on, (usually to override default options). This is
266 * usually done by using a `%|+|%' instead of a `%|-|%' to
267 * introduce the option.
269 * Long options, as popularized by the GNU utilities, are given
270 * long-ish memorable names, preceded by a double-dash `%|--|%'.
271 * Since their names are more than a single character, long
272 * options can't be combined in the same way as short options.
273 * Arguments to long options may be given either in the same
274 * `word', separated from the option name by an equals sign, or
275 * in the following `word'.
277 * Long option names can be abbreviated if necessary, as long
278 * as the abbreviation is unique. This means that options can
279 * have sensible and memorable names but still not require much
280 * typing from an experienced user.
282 * Like short options, long options can control flag settings.
283 * The options to manipulate these settings come in pairs: an
284 * option of the form `%|--set-flag|%' might set the flag, while
285 * an option of the form `%|--no-set-flag|%' might clear it.
287 * It is usual for applications to provide both short and long
288 * options with identical behaviour. Some applications with
289 * lots of options may only provide long options (although they
290 * will often be only two or three characters long). In this
291 * case, long options can be preceded with a single `%|-|%'
292 * character, and negated by a `%|+|%' character.
294 * Finally, some (older) programs accept arguments of the form
295 * `%%@.{"-"<number>}%%', to set some numerical parameter,
296 * typically a line count of some kind.
299 * How programs parse options
301 * An application parses its options by calling mdwopt
302 * repeatedly. Each time it is called, mdwopt returns a value
303 * describing the option just read, and stores information about
304 * the option in a data block. The value %$-1$% is returned
305 * when there are no more options to be read. The `%|?|%'
306 * character is returned when an error is encountered.
308 * Before starting to parse options, the value @data->ind@ must
309 * be set to 0 or 1. The value of @data->err@ can also be set,
310 * to choose whether errors are reported by mdwopt.
312 * The program's `@argc@' and `@argv@' arguments are passed to
313 * the options parser, so that it can read the command line. A
314 * flags word is also passed, allowing the program fine control
315 * over parsing. The flags are described above.
317 * Short options are described by a string, which once upon a
318 * time just contained the permitted option characters. Now the
319 * options string begins with a collection of flag characters,
320 * and various flag characters can be put after options
321 * characters to change their properties.
323 * If the first character of the short options string is
324 * `%|+|%', `%|-|%' or `%|!|%', the order in which options are
325 * read is modified, as follows:
327 * `%|+|%' forces the POSIX order to be used. As soon as a non-
328 * option is found, mdwopt returns %$-1$%.
330 * `%|-|%' makes mdwopt treat non-options as being `special'
331 * sorts of option. When a non-option word is found, the
332 * value 0 is returned, and the actual text of the word
333 * is stored as being the option's argument.
335 * `%|!|%' forces the default order to be used. The entire
336 * command line is scanned for options, which are
337 * returned in order. However, during this process,
338 * the options are moved in the @argv@ array, so that
339 * they appear before the non- options.
341 * A `%|:|%' character may be placed after the ordering flag (or
342 * at the very beginning if no ordering flag is given) which
343 * indicates that the character `%|:|%', rather than `%|?|%',
344 * should be returned if a missing argument error is detected.
346 * Each option in the string can be followed by a `%|+|%' sign,
347 * indicating that it can be negated, a `%|:|%' sign indicating
348 * that it requires an argument, or a `%|::|%' string,
349 * indicating an optional argument. Both `%|+|%' and `%|:|%' or
350 * `%|::|%' may be given, although the `%|+|%' must come first.
352 * If an option is found, the option character is returned to
353 * the caller. A pointer to an argument is stored in
354 * @data->arg@, or @NULL@ is stored if there was no argument.
355 * If a negated option was found, the option character is
356 * returned ORred with @OPTF_NEGATED@ (bit 8 set).
358 * Long options are described in a table. Each entry in the
359 * table is of type @struct option@, and the table is terminated
360 * by an entry whose @name@ field is null. Each option has
361 * a flags word which, due to historical reasons, is called
362 * @has_arg@. This describes various properties of the option,
363 * such as what sort of argument it takes, and whether it can
366 * When mdwopt finds a long option, it looks the name up in the
367 * table. The index of the matching entry is stored in the
368 * @longind@ variable, passed to mdwopt (unless @longind@ is 0):
369 * a value of %$-1$% indicates that no long option was
370 * found. The behaviour is then dependent on the values in the
371 * table entry. If @flag@ is nonzero, it points to an integer
372 * to be modified by mdwopt. Usually the value in the @val@
373 * field is simply stored in the @flag@ variable. If the flag
374 * @OPTF_SWITCH@ is set, however, the value is combined with
375 * the existing value of the flags using a bitwise OR. If
376 * @OPTF_NEGATE@ is set, then the flag bit will be cleared if a
377 * matching negated long option is found. The value 0 is
380 * If @flag@ is zero, the value in @val@ is returned by mdwopt,
381 * possibly with bit 8 set if the option was negated.
383 * Arguments for long options are stored in @data->arg@, as
386 * Numeric options, if enabled, cause the value `%|#|%' to be
387 * returned, and the numeric value to be stored in @data->opt@.
389 * If the flag @OPTF_ENVVAR@ is set on entry, options will be
390 * extracted from an environment variable whose name is built by
391 * capitalising all the letters of the program's name. (This
392 * allows a user to have different default settings for a
393 * program, by calling it through different symbolic links.) */
395 int mdwopt(int argc
, char *const *argv
,
396 const char *shortopt
,
397 const struct option
*longopts
, int *longind
,
398 mdwopt_data
*data
, int flags
)
400 /* --- Local variables --- */
402 char *p
, *q
, *r
; /* Some useful things to have */
403 char *prefix
; /* Prefix from this option */
404 int i
; /* Always useful */
405 char noarg
= '?'; /* Standard missing-arg char */
407 /* --- Sort out our data --- */
409 if (!data
) /* If default data requested */
410 data
= &mdwopt_global
; /* Then use the global stuff */
412 /* --- See if this is the first time --- */
414 if (data
->ind
== 0 || (data
->ind
== 1 && ~flags
& OPTF_NOPROGNAME
)) {
416 /* --- Sort out default returning order --- */
418 if (getenv("_POSIX_OPTION_ORDER") || /* Examine environment for opts */
419 getenv("POSIXLY_CORRECT")) /* To see if we disable features */
420 data
->order
= ORD_POSIX
; /* If set, use POSIX ordering */
422 data
->order
= ORD_PERMUTE
; /* Otherwise mangle the options */
424 /* --- Now see what the caller actually wants --- */
426 switch (shortopt
[0]) { /* Look at the first character */
427 case '-': /* `%|-|%' turns on in-orderness */
428 data
->order
= ORD_RETURN
;
430 case '+': /* `%|+|%' turns on POSIXness */
431 data
->order
= ORD_POSIX
;
433 case '!': /* `%|!|%' ignores POSIXness */
434 data
->order
= ORD_PERMUTE
;
438 /* --- Now decide on the program's name --- */
440 if (~flags
& OPTF_NOPROGNAME
) {
441 p
= q
= (char *)argv
[0];
448 data
->ind
= data
->next
= 1;
451 /* --- See about environment variables --- *
453 * Be careful. The program may be setuid, and an attacker might have
454 * given us a long name in @argv[0]@. If the name is very long, don't
455 * support this option.
458 if (flags
& OPTF_ENVVAR
&& strlen(data
->prog
) < 48) {
462 /* --- For RISC OS, support a different format --- *
464 * Acorn's RISC OS tends to put settings in variables named
465 * `App$Options' rather than `APP'. Under RISC OS, I'll support
466 * both methods, just to avoid confuddlement.
470 sprintf(buf
, "%s$Options", data
->prog
);
475 p
= buf
; /* Point to a buffer */
476 q
= data
->prog
; /* Point to program name */
477 while (*q
) /* While characters left here */
478 *p
++ = toupper(*q
++); /* Copy and uppercase */
479 *p
++ = 0; /* Terminate my copy of this */
480 p
= getenv(buf
); /* Get the value of the variable */
486 /* --- Copy the options string into a buffer --- */
488 if (p
) { /* If it is defined */
489 q
= malloc(strlen(p
) + 1); /* Allocate space for a copy */
490 if (!q
) { /* If that failed */
491 fprintf(stderr
, /* Report a nice error */
492 "%s: Not enough memory to read settings in "
493 "environment variable\n",
495 } else { /* Otherwise */
496 strcpy(q
, p
); /* Copy the text over */
497 data
->ind
= -1; /* Mark that we're parsing envvar */
498 data
->env
= data
->estart
= q
; /* And store the pointer away */
505 data
->ind
= data
->next
= 0;
508 /* --- Do some initial bodgery --- *
510 * The @shortopt@ string can have some interesting characters at the
511 * beginning. We'll skip past them.
514 switch (shortopt
[0]) {
522 if (shortopt
[0] == ':') {
527 if (longind
) /* Allow longind to be null */
528 *longind
= -1; /* Clear this to avoid confusion */
529 data
->opt
= -1; /* And this too */
530 data
->arg
= 0; /* No option set up here */
532 /* --- Now go off and search for an option --- */
534 if (!data
->list
|| !*data
->list
) {
535 data
->order
&= 3; /* Clear negation flag */
537 /* --- Now we need to find the next option --- *
539 * Exactly how we do this depends on the settings of the order variable.
540 * We identify options as being things starting with `%|-|%', and which
541 * aren't equal to `%|-|%' or `%|--|%'. We'll look for options until:
543 * * We find something which isn't an option AND @order == ORD_POSIX@
544 * * We find a `%|--|%'
545 * * We reach the end of the list
547 * There are some added little wrinkles, which we'll meet as we go.
550 for (;;) { /* Keep looping for a while */
551 p
= nextWord(argc
, argv
, data
); /* Get the next word out */
552 if (!p
) /* If there's no next word */
553 return (EOF
); /* There's no more now */
555 /* --- See if we've found an option --- */
557 if ((p
[0] == '-' || (p
[0] == '+' && flags
& OPTF_NEGATION
)) &&
559 if (strcmp(p
, "--") == 0) { /* If this is the magic marker */
560 permute(argv
, data
); /* Stow the magic marker item */
561 return (EOF
); /* There's nothing else to do */
563 break; /* We've found something! */
566 /* --- Figure out how to proceed --- */
568 switch (data
->order
& 3) {
569 case ORD_POSIX
: /* POSIX option order */
570 return (EOF
); /* This is easy */
572 case ORD_PERMUTE
: /* Permute the option order */
574 case ORD_RETURN
: /* Return each argument */
575 permute(argv
, data
); /* Insert word in same place */
576 data
->arg
= p
; /* Point to the argument */
577 return (0); /* Return the value */
581 /* --- We found an option --- */
583 permute(argv
, data
); /* Do any permuting necessary */
585 /* --- Check for a numeric option --- *
587 * We only check the first character (or the second if the first is a
588 * sign). This ought to be enough.
591 if (flags
& OPTF_NUMBERS
&& (p
[0] == '-' || flags
& OPTF_NEGNUMBER
)) {
592 if (((p
[1] == '+' || p
[1] == '-') && isdigit((unsigned char)p
[2])) ||
593 isdigit((unsigned char)p
[1])) {
594 data
->opt
= strtol(p
+ 1, &data
->arg
, 10);
595 while (isspace((unsigned char)data
->arg
[0]))
599 return (p
[0] == '-' ?
'#' : '#' | OPTF_NEGATED
);
603 /* --- Check for a long option --- */
606 data
->order
|= ORD_NEGATE
;
608 if (((p
[0] == '-' && p
[1] == '-') ||
609 (flags
& OPTF_NOSHORTS
&& !findOpt(p
[1], shortopt
, data
))) &&
610 (~flags
& OPTF_NOLONGS
)) /* Is this a long option? */
612 int match
= -1; /* Count matches as we go */
614 if (p
[0] == '+') { /* If it's negated */
615 data
->order
|= ORD_NEGATE
; /* Set the negate flag */
616 p
++; /* Point to the main text */
617 prefix
= "+"; /* Set the prefix string up */
618 } else if (p
[1] == '-') { /* If this is a `%|--|%' option */
619 if ((flags
& OPTF_NEGATION
) && strncmp(p
+ 2, "no-", 3) == 0) {
620 p
+= 5; /* Point to main text */
621 prefix
= "--no-"; /* And set the prefix */
622 data
->order
|= ORD_NEGATE
; /* Set the negatedness flag */
624 p
+= 2; /* Point to the main text */
625 prefix
= "--"; /* Remember the prefix string */
628 if ((flags
& OPTF_NEGATION
) && strncmp(p
+ 1, "no-", 3) == 0) {
629 p
+= 4; /* Find the text */
630 prefix
= "-no-"; /* Set the prefix */
631 data
->order
|= ORD_NEGATE
; /* Set negatedness flag */
633 p
++; /* Otherwise find the text */
634 prefix
= "-"; /* And remember the prefix */
638 for (i
= 0; longopts
[i
].name
; i
++) { /* Loop through the options */
639 if ((data
->order
& ORD_NEGATE
) &&
640 (~longopts
[i
].has_arg
& OPTF_NEGATE
))
641 continue; /* If neg and opt doesn't allow */
643 r
= (char *) longopts
[i
].name
; /* Point to the name string */
644 q
= p
; /* Point to the string start */
645 for (;;) { /* Do a loop here */
646 if (*q
== 0 || *q
== '=') { /* End of the option string? */
647 if (*r
== 0) { /* If end of other string */
648 match
= i
; /* This is the match */
649 goto botched
; /* And exit the loop now */
651 if (match
== -1) { /* If no match currently */
652 match
= i
; /* Then this is it, here */
653 break; /* Stop looking now */
655 match
= -1; /* Else it's ambiguous */
656 goto botched
; /* So give up right now */
659 else if (*q
!= *r
) /* Otherwise if mismatch */
660 break; /* Abort this loop */
661 q
++, r
++; /* Increment the counters */
666 if (match
== -1) { /* If we couldn't find a match */
668 fprintf(stderr
, "%s: unrecognized option `%s%s'\n",
675 if (longind
) /* Allow longind to be null */
676 *longind
= match
; /* Store the match away */
678 /* --- Handle argument behaviour --- */
680 while (*p
!= 0 && *p
!= '=') /* Find the argument string */
682 p
= (*p ? p
+ 1 : 0); /* Sort out argument presence */
683 q
= (char *) longopts
[match
].name
; /* Remember the name here */
685 switch (longopts
[match
].has_arg
& 3) {
690 "%s: option `%s%s' does not accept arguments\n",
698 case required_argument
:
699 if (!p
) { /* If no argument given */
700 p
= nextWord(argc
, argv
, data
);
702 if (!p
) { /* If no more arguments */
704 fprintf(stderr
, "%s: option `%s%s' requires an argument\n",
715 case optional_argument
:
721 /* --- Do correct things now we have a match --- */
723 if (longopts
[match
].flag
) { /* If he has a @flag@ argument */
724 if (longopts
[match
].has_arg
& OPTF_SWITCH
) {
725 if (data
->order
& ORD_NEGATE
)
726 *longopts
[match
].flag
&= ~longopts
[match
].val
;
728 *longopts
[match
].flag
|= longopts
[match
].val
;
730 if (data
->order
& ORD_NEGATE
)
731 *longopts
[match
].flag
= 0;
733 *longopts
[match
].flag
= longopts
[match
].val
;
735 return (0); /* And return something */
737 if (data
->order
& ORD_NEGATE
)
738 return (longopts
[match
].val
| OPTF_NEGATED
);
740 return (longopts
[match
].val
);
744 /* --- Do short options things --- */
747 if (p
[0] == '+') /* If starts with a `%|+|%' */
748 data
->order
|= ORD_NEGATE
;
749 data
->list
= p
+ 1; /* Omit leading `%|-|%'/`%|+|%' */
753 /* --- Now process the short options --- */
755 i
= *data
->list
++; /* Get the next option letter */
756 data
->opt
= i
; /* Store this away nicely */
758 p
= (char *) findOpt(i
, shortopt
, data
);
759 if (!p
) { /* No more options left */
761 fprintf(stderr
, "%s: unknown option `%c%c'\n",
763 data
->order
& ORD_NEGATE ?
'+' : '-',
769 data
->opt
= i
; /* Store this for the caller */
771 /* --- Sort out an argument, if we expect one --- */
773 if (p
[0] == ':') { /* If we expect an option */
774 q
= (data
->list
[0] ? data
->list
: 0); /* If argument expected, use it */
775 data
->list
= 0; /* Kill the remaining options */
776 if (p
[1] != ':' && !q
) { /* If no arg, and not optional */
778 /* --- Same code as before --- */
780 q
= nextWord(argc
, argv
, data
); /* Read the next word */
781 if (!q
) { /* If no more arguments */
783 fprintf(stderr
, "%s: option `%c%c' requires an argument\n",
785 data
->order
& ORD_NEGATE ?
'+' : '-',
795 return ((data
->order
& ORD_NEGATE
) ? i
| OPTF_NEGATED
: i
);
798 /*----- That's all, folks -------------------------------------------------*/