3 * $Id: mdwopt.c,v 1.1 1997/07/21 13:47:47 mdw Exp $
5 * Options parsing, similar to GNU @getopt_long@
7 * (c) 1996 Mark Wooding
10 /*----- Licencing 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 General Public License as published by
16 * the Free Software Foundation; either version 2 of the License, or
17 * (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 General Public License for more details.
24 * You should have received a copy of the GNU General Public License
25 * along with `mdwopt'; if not, write to the Free Software
26 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
29 /*----- Revision history --------------------------------------------------*
32 * Revision 1.1 1997/07/21 13:47:47 mdw
35 * Revision 1.3 1997/02/26 00:41:10 mdw
36 * Added GPL notice to the top. Slight formatting changes.
38 * Revision 1.2 1996/10/28 13:12:13 mdw
39 * Fixed calls to ctype.h routines. Arguments are cast to unsigned char
40 * to avoid invoking undefined behaviour caused by signedness of chars.
42 * Revision 1.1 1996/09/24 18:01:28 mdw
47 /*----- External dependencies ---------------------------------------------*/
56 /*----- Configuration things ----------------------------------------------*/
60 #elif defined(__OS2__) || defined(__MSDOS__)
62 #else /* Assume a sane filing system */
66 /*----- Global variables --------------------------------------------------*/
68 mdwopt_data mdwopt_global
= {0, 0, 0, 1, 0, 0, 0, 0, 0};
71 ord__permute
= 0, /* Permute the options (default) */
72 ord__return
= 1, /* Return non-option things */
73 ord__posix
= 2, /* Do POSIX-type hacking */
74 ord__negate
= 4 /* Magic negate-next-thing flag */
77 /*----- Main code ---------------------------------------------------------*/
79 /* --- @mo__nextWord@ --- *
81 * Arguments: @int argc@ = number of command line options
82 * @char *argv[]@ = pointer to command line options
83 * @mdwopt_data *data@ = pointer to persistent state
85 * Returns: Pointer to the next word to handle, or 0
87 * Use: Extracts the next word from the command line or environment
91 static char *mo__nextWord(int argc
, char *const *argv
, mdwopt_data
*data
)
93 if (data
->ind
== -1) {
96 while (isspace((unsigned char)*p
))
99 while (*p
&& !isspace((unsigned char)*p
))
110 if (data
->next
== argc
)
112 return (argv
[data
->next
++]);
115 /* --- @mo__permute@ --- *
117 * Arguments: @char *argv[]@ = pointer to command line arguments
118 * @mdwopt_data *data@ = pointer to persistent data
122 * Use: Moves a command line option into the right place.
125 static void mo__permute(char *const *argv
, mdwopt_data
*data
)
127 char **v
= (char **)argv
;
128 if (data
->ind
!= -1) {
129 int i
= data
->next
- 1;
131 while (i
> data
->ind
) {
140 /* --- @mo__findOpt@ --- *
142 * Arguments: @int o@ = which option to search for
143 * @const char *shortopt@ = short options string to search
144 * @mdwopt_data *data@ = pointer to persistant state
146 * Returns: Pointer to rest of short options string (including magic
149 * Use: Looks up a short option in the given string.
152 static const char *mo__findOpt(int o
, const char *shortopt
,
155 const char *p
= shortopt
; /* Point to short opts table */
157 if (!*p
) /* No more options left */
160 if (o
!= *p
|| (p
[1] != '+' && data
->order
& ord__negate
)) {
161 p
++; /* Skip this option entry */
162 while (*p
== '+') /* Jump a `%|+|%' sign */
164 while (*p
== ':') /* And jump any `%|:|%' characters */
165 p
++; /* Just in case there are any */
172 /* --- @mdwopt@ --- *
174 * Arguments: @int argc@ = number of command line arguments
175 * @char * const *argv@ = pointer to command line arguments
176 * @const char *shortopt@ = pointer to short options information
177 * @const struct option *longopts@ = pointer to long opts info
178 * @int *longind@ = where to store matched longopt
179 * @mdwopt_data *data@ = persistent state for the parser
180 * @int flags@ = various useful flags
182 * Returns: Value of option found next, or an error character, or
183 * @EOF@ for the last thing.
185 * Use: Reads options. The routine should be more-or-less compatible
186 * with standard getopts, although it provides many more
187 * features even than the standard GNU implementation.
189 * The precise manner of options parsing is determined by
190 * various flag settings, which are described below. By setting
191 * flag values appropriately, you can achieve behaviour very
192 * similar to most other getopt routines.
195 * How options parsing appears to users
197 * A command line consists of a number of `words' (which may
198 * contain spaces, according to various shell quoting
199 * conventions). A word may be an option, an argument to an
200 * option, or a non-option. An option begins with a special
201 * character, usually `%|-|%', although `%|+|%' is also used
202 * sometimes. As special exceptions, the word containing only a
203 * `%|-|%' is considered to be a non-option, since it usually
204 * represents standard input or output as a filename, and the
205 * word containing a double-dash `%|--|%' is used to mark all
206 * following words as being non-options regardless of their
209 * Traditionally, all words after the first non-option have been
210 * considered to be non-options automatically, so that options
211 * must be specified before filenames. However, this
212 * implementation can extract all the options from the command
213 * line regardless of their position. This can usually be
214 * disabled by setting one of the environment variables
215 * `%|POSIXLY_CORRECT|%' or `%|_POSIX_OPTION_ORDER|%'.
217 * There are two different styles of options: `short' and
220 * Short options are the sort which Unix has known for ages: an
221 * option is a single letter, preceded by a `%|-|%'. Short
222 * options can be joined together to save space (and possibly to
223 * make silly words): e.g., instead of giving options
224 * `%|-x -y|%', a user could write `%|-xy|%'. Some short
225 * options can have arguments, which appear after the option
226 * letter, either immediately following, or in the next `word'
227 * (so an option with an argument could be written as
228 * `%|-o foo|%' or as `%|-ofoo|%'). Note that options with
229 * optional arguments must be written in the second style.
231 * When a short option controls a flag setting, it is sometimes
232 * possible to explicitly turn the flag off, as well as turning
233 * it on, (usually to override default options). This is
234 * usually done by using a `%|+|%' instead of a `%|-|%' to
235 * introduce the option.
237 * Long options, as popularised by the GNU utilities, are given
238 * long-ish memorable names, preceded by a double-dash `%|--|%'.
239 * Since their names are more than a single character, long
240 * options can't be combined in the same way as short options.
241 * Arguments to long options may be given either in the same
242 * `word', separated from the option name by an equals sign, or
243 * in the following `word'.
245 * Long option names can be abbreviated if necessary, as long
246 * as the abbreviation is unique. This means that options can
247 * have sensible and memorable names but still not require much
248 * typing from an experienced user.
250 * Like short options, long options can control flag settings.
251 * The options to manipulate these settings come in pairs: an
252 * option of the form `%|--set-flag|%' might set the flag, while
253 * an option of the form `%|--no-set-flag|%' might clear it.
255 * It is usual for applications to provide both short and long
256 * options with identical behaviour. Some applications with
257 * lots of options may only provide long options (although they
258 * will often be only two or three characters long). In this
259 * case, long options can be preceded with a single `%|-|%'
260 * character, and negated by a `%|+|%' character.
262 * Finally, some (older) programs accept arguments of the form
263 * `%%@.{"-"<number>}%%', to set some numerical parameter,
264 * typically a line count of some kind.
267 * How programs parse options
269 * An application parses its options by calling mdwopt
270 * repeatedly. Each time it is called, mdwopt returns a value
271 * describing the option just read, and stores information about
272 * the option in a data block. The value %$-1$% is returned
273 * when there are no more options to be read. The `%|?|%'
274 * character is returned when an error is encountered.
276 * Before starting to parse options, the value @data->ind@ must
277 * be set to 0 or 1. The value of @data->err@ can also be set,
278 * to choose whether errors are reported by mdwopt.
280 * The program's `@argc@' and `@argv@' arguments are passed to
281 * the options parser, so that it can read the command line. A
282 * flags word is also passed, allowing the program fine control
283 * over parsing. The flags are described above.
285 * Short options are described by a string, which once upon a
286 * time just contained the permitted option characters. Now the
287 * options string begins with a collection of flag characters,
288 * and various flag characters can be put after options
289 * characters to change their properties.
291 * If the first character of the short options string is
292 * `%|+|%', `%|-|%' or `%|!|%', the order in which options are
293 * read is modified, as follows:
295 * `%|+|%' forces the POSIX order to be used. As soon as a non-
296 * option is found, mdwopt returns %$-1$%.
298 * `%|-|%' makes mdwopt treat non-options as being `special'
299 * sorts of option. When a non-option word is found, the
300 * value 0 is returned, and the actual text of the word
301 * is stored as being the option's argument.
303 * `%|!|%' forces the default order to be used. The entire
304 * command line is scanned for options, which are
305 * returned in order. However, during this process,
306 * the options are moved in the @argv@ array, so that
307 * they appear before the non- options.
309 * A `%|:|%' character may be placed after the ordering flag (or
310 * at the very beginning if no ordering flag is given) which
311 * indicates that the character `%|:|%', rather than `%|?|%',
312 * should be returned if a missing argument error is detected.
314 * Each option in the string can be followed by a `%|+|%' sign,
315 * indicating that it can be negated, a `%|:|%' sign indicating
316 * that it requires an argument, or a `%|::|%' string,
317 * indicating an optional argument. Both `%|+|%' and `%|:|%' or
318 * `%|::|%' may be given, although the `%|+|%' must come first.
320 * If an option is found, the option character is returned to
321 * the caller. A pointer to an argument is stored in
322 * @data->arg@, or @NULL@ is stored if there was no argument.
323 * If a negated option was found, the option character is
324 * returned ORred with @gFlag_negated@ (bit 8 set).
326 * Long options are described in a table. Each entry in the
327 * table is of type @struct option@, and the table is terminated
328 * by an entry whose @name@ field is null. Each option has
329 * a flags word which, due to historical reasons, is called
330 * @has_arg@. This describes various properties of the option,
331 * such as what sort of argument it takes, and whether it can
334 * When mdwopt finds a long option, it looks the name up in the
335 * table. The index of the matching entry is stored in the
336 * @longind@ variable, passed to mdwopt (unless @longind@ is 0):
337 * a value of %$-1$% indicates that no long option was
338 * found. The behaviour is then dependent on the values in the
339 * table entry. If @flag@ is nonzero, it points to an integer
340 * to be modified by mdwopt. Usually the value in the @val@
341 * field is simply stored in the @flag@ variable. If the flag
342 * @gFlag_switch@ is set, however, the value is combined with
343 * the existing value of the flags using a bitwise OR. If
344 * @gFlag_negate@ is set, then the flag bit will be cleared if a
345 * matching negated long option is found. The value 0 is
348 * If @flag@ is zero, the value in @val@ is returned by mdwopt,
349 * possibly with bit 8 set if the option was negated.
351 * Arguments for long options are stored in @data->arg@, as
354 * Numeric options, if enabled, cause the value `%|#|%' to be
355 * returned, and the numeric value to be stored in @data->opt@.
357 * If the flag @gFlag_envVar@ is set on entry, options will be
358 * extracted from an environment variable whose name is built by
359 * capitalising all the letters of the program's name. (This
360 * allows a user to have different default settings for a
361 * program, by calling it through different symbolic links.) */
363 int mdwopt(int argc
, char *const *argv
,
364 const char *shortopt
,
365 const struct option
*longopts
, int *longind
,
366 mdwopt_data
*data
, int flags
)
368 /* --- Local variables --- */
370 char *p
, *q
, *r
; /* Some useful things to have */
371 char *prefix
; /* Prefix from this option */
372 int i
; /* Always useful */
373 char noarg
= '?'; /* Standard missing-arg char */
375 /* --- Sort out our data --- */
377 if (!data
) /* If default data requested */
378 data
= &mdwopt_global
; /* Then use the global stuff */
380 /* --- See if this is the first time --- */
382 if (data
->ind
== 0 || (data
->ind
== 1 && ~flags
& gFlag_noProgName
)) {
384 /* --- Sort out default returning order --- */
386 if (getenv("_POSIX_OPTION_ORDER") || /* Examine environment for opts */
387 getenv("POSIXLY_CORRECT")) /* To see if we disable features */
388 data
->order
= ord__posix
; /* If set, use POSIX ordering */
390 data
->order
= ord__permute
; /* Otherwise mangle the options */
392 /* --- Now see what the caller actually wants --- */
394 switch (shortopt
[0]) { /* Look at the first character */
395 case '-': /* `%|-|%' turns on in-orderness */
396 data
->order
= ord__return
;
398 case '+': /* `%|+|%' turns on POSIXness */
399 data
->order
= ord__posix
;
401 case '!': /* `%|!|%' ignores POSIXness */
402 data
->order
= ord__permute
;
406 /* --- Now decide on the program's name --- */
408 if (~flags
& gFlag_noProgName
) {
409 p
= q
= (char *)argv
[0];
416 data
->ind
= data
->next
= 1;
419 /* --- See about environment variables --- *
421 * Be careful. The program may be setuid, and an attacker might have
422 * given us a long name in @argv[0]@. If the name is very long, don't
423 * support this option.
426 if (flags
& gFlag_envVar
&& strlen(data
->prog
) < 48) {
430 /* --- For RISC OS, support a different format --- *
432 * Acorn's RISC OS tends to put settings in variables named
433 * `App$Options' rather than `APP'. Under RISC OS, I'll support
434 * both methods, just to avoid confuddlement.
438 sprintf(buf
, "%s$Options", data
->prog
);
443 p
= buf
; /* Point to a buffer */
444 q
= data
->prog
; /* Point to program name */
445 while (*q
) /* While characters left here */
446 *p
++ = toupper(*q
++); /* Copy and uppercase */
447 *p
++ = 0; /* Terminate my copy of this */
448 p
= getenv(buf
); /* Get the value of the variable */
454 /* --- Copy the options string into a buffer --- */
456 if (p
) { /* If it is defined */
457 q
= malloc(strlen(p
) + 1); /* Allocate space for a copy */
458 if (!q
) { /* If that failed */
459 fprintf(stderr
, /* Report a nice error */
460 "%s: Not enough memory to read settings in "
461 "environment variable\n",
463 } else { /* Otherwise */
464 strcpy(q
, p
); /* Copy the text over */
465 data
->ind
= -1; /* Mark that we're parsing envvar */
466 data
->env
= data
->estart
= q
; /* And store the pointer away */
473 data
->ind
= data
->next
= 0;
476 /* --- Do some initial bodgery --- *
478 * The @shortopt@ string can have some interesting characters at the
479 * beginning. We'll skip past them.
482 switch (shortopt
[0]) {
490 if (shortopt
[0] == ':') {
495 if (longind
) /* Allow longind to be null */
496 *longind
= -1; /* Clear this to avoid confusion */
497 data
->opt
= -1; /* And this too */
498 data
->arg
= 0; /* No option set up here */
500 /* --- Now go off and search for an option --- */
502 if (!data
->list
|| !*data
->list
) {
503 data
->order
&= 3; /* Clear negation flag */
505 /* --- Now we need to find the next option --- *
507 * Exactly how we do this depends on the settings of the order variable.
508 * We identify options as being things starting with `%|-|%', and which
509 * aren't equal to `%|-|%' or `%|--|%'. We'll look for options until:
511 * * We find something which isn't an option AND @order == ord__posix@
512 * * We find a `%|--|%'
513 * * We reach the end of the list
515 * There are some added little wrinkles, which we'll meet as we go.
518 for (;;) { /* Keep looping for a while */
519 p
= mo__nextWord(argc
, argv
, data
); /* Get the next word out */
520 if (!p
) /* If there's no next word */
521 return (EOF
); /* There's no more now */
523 /* --- See if we've found an option --- */
525 if ((p
[0] == '-' || (p
[0] == '+' && flags
& gFlag_negation
)) &&
527 if (strcmp(p
, "--") == 0) { /* If this is the magic marker */
528 mo__permute(argv
, data
); /* Stow the magic marker item */
529 return (EOF
); /* There's nothing else to do */
531 break; /* We've found something! */
534 /* --- Figure out how to proceed --- */
536 switch (data
->order
& 3) {
537 case ord__posix
: /* POSIX option order */
538 return (EOF
); /* This is easy */
540 case ord__permute
: /* Permute the option order */
542 case ord__return
: /* Return each argument */
543 mo__permute(argv
, data
); /* Insert word in same place */
544 data
->arg
= p
; /* Point to the argument */
545 return (0); /* Return the value */
549 /* --- We found an option --- */
551 mo__permute(argv
, data
); /* Do any permuting necessary */
553 /* --- Check for a numeric option --- *
555 * We only check the first character (or the second if the first is a
556 * sign). This ought to be enough.
559 if (flags
& gFlag_numbers
&& (p
[0] == '-' || flags
& gFlag_negNumber
)) {
560 if (((p
[1] == '+' || p
[1] == '-') && isdigit((unsigned char)p
[2])) ||
561 isdigit((unsigned char)p
[1])) {
562 data
->opt
= strtol(p
+ 1, &data
->arg
, 10);
563 while (isspace((unsigned char)data
->arg
[0]))
567 return (p
[0] == '-' ?
'#' : '#' | gFlag_negated
);
571 /* --- Check for a long option --- */
574 data
->order
|= ord__negate
;
576 if (((p
[0] == '-' && p
[1] == '-') ||
577 (flags
& gFlag_noShorts
&& !mo__findOpt(p
[1], shortopt
, data
))) &&
578 (~flags
& gFlag_noLongs
)) /* Is this a long option? */
580 int match
= -1; /* Count matches as we go */
582 if (p
[0] == '+') { /* If it's negated */
583 data
->order
|= ord__negate
; /* Set the negate flag */
584 p
++; /* Point to the main text */
585 prefix
= "+"; /* Set the prefix string up */
586 } else if (p
[1] == '-') { /* If this is a `%|--|%' option */
587 if ((flags
& gFlag_negation
) && strncmp(p
+ 2, "no-", 3) == 0) {
588 p
+= 5; /* Point to main text */
589 prefix
= "--no-"; /* And set the prefix */
590 data
->order
|= ord__negate
; /* Set the negatedness flag */
592 p
+= 2; /* Point to the main text */
593 prefix
= "--"; /* Remember the prefix string */
596 if ((flags
& gFlag_negation
) && strncmp(p
+ 1, "no-", 3) == 0) {
597 p
+= 4; /* Find the text */
598 prefix
= "-no-"; /* Set the prefix */
599 data
->order
|= ord__negate
; /* Set negatedness flag */
601 p
++; /* Otherwise find the text */
602 prefix
= "-"; /* And remember the prefix */
606 for (i
= 0; longopts
[i
].name
; i
++) { /* Loop through the options */
607 if ((data
->order
& ord__negate
) &&
608 (~longopts
[i
].has_arg
& gFlag_negate
))
609 continue; /* If neg and opt doesn't allow */
611 r
= (char *) longopts
[i
].name
; /* Point to the name string */
612 q
= p
; /* Point to the string start */
613 for (;;) { /* Do a loop here */
614 if (*q
== 0 || *q
== '=') { /* End of the option string? */
615 if (*r
== 0) { /* If end of other string */
616 match
= i
; /* This is the match */
617 goto botched
; /* And exit the loop now */
619 if (match
== -1) { /* If no match currently */
620 match
= i
; /* Then this is it, here */
621 break; /* Stop looking now */
623 match
= -1; /* Else it's ambiguous */
624 goto botched
; /* So give up right now */
627 else if (*q
!= *r
) /* Otherwise if mismatch */
628 break; /* Abort this loop */
629 q
++, r
++; /* Increment the counters */
634 if (match
== -1) { /* If we couldn't find a match */
636 fprintf(stderr
, "%s: unrecognised option `%s%s'\n",
643 if (longind
) /* Allow longind to be null */
644 *longind
= match
; /* Store the match away */
646 /* --- Handle argument behaviour --- */
648 while (*p
!= 0 && *p
!= '=') /* Find the argument string */
650 p
= (*p ? p
+ 1 : 0); /* Sort out argument presence */
651 q
= (char *) longopts
[match
].name
; /* Remember the name here */
653 switch (longopts
[match
].has_arg
& 3) {
658 "%s: option `%s%s' does not accept arguments\n",
666 case required_argument
:
667 if (!p
) { /* If no argument given */
668 p
= mo__nextWord(argc
, argv
, data
);
670 if (!p
) { /* If no more arguments */
672 fprintf(stderr
, "%s: option `%s%s' requires an argument\n",
679 mo__permute(argv
, data
);
683 case optional_argument
:
689 /* --- Do correct things now we have a match --- */
691 if (longopts
[match
].flag
) { /* If he has a @flag@ argument */
692 if (longopts
[match
].has_arg
& gFlag_switch
) {
693 if (data
->order
& ord__negate
)
694 *longopts
[match
].flag
&= ~longopts
[match
].val
;
696 *longopts
[match
].flag
|= longopts
[match
].val
;
698 if (data
->order
& ord__negate
)
699 *longopts
[match
].flag
= 0;
701 *longopts
[match
].flag
= longopts
[match
].val
;
703 return (0); /* And return something */
705 if (data
->order
& ord__negate
)
706 return (longopts
[match
].val
| gFlag_negated
);
708 return (longopts
[match
].val
);
712 /* --- Do short options things --- */
715 if (p
[0] == '+') /* If starts with a `%|+|%' */
716 data
->order
|= ord__negate
;
717 data
->list
= p
+ 1; /* Omit leading `%|-|%'/`%|+|%' */
721 /* --- Now process the short options --- */
723 i
= *data
->list
++; /* Get the next option letter */
724 data
->opt
= i
; /* Store this away nicely */
726 p
= (char *) mo__findOpt(i
, shortopt
, data
);
727 if (!p
) { /* No more options left */
729 fprintf(stderr
, "%s: unknown option `%c%c'\n",
731 data
->order
& ord__negate ?
'+' : '-',
737 data
->opt
= i
; /* Store this for the caller */
739 /* --- Sort out an argument, if we expect one --- */
741 if (p
[0] == ':') { /* If we expect an option */
742 q
= (data
->list
[0] ? data
->list
: 0); /* If argument expected, use it */
743 data
->list
= 0; /* Kill the remaining options */
744 if (p
[1] != ':' && !q
) { /* If no arg, and not optional */
746 /* --- Same code as before --- */
748 q
= mo__nextWord(argc
, argv
, data
); /* Read the next word */
749 if (!q
) { /* If no more arguments */
751 fprintf(stderr
, "%s: option `%c%c' requires an argument\n",
753 data
->order
& ord__negate ?
'+' : '-',
758 mo__permute(argv
, data
);
763 return ((data
->order
& ord__negate
) ? i
| gFlag_negated
: i
);
766 /*----- That's all, folks -------------------------------------------------*/