3 * $Id: mdwopt.c,v 1.6 1999/05/20 23:00:42 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 1.6 1999/05/20 23:00:42 mdw
34 * Little formatting things.
36 * Revision 1.5 1999/05/19 20:23:59 mdw
37 * Change naming to match newer mLib conventions.
39 * Revision 1.4 1999/05/15 10:25:38 mdw
40 * Fix copyright information.
42 * Revision 1.3 1999/05/14 18:51:42 mdw
43 * Reformat the LGPL notice slightly.
45 * Revision 1.2 1999/05/13 22:57:23 mdw
46 * Change `-ise' to `-ize' throughout.
48 * Revision 1.1.1.1 1999/05/05 19:23:47 mdw
49 * New import. The old CVS repository was lost in a disk disaster.
51 * --- Previous lives ---
54 * Revision 1.7 1997/09/11 09:19:11 mdw
55 * (mo__nextWord): Arrrgh. Don't free the environment variable buffer!
56 * People are still using it!
58 * Revision 1.6 1997/09/11 09:05:54 mdw
59 * (mo__nextWord): Fix bug which returns too many words from environment
62 * Revision 1.5 1997/08/09 20:27:59 mdw
63 * Fix spelling of `Licensing'.
65 * Revision 1.4 1997/07/29 21:11:35 mdw
66 * Reformatted. Fixed buffer overflow when dealing with environment
67 * variables. Included NT in list of daft operating systems with `\' as a
68 * path separator. Fixed address of the FSF.
70 * Revision 1.3 1997/02/26 00:41:10 mdw
71 * Added GPL notice to the top. Slight formatting changes.
73 * Revision 1.2 1996/10/28 13:12:13 mdw
74 * Fixed calls to ctype.h routines. Arguments are cast to unsigned char
75 * to avoid invoking undefined behaviour caused by signedness of chars.
77 * Revision 1.1 1996/09/24 18:01:28 mdw
82 /*----- External dependencies ---------------------------------------------*/
91 /*----- Configuration things ----------------------------------------------*/
95 #elif defined(__OS2__) || defined(__MSDOS__) || defined(__WINNT__)
97 #else /* Assume a sane filing system */
101 /*----- Global variables --------------------------------------------------*/
103 mdwopt_data mdwopt_global = {0, 0, 0, 1, 0, 0, 0, 0, 0};
106 ORD_PERMUTE = 0, /* Permute the options (default) */
107 ORD_RETURN = 1, /* Return non-option things */
108 ORD_POSIX = 2, /* Do POSIX-type hacking */
109 ORD_NEGATE = 4 /* Magic negate-next-thing flag */
112 /*----- Main code ---------------------------------------------------------*/
114 /* --- @nextWord@ --- *
116 * Arguments: @int argc@ = number of command line options
117 * @char *argv[]@ = pointer to command line options
118 * @mdwopt_data *data@ = pointer to persistent state
120 * Returns: Pointer to the next word to handle, or 0
122 * Use: Extracts the next word from the command line or environment
126 static char *nextWord(int argc, char *const *argv, mdwopt_data *data)
128 if (data->ind == -1) {
131 while (isspace((unsigned char)*p))
134 while (*p && !isspace((unsigned char)*p))
145 if (data->next == argc)
147 return (argv[data->next++]);
150 /* --- @permute@ --- *
152 * Arguments: @char *argv[]@ = pointer to command line arguments
153 * @mdwopt_data *data@ = pointer to persistent data
157 * Use: Moves a command line option into the right place.
160 static void permute(char *const *argv, mdwopt_data *data)
162 char **v = (char **)argv;
163 if (data->ind != -1) {
164 int i = data->next - 1;
166 while (i > data->ind) {
175 /* --- @findOpt@ --- *
177 * Arguments: @int o@ = which option to search for
178 * @const char *shortopt@ = short options string to search
179 * @mdwopt_data *data@ = pointer to persistant state
181 * Returns: Pointer to rest of short options string (including magic
184 * Use: Looks up a short option in the given string.
187 static const char *findOpt(int o, const char *shortopt,
190 const char *p = shortopt; /* Point to short opts table */
192 if (!*p) /* No more options left */
195 if (o != *p || (p[1] != '+' && data->order & ORD_NEGATE)) {
196 p++; /* Skip this option entry */
197 while (*p == '+') /* Jump a `%|+|%' sign */
199 while (*p == ':') /* And jump any `%|:|%' characters */
200 p++; /* Just in case there are any */
207 /* --- @mdwopt@ --- *
209 * Arguments: @int argc@ = number of command line arguments
210 * @char * const *argv@ = pointer to command line arguments
211 * @const char *shortopt@ = pointer to short options information
212 * @const struct option *longopts@ = pointer to long opts info
213 * @int *longind@ = where to store matched longopt
214 * @mdwopt_data *data@ = persistent state for the parser
215 * @int flags@ = various useful flags
217 * Returns: Value of option found next, or an error character, or
218 * @EOF@ for the last thing.
220 * Use: Reads options. The routine should be more-or-less compatible
221 * with standard getopts, although it provides many more
222 * features even than the standard GNU implementation.
224 * The precise manner of options parsing is determined by
225 * various flag settings, which are described below. By setting
226 * flag values appropriately, you can achieve behaviour very
227 * similar to most other getopt routines.
230 * How options parsing appears to users
232 * A command line consists of a number of `words' (which may
233 * contain spaces, according to various shell quoting
234 * conventions). A word may be an option, an argument to an
235 * option, or a non-option. An option begins with a special
236 * character, usually `%|-|%', although `%|+|%' is also used
237 * sometimes. As special exceptions, the word containing only a
238 * `%|-|%' is considered to be a non-option, since it usually
239 * represents standard input or output as a filename, and the
240 * word containing a double-dash `%|--|%' is used to mark all
241 * following words as being non-options regardless of their
244 * Traditionally, all words after the first non-option have been
245 * considered to be non-options automatically, so that options
246 * must be specified before filenames. However, this
247 * implementation can extract all the options from the command
248 * line regardless of their position. This can usually be
249 * disabled by setting one of the environment variables
250 * `%|POSIXLY_CORRECT|%' or `%|_POSIX_OPTION_ORDER|%'.
252 * There are two different styles of options: `short' and
255 * Short options are the sort which Unix has known for ages: an
256 * option is a single letter, preceded by a `%|-|%'. Short
257 * options can be joined together to save space (and possibly to
258 * make silly words): e.g., instead of giving options
259 * `%|-x -y|%', a user could write `%|-xy|%'. Some short
260 * options can have arguments, which appear after the option
261 * letter, either immediately following, or in the next `word'
262 * (so an option with an argument could be written as
263 * `%|-o foo|%' or as `%|-ofoo|%'). Note that options with
264 * optional arguments must be written in the second style.
266 * When a short option controls a flag setting, it is sometimes
267 * possible to explicitly turn the flag off, as well as turning
268 * it on, (usually to override default options). This is
269 * usually done by using a `%|+|%' instead of a `%|-|%' to
270 * introduce the option.
272 * Long options, as popularized by the GNU utilities, are given
273 * long-ish memorable names, preceded by a double-dash `%|--|%'.
274 * Since their names are more than a single character, long
275 * options can't be combined in the same way as short options.
276 * Arguments to long options may be given either in the same
277 * `word', separated from the option name by an equals sign, or
278 * in the following `word'.
280 * Long option names can be abbreviated if necessary, as long
281 * as the abbreviation is unique. This means that options can
282 * have sensible and memorable names but still not require much
283 * typing from an experienced user.
285 * Like short options, long options can control flag settings.
286 * The options to manipulate these settings come in pairs: an
287 * option of the form `%|--set-flag|%' might set the flag, while
288 * an option of the form `%|--no-set-flag|%' might clear it.
290 * It is usual for applications to provide both short and long
291 * options with identical behaviour. Some applications with
292 * lots of options may only provide long options (although they
293 * will often be only two or three characters long). In this
294 * case, long options can be preceded with a single `%|-|%'
295 * character, and negated by a `%|+|%' character.
297 * Finally, some (older) programs accept arguments of the form
298 * `%%@.{"-"<number>}%%', to set some numerical parameter,
299 * typically a line count of some kind.
302 * How programs parse options
304 * An application parses its options by calling mdwopt
305 * repeatedly. Each time it is called, mdwopt returns a value
306 * describing the option just read, and stores information about
307 * the option in a data block. The value %$-1$% is returned
308 * when there are no more options to be read. The `%|?|%'
309 * character is returned when an error is encountered.
311 * Before starting to parse options, the value @data->ind@ must
312 * be set to 0 or 1. The value of @data->err@ can also be set,
313 * to choose whether errors are reported by mdwopt.
315 * The program's `@argc@' and `@argv@' arguments are passed to
316 * the options parser, so that it can read the command line. A
317 * flags word is also passed, allowing the program fine control
318 * over parsing. The flags are described above.
320 * Short options are described by a string, which once upon a
321 * time just contained the permitted option characters. Now the
322 * options string begins with a collection of flag characters,
323 * and various flag characters can be put after options
324 * characters to change their properties.
326 * If the first character of the short options string is
327 * `%|+|%', `%|-|%' or `%|!|%', the order in which options are
328 * read is modified, as follows:
330 * `%|+|%' forces the POSIX order to be used. As soon as a non-
331 * option is found, mdwopt returns %$-1$%.
333 * `%|-|%' makes mdwopt treat non-options as being `special'
334 * sorts of option. When a non-option word is found, the
335 * value 0 is returned, and the actual text of the word
336 * is stored as being the option's argument.
338 * `%|!|%' forces the default order to be used. The entire
339 * command line is scanned for options, which are
340 * returned in order. However, during this process,
341 * the options are moved in the @argv@ array, so that
342 * they appear before the non- options.
344 * A `%|:|%' character may be placed after the ordering flag (or
345 * at the very beginning if no ordering flag is given) which
346 * indicates that the character `%|:|%', rather than `%|?|%',
347 * should be returned if a missing argument error is detected.
349 * Each option in the string can be followed by a `%|+|%' sign,
350 * indicating that it can be negated, a `%|:|%' sign indicating
351 * that it requires an argument, or a `%|::|%' string,
352 * indicating an optional argument. Both `%|+|%' and `%|:|%' or
353 * `%|::|%' may be given, although the `%|+|%' must come first.
355 * If an option is found, the option character is returned to
356 * the caller. A pointer to an argument is stored in
357 * @data->arg@, or @NULL@ is stored if there was no argument.
358 * If a negated option was found, the option character is
359 * returned ORred with @OPTF_NEGATED@ (bit 8 set).
361 * Long options are described in a table. Each entry in the
362 * table is of type @struct option@, and the table is terminated
363 * by an entry whose @name@ field is null. Each option has
364 * a flags word which, due to historical reasons, is called
365 * @has_arg@. This describes various properties of the option,
366 * such as what sort of argument it takes, and whether it can
369 * When mdwopt finds a long option, it looks the name up in the
370 * table. The index of the matching entry is stored in the
371 * @longind@ variable, passed to mdwopt (unless @longind@ is 0):
372 * a value of %$-1$% indicates that no long option was
373 * found. The behaviour is then dependent on the values in the
374 * table entry. If @flag@ is nonzero, it points to an integer
375 * to be modified by mdwopt. Usually the value in the @val@
376 * field is simply stored in the @flag@ variable. If the flag
377 * @OPTF_SWITCH@ is set, however, the value is combined with
378 * the existing value of the flags using a bitwise OR. If
379 * @OPTF_NEGATE@ is set, then the flag bit will be cleared if a
380 * matching negated long option is found. The value 0 is
383 * If @flag@ is zero, the value in @val@ is returned by mdwopt,
384 * possibly with bit 8 set if the option was negated.
386 * Arguments for long options are stored in @data->arg@, as
389 * Numeric options, if enabled, cause the value `%|#|%' to be
390 * returned, and the numeric value to be stored in @data->opt@.
392 * If the flag @OPTF_ENVVAR@ is set on entry, options will be
393 * extracted from an environment variable whose name is built by
394 * capitalizing all the letters of the program's name. (This
395 * allows a user to have different default settings for a
396 * program, by calling it through different symbolic links.)
399 int mdwopt(int argc, char *const *argv,
400 const char *shortopt,
401 const struct option *longopts, int *longind,
402 mdwopt_data *data, int flags)
404 /* --- Local variables --- */
406 char *p, *q, *r; /* Some useful things to have */
407 char *prefix; /* Prefix from this option */
408 int i; /* Always useful */
409 char noarg = '?'; /* Standard missing-arg char */
411 /* --- Sort out our data --- */
413 if (!data) /* If default data requested */
414 data = &mdwopt_global; /* Then use the global stuff */
416 /* --- See if this is the first time --- */
418 if (data->ind == 0 || (data->ind == 1 && ~flags & OPTF_NOPROGNAME)) {
420 /* --- Sort out default returning order --- */
422 if (getenv("_POSIX_OPTION_ORDER") || /* Examine environment for opts */
423 getenv("POSIXLY_CORRECT")) /* To see if we disable features */
424 data->order = ORD_POSIX; /* If set, use POSIX ordering */
426 data->order = ORD_PERMUTE; /* Otherwise mangle the options */
428 /* --- Now see what the caller actually wants --- */
430 switch (shortopt[0]) { /* Look at the first character */
431 case '-': /* `%|-|%' turns on in-orderness */
432 data->order = ORD_RETURN;
434 case '+': /* `%|+|%' turns on POSIXness */
435 data->order = ORD_POSIX;
437 case '!': /* `%|!|%' ignores POSIXness */
438 data->order = ORD_PERMUTE;
442 /* --- Now decide on the program's name --- */
444 if (~flags & OPTF_NOPROGNAME) {
445 p = q = (char *)argv[0];
452 data->ind = data->next = 1;
455 /* --- See about environment variables --- *
457 * Be careful. The program may be setuid, and an attacker might have
458 * given us a long name in @argv[0]@. If the name is very long, don't
459 * support this option.
462 if (flags & OPTF_ENVVAR && strlen(data->prog) < 48) {
466 /* --- For RISC OS, support a different format --- *
468 * Acorn's RISC OS tends to put settings in variables named
469 * `App$Options' rather than `APP'. Under RISC OS, I'll support
470 * both methods, just to avoid confuddlement.
474 sprintf(buf, "%s$Options", data->prog);
479 p = buf; /* Point to a buffer */
480 q = data->prog; /* Point to program name */
481 while (*q) /* While characters left here */
482 *p++ = toupper(*q++); /* Copy and uppercase */
483 *p++ = 0; /* Terminate my copy of this */
484 p = getenv(buf); /* Get the value of the variable */
490 /* --- Copy the options string into a buffer --- */
492 if (p) { /* If it is defined */
493 q = malloc(strlen(p) + 1); /* Allocate space for a copy */
494 if (!q) { /* If that failed */
495 fprintf(stderr, /* Report a nice error */
496 "%s: Not enough memory to read settings in "
497 "environment variable\n",
499 } else { /* Otherwise */
500 strcpy(q, p); /* Copy the text over */
501 data->ind = -1; /* Mark that we're parsing envvar */
502 data->env = data->estart = q; /* And store the pointer away */
509 data->ind = data->next = 0;
512 /* --- Do some initial bodgery --- *
514 * The @shortopt@ string can have some interesting characters at the
515 * beginning. We'll skip past them.
518 switch (shortopt[0]) {
526 if (shortopt[0] == ':') {
531 if (longind) /* Allow longind to be null */
532 *longind = -1; /* Clear this to avoid confusion */
533 data->opt = -1; /* And this too */
534 data->arg = 0; /* No option set up here */
536 /* --- Now go off and search for an option --- */
538 if (!data->list || !*data->list) {
539 data->order &= 3; /* Clear negation flag */
541 /* --- Now we need to find the next option --- *
543 * Exactly how we do this depends on the settings of the order variable.
544 * We identify options as being things starting with `%|-|%', and which
545 * aren't equal to `%|-|%' or `%|--|%'. We'll look for options until:
547 * * We find something which isn't an option AND @order == ORD_POSIX@
548 * * We find a `%|--|%'
549 * * We reach the end of the list
551 * There are some added little wrinkles, which we'll meet as we go.
554 for (;;) { /* Keep looping for a while */
555 p = nextWord(argc, argv, data); /* Get the next word out */
556 if (!p) /* If there's no next word */
557 return (EOF); /* There's no more now */
559 /* --- See if we've found an option --- */
561 if ((p[0] == '-' || (p[0] == '+' && flags & OPTF_NEGATION)) &&
563 if (strcmp(p, "--") == 0) { /* If this is the magic marker */
564 permute(argv, data); /* Stow the magic marker item */
565 return (EOF); /* There's nothing else to do */
567 break; /* We've found something! */
570 /* --- Figure out how to proceed --- */
572 switch (data->order & 3) {
573 case ORD_POSIX: /* POSIX option order */
574 return (EOF); /* This is easy */
576 case ORD_PERMUTE: /* Permute the option order */
578 case ORD_RETURN: /* Return each argument */
579 permute(argv, data); /* Insert word in same place */
580 data->arg = p; /* Point to the argument */
581 return (0); /* Return the value */
585 /* --- We found an option --- */
587 permute(argv, data); /* Do any permuting necessary */
589 /* --- Check for a numeric option --- *
591 * We only check the first character (or the second if the first is a
592 * sign). This ought to be enough.
595 if (flags & OPTF_NUMBERS && (p[0] == '-' || flags & OPTF_NEGNUMBER)) {
596 if (((p[1] == '+' || p[1] == '-') && isdigit((unsigned char)p[2])) ||
597 isdigit((unsigned char)p[1])) {
598 data->opt = strtol(p + 1, &data->arg, 10);
599 while (isspace((unsigned char)data->arg[0]))
603 return (p[0] == '-' ? '#' : '#' | OPTF_NEGATED);
607 /* --- Check for a long option --- */
610 data->order |= ORD_NEGATE;
612 if (((p[0] == '-' && p[1] == '-') ||
613 (flags & OPTF_NOSHORTS && !findOpt(p[1], shortopt, data))) &&
614 (~flags & OPTF_NOLONGS)) /* Is this a long option? */
616 int match = -1; /* Count matches as we go */
618 if (p[0] == '+') { /* If it's negated */
619 data->order |= ORD_NEGATE; /* Set the negate flag */
620 p++; /* Point to the main text */
621 prefix = "+"; /* Set the prefix string up */
622 } else if (p[1] == '-') { /* If this is a `%|--|%' option */
623 if ((flags & OPTF_NEGATION) && strncmp(p + 2, "no-", 3) == 0) {
624 p += 5; /* Point to main text */
625 prefix = "--no-"; /* And set the prefix */
626 data->order |= ORD_NEGATE; /* Set the negatedness flag */
628 p += 2; /* Point to the main text */
629 prefix = "--"; /* Remember the prefix string */
632 if ((flags & OPTF_NEGATION) && strncmp(p + 1, "no-", 3) == 0) {
633 p += 4; /* Find the text */
634 prefix = "-no-"; /* Set the prefix */
635 data->order |= ORD_NEGATE; /* Set negatedness flag */
637 p++; /* Otherwise find the text */
638 prefix = "-"; /* And remember the prefix */
642 for (i = 0; longopts[i].name; i++) { /* Loop through the options */
643 if ((data->order & ORD_NEGATE) &&
644 (~longopts[i].has_arg & OPTF_NEGATE))
645 continue; /* If neg and opt doesn't allow */
647 r = (char *) longopts[i].name; /* Point to the name string */
648 q = p; /* Point to the string start */
649 for (;;) { /* Do a loop here */
650 if (*q == 0 || *q == '=') { /* End of the option string? */
651 if (*r == 0) { /* If end of other string */
652 match = i; /* This is the match */
653 goto botched; /* And exit the loop now */
655 if (match == -1) { /* If no match currently */
656 match = i; /* Then this is it, here */
657 break; /* Stop looking now */
659 match = -1; /* Else it's ambiguous */
660 goto botched; /* So give up right now */
663 else if (*q != *r) /* Otherwise if mismatch */
664 break; /* Abort this loop */
665 q++, r++; /* Increment the counters */
670 if (match == -1) { /* If we couldn't find a match */
672 fprintf(stderr, "%s: unrecognized option `%s%s'\n",
679 if (longind) /* Allow longind to be null */
680 *longind = match; /* Store the match away */
682 /* --- Handle argument behaviour --- */
684 while (*p != 0 && *p != '=') /* Find the argument string */
686 p = (*p ? p + 1 : 0); /* Sort out argument presence */
687 q = (char *) longopts[match].name; /* Remember the name here */
689 switch (longopts[match].has_arg & 3) {
694 "%s: option `%s%s' does not accept arguments\n",
702 case required_argument:
703 if (!p) { /* If no argument given */
704 p = nextWord(argc, argv, data);
706 if (!p) { /* If no more arguments */
708 fprintf(stderr, "%s: option `%s%s' requires an argument\n",
719 case optional_argument:
725 /* --- Do correct things now we have a match --- */
727 if (longopts[match].flag) { /* If he has a @flag@ argument */
728 if (longopts[match].has_arg & OPTF_SWITCH) {
729 if (data->order & ORD_NEGATE)
730 *longopts[match].flag &= ~longopts[match].val;
732 *longopts[match].flag |= longopts[match].val;
734 if (data->order & ORD_NEGATE)
735 *longopts[match].flag = 0;
737 *longopts[match].flag = longopts[match].val;
739 return (0); /* And return something */
741 if (data->order & ORD_NEGATE)
742 return (longopts[match].val | OPTF_NEGATED);
744 return (longopts[match].val);
748 /* --- Do short options things --- */
751 if (p[0] == '+') /* If starts with a `%|+|%' */
752 data->order |= ORD_NEGATE;
753 data->list = p + 1; /* Omit leading `%|-|%'/`%|+|%' */
757 /* --- Now process the short options --- */
759 i = *data->list++; /* Get the next option letter */
760 data->opt = i; /* Store this away nicely */
762 p = (char *) findOpt(i, shortopt, data);
763 if (!p) { /* No more options left */
765 fprintf(stderr, "%s: unknown option `%c%c'\n",
767 data->order & ORD_NEGATE ? '+' : '-',
773 data->opt = i; /* Store this for the caller */
775 /* --- Sort out an argument, if we expect one --- */
777 if (p[0] == ':') { /* If we expect an option */
778 q = (data->list[0] ? data->list : 0); /* If argument expected, use it */
779 data->list = 0; /* Kill the remaining options */
780 if (p[1] != ':' && !q) { /* If no arg, and not optional */
782 /* --- Same code as before --- */
784 q = nextWord(argc, argv, data); /* Read the next word */
785 if (!q) { /* If no more arguments */
787 fprintf(stderr, "%s: option `%c%c' requires an argument\n",
789 data->order & ORD_NEGATE ? '+' : '-',
799 return ((data->order & ORD_NEGATE) ? i | OPTF_NEGATED : i);
802 /*----- That's all, folks -------------------------------------------------*/