xref: /freebsd/usr.bin/locale/locale.c (revision 9e5787d2284e187abb5b654d924394a65772e004)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2002, 2003 Alexey Zelkin <phantom@FreeBSD.org>
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26  * SUCH DAMAGE.
27  *
28  * $FreeBSD$
29  */
30 
31 /*
32  * XXX: implement missing era_* (LC_TIME) keywords (require libc &
33  *	nl_langinfo(3) extensions)
34  *
35  * XXX: correctly handle reserved 'charmap' keyword and '-m' option (require
36  *	localedef(1) implementation).  Currently it's handled via
37  *	nl_langinfo(CODESET).
38  */
39 
40 #include <sys/param.h>
41 #include <sys/types.h>
42 #include <sys/sbuf.h>
43 
44 #include <dirent.h>
45 #include <err.h>
46 #include <limits.h>
47 #include <locale.h>
48 #include <langinfo.h>
49 #include <stdio.h>
50 #include <stdlib.h>
51 #include <string.h>
52 #include <stringlist.h>
53 #include <unistd.h>
54 #include "setlocale.h"
55 
56 /* Local prototypes */
57 char	*format_grouping(char *);
58 void	init_locales_list(void);
59 void	list_charmaps(void);
60 void	list_locales(void);
61 const char *lookup_localecat(int);
62 char	*kwval_lconv(int);
63 int	kwval_lookup(const char *, char **, int *, int *, int *);
64 int	showdetails(const char *);
65 void	showkeywordslist(char *substring);
66 void	showlocale(void);
67 void	usage(void);
68 
69 /* Global variables */
70 static StringList *locales = NULL;
71 
72 static int	all_locales = 0;
73 static int	all_charmaps = 0;
74 static int	prt_categories = 0;
75 static int	prt_keywords = 0;
76 
77 static const struct _lcinfo {
78 	const char	*name;
79 	int		id;
80 } lcinfo [] = {
81 	{ "LC_CTYPE",		LC_CTYPE },
82 	{ "LC_COLLATE",		LC_COLLATE },
83 	{ "LC_TIME",		LC_TIME },
84 	{ "LC_NUMERIC",		LC_NUMERIC },
85 	{ "LC_MONETARY",	LC_MONETARY },
86 	{ "LC_MESSAGES",	LC_MESSAGES }
87 };
88 #define	NLCINFO nitems(lcinfo)
89 
90 /* ids for values not referenced by nl_langinfo() */
91 enum {
92 	KW_GROUPING,
93 	KW_INT_CURR_SYMBOL,
94 	KW_CURRENCY_SYMBOL,
95 	KW_MON_DECIMAL_POINT,
96 	KW_MON_THOUSANDS_SEP,
97 	KW_MON_GROUPING,
98 	KW_POSITIVE_SIGN,
99 	KW_NEGATIVE_SIGN,
100 	KW_INT_FRAC_DIGITS,
101 	KW_FRAC_DIGITS,
102 	KW_P_CS_PRECEDES,
103 	KW_P_SEP_BY_SPACE,
104 	KW_N_CS_PRECEDES,
105 	KW_N_SEP_BY_SPACE,
106 	KW_P_SIGN_POSN,
107 	KW_N_SIGN_POSN,
108 	KW_INT_P_CS_PRECEDES,
109 	KW_INT_P_SEP_BY_SPACE,
110 	KW_INT_N_CS_PRECEDES,
111 	KW_INT_N_SEP_BY_SPACE,
112 	KW_INT_P_SIGN_POSN,
113 	KW_INT_N_SIGN_POSN,
114 	KW_TIME_DAY,
115 	KW_TIME_ABDAY,
116 	KW_TIME_MON,
117 	KW_TIME_ABMON,
118 	KW_TIME_AM_PM
119 };
120 
121 enum {
122 	TYPE_NUM,
123 	TYPE_STR,
124 	TYPE_UNQ
125 };
126 
127 enum {
128 	SRC_LINFO,
129 	SRC_LCONV,
130 	SRC_LTIME
131 };
132 
133 static const struct _kwinfo {
134 	const char	*name;
135 	int		type;
136 	int		catid;		/* LC_* */
137 	int		source;
138 	int		value_ref;
139 	const char	*comment;
140 } kwinfo [] = {
141 	{ "charmap",		TYPE_STR, LC_CTYPE,	SRC_LINFO,
142 	  CODESET, "" },					/* hack */
143 
144 	/* LC_MONETARY - POSIX */
145 	{ "int_curr_symbol",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
146 	  KW_INT_CURR_SYMBOL, "" },
147 	{ "currency_symbol",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
148 	  KW_CURRENCY_SYMBOL, "" },
149 	{ "mon_decimal_point",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
150 	  KW_MON_DECIMAL_POINT, "" },
151 	{ "mon_thousands_sep",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
152 	  KW_MON_THOUSANDS_SEP, "" },
153 	{ "mon_grouping",	TYPE_UNQ, LC_MONETARY,	SRC_LCONV,
154 	  KW_MON_GROUPING, "" },
155 	{ "positive_sign",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
156 	  KW_POSITIVE_SIGN, "" },
157 	{ "negative_sign",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
158 	  KW_NEGATIVE_SIGN, "" },
159 	{ "int_frac_digits",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
160 	  KW_INT_FRAC_DIGITS, "" },
161 	{ "frac_digits",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
162 	  KW_FRAC_DIGITS, "" },
163 	{ "p_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
164 	  KW_P_CS_PRECEDES, "" },
165 	{ "p_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
166 	  KW_P_SEP_BY_SPACE, "" },
167 	{ "n_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
168 	  KW_N_CS_PRECEDES, "" },
169 	{ "n_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
170 	  KW_N_SEP_BY_SPACE, "" },
171 	{ "p_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
172 	  KW_P_SIGN_POSN, "" },
173 	{ "n_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
174 	  KW_N_SIGN_POSN, "" },
175 	{ "int_p_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
176 	  KW_INT_P_CS_PRECEDES, "" },
177 	{ "int_p_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
178 	  KW_INT_P_SEP_BY_SPACE, "" },
179 	{ "int_n_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
180 	  KW_INT_N_CS_PRECEDES, "" },
181 	{ "int_n_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
182 	  KW_INT_N_SEP_BY_SPACE, "" },
183 	{ "int_p_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
184 	  KW_INT_P_SIGN_POSN, "" },
185 	{ "int_n_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
186 	  KW_INT_N_SIGN_POSN, "" },
187 
188 	/* LC_NUMERIC - POSIX */
189 	{ "decimal_point",	TYPE_STR, LC_NUMERIC,	SRC_LINFO,
190 	  RADIXCHAR, "" },
191 	{ "thousands_sep",	TYPE_STR, LC_NUMERIC,	SRC_LINFO,
192 	  THOUSEP, "" },
193 	{ "grouping",		TYPE_UNQ, LC_NUMERIC,	SRC_LCONV,
194 	  KW_GROUPING, "" },
195 	/* LC_NUMERIC - local additions */
196 	{ "radixchar",		TYPE_STR, LC_NUMERIC,	SRC_LINFO,
197 	  RADIXCHAR, "Same as decimal_point (FreeBSD only)" },	/* compat */
198 	{ "thousep",		TYPE_STR, LC_NUMERIC,	SRC_LINFO,
199 	  THOUSEP, "Same as thousands_sep (FreeBSD only)" },	/* compat */
200 
201 	/* LC_TIME - POSIX */
202 	{ "abday",		TYPE_STR, LC_TIME,	SRC_LTIME,
203 	  KW_TIME_ABDAY, "" },
204 	{ "day",		TYPE_STR, LC_TIME,	SRC_LTIME,
205 	  KW_TIME_DAY, "" },
206 	{ "abmon",		TYPE_STR, LC_TIME,	SRC_LTIME,
207 	  KW_TIME_ABMON, "" },
208 	{ "mon",		TYPE_STR, LC_TIME,	SRC_LTIME,
209 	  KW_TIME_MON, "" },
210 	{ "d_t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
211 	  D_T_FMT, "" },
212 	{ "d_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
213 	  D_FMT, "" },
214 	{ "t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
215 	  T_FMT, "" },
216 	{ "am_pm",		TYPE_STR, LC_TIME,	SRC_LTIME,
217 	  KW_TIME_AM_PM, "" },
218 	{ "t_fmt_ampm",		TYPE_STR, LC_TIME,	SRC_LINFO,
219 	  T_FMT_AMPM, "" },
220 	{ "era",		TYPE_UNQ, LC_TIME,	SRC_LINFO,
221 	  ERA, "(unavailable)" },
222 	{ "era_d_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
223 	  ERA_D_FMT, "(unavailable)" },
224 	{ "era_d_t_fmt",	TYPE_STR, LC_TIME,	SRC_LINFO,
225 	  ERA_D_T_FMT, "(unavailable)" },
226 	{ "era_t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
227 	  ERA_T_FMT, "(unavailable)" },
228 	{ "alt_digits",		TYPE_UNQ, LC_TIME,	SRC_LINFO,
229 	  ALT_DIGITS, "" },
230 	/* LC_TIME - local additions */
231 	{ "abday_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
232 	  ABDAY_1, "(FreeBSD only)" },
233 	{ "abday_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
234 	  ABDAY_2, "(FreeBSD only)" },
235 	{ "abday_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
236 	  ABDAY_3, "(FreeBSD only)" },
237 	{ "abday_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
238 	  ABDAY_4, "(FreeBSD only)" },
239 	{ "abday_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
240 	  ABDAY_5, "(FreeBSD only)" },
241 	{ "abday_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
242 	  ABDAY_6, "(FreeBSD only)" },
243 	{ "abday_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
244 	  ABDAY_7, "(FreeBSD only)" },
245 	{ "day_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
246 	  DAY_1, "(FreeBSD only)" },
247 	{ "day_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
248 	  DAY_2, "(FreeBSD only)" },
249 	{ "day_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
250 	  DAY_3, "(FreeBSD only)" },
251 	{ "day_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
252 	  DAY_4, "(FreeBSD only)" },
253 	{ "day_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
254 	  DAY_5, "(FreeBSD only)" },
255 	{ "day_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
256 	  DAY_6, "(FreeBSD only)" },
257 	{ "day_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
258 	  DAY_7, "(FreeBSD only)" },
259 	{ "abmon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
260 	  ABMON_1, "(FreeBSD only)" },
261 	{ "abmon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
262 	  ABMON_2, "(FreeBSD only)" },
263 	{ "abmon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
264 	  ABMON_3, "(FreeBSD only)" },
265 	{ "abmon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
266 	  ABMON_4, "(FreeBSD only)" },
267 	{ "abmon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
268 	  ABMON_5, "(FreeBSD only)" },
269 	{ "abmon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
270 	  ABMON_6, "(FreeBSD only)" },
271 	{ "abmon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
272 	  ABMON_7, "(FreeBSD only)" },
273 	{ "abmon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
274 	  ABMON_8, "(FreeBSD only)" },
275 	{ "abmon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
276 	  ABMON_9, "(FreeBSD only)" },
277 	{ "abmon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
278 	  ABMON_10, "(FreeBSD only)" },
279 	{ "abmon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
280 	  ABMON_11, "(FreeBSD only)" },
281 	{ "abmon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
282 	  ABMON_12, "(FreeBSD only)" },
283 	{ "mon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
284 	  MON_1, "(FreeBSD only)" },
285 	{ "mon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
286 	  MON_2, "(FreeBSD only)" },
287 	{ "mon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
288 	  MON_3, "(FreeBSD only)" },
289 	{ "mon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
290 	  MON_4, "(FreeBSD only)" },
291 	{ "mon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
292 	  MON_5, "(FreeBSD only)" },
293 	{ "mon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
294 	  MON_6, "(FreeBSD only)" },
295 	{ "mon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
296 	  MON_7, "(FreeBSD only)" },
297 	{ "mon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
298 	  MON_8, "(FreeBSD only)" },
299 	{ "mon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
300 	  MON_9, "(FreeBSD only)" },
301 	{ "mon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
302 	  MON_10, "(FreeBSD only)" },
303 	{ "mon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
304 	  MON_11, "(FreeBSD only)" },
305 	{ "mon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
306 	  MON_12, "(FreeBSD only)" },
307 	{ "altmon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
308 	  ALTMON_1, "(FreeBSD only)" },
309 	{ "altmon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
310 	  ALTMON_2, "(FreeBSD only)" },
311 	{ "altmon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
312 	  ALTMON_3, "(FreeBSD only)" },
313 	{ "altmon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
314 	  ALTMON_4, "(FreeBSD only)" },
315 	{ "altmon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
316 	  ALTMON_5, "(FreeBSD only)" },
317 	{ "altmon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
318 	  ALTMON_6, "(FreeBSD only)" },
319 	{ "altmon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
320 	  ALTMON_7, "(FreeBSD only)" },
321 	{ "altmon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
322 	  ALTMON_8, "(FreeBSD only)" },
323 	{ "altmon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
324 	  ALTMON_9, "(FreeBSD only)" },
325 	{ "altmon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
326 	  ALTMON_10, "(FreeBSD only)" },
327 	{ "altmon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
328 	  ALTMON_11, "(FreeBSD only)" },
329 	{ "altmon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
330 	  ALTMON_12, "(FreeBSD only)" },
331 	{ "am_str",		TYPE_STR, LC_TIME,	SRC_LINFO,
332 	  AM_STR, "(FreeBSD only)" },
333 	{ "pm_str",		TYPE_STR, LC_TIME,	SRC_LINFO,
334 	  PM_STR, "(FreeBSD only)" },
335 	{ "d_md_order",		TYPE_STR, LC_TIME,	SRC_LINFO,
336 	  D_MD_ORDER, "(FreeBSD only)" },			/* local */
337 
338 	/* LC_MESSAGES - POSIX */
339 	{ "yesexpr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
340 	  YESEXPR, "" },
341 	{ "noexpr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
342 	  NOEXPR, "" },
343 	/* LC_MESSAGES - local additions */
344 	{ "yesstr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
345 	  YESSTR, "(POSIX legacy)" },				/* compat */
346 	{ "nostr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
347 	  NOSTR, "(POSIX legacy)" }				/* compat */
348 
349 };
350 #define	NKWINFO (nitems(kwinfo))
351 
352 static const char *boguslocales[] = { "UTF-8" };
353 #define	NBOGUS	(nitems(boguslocales))
354 
355 int
356 main(int argc, char *argv[])
357 {
358 	int	ch;
359 	int	tmp;
360 
361 	while ((ch = getopt(argc, argv, "ackms:")) != -1) {
362 		switch (ch) {
363 		case 'a':
364 			all_locales = 1;
365 			break;
366 		case 'c':
367 			prt_categories = 1;
368 			break;
369 		case 'k':
370 			prt_keywords = 1;
371 			break;
372 		case 'm':
373 			all_charmaps = 1;
374 			break;
375 		default:
376 			usage();
377 		}
378 	}
379 	argc -= optind;
380 	argv += optind;
381 
382 	/* validate arguments */
383 	if (all_locales && all_charmaps)
384 		usage();
385 	if ((all_locales || all_charmaps) && argc > 0)
386 		usage();
387 	if ((all_locales || all_charmaps) && (prt_categories || prt_keywords))
388 		usage();
389 
390 	/* process '-a' */
391 	if (all_locales) {
392 		list_locales();
393 		exit(0);
394 	}
395 
396 	/* process '-m' */
397 	if (all_charmaps) {
398 		list_charmaps();
399 		exit(0);
400 	}
401 
402 	/* check for special case '-k list' */
403 	tmp = 0;
404 	if (prt_keywords && argc > 0)
405 		while (tmp < argc)
406 			if (strcasecmp(argv[tmp++], "list") == 0) {
407 				showkeywordslist(argv[tmp]);
408 				exit(0);
409 			}
410 
411 	/* process '-c', '-k', or command line arguments. */
412 	if (prt_categories || prt_keywords || argc > 0) {
413 		if (prt_keywords || argc > 0)
414 			setlocale(LC_ALL, "");
415 		if (argc > 0) {
416 			while (argc > 0) {
417 				if (showdetails(*argv) != 0)
418 					exit(EXIT_FAILURE);
419 				argv++;
420 				argc--;
421 			}
422 		} else {
423 			uint i;
424 			for (i = 0; i < nitems(kwinfo); i++)
425 				showdetails(kwinfo[i].name);
426 		}
427 		exit(0);
428 	}
429 
430 	/* no arguments, show current locale state */
431 	showlocale();
432 
433 	return (0);
434 }
435 
436 void
437 usage(void)
438 {
439 	printf("Usage: locale [ -a | -m ]\n"
440 	       "       locale -k list [prefix]\n"
441 	       "       locale [ -ck ] [keyword ...]\n");
442 	exit(1);
443 }
444 
445 /*
446  * Output information about all available locales
447  *
448  * XXX actually output of this function does not guarantee that locale
449  *     is really available to application, since it can be broken or
450  *     inconsistent thus setlocale() will fail.  Maybe add '-V' function to
451  *     also validate these locales?
452  */
453 void
454 list_locales(void)
455 {
456 	size_t i;
457 
458 	init_locales_list();
459 	for (i = 0; i < locales->sl_cur; i++) {
460 		printf("%s\n", locales->sl_str[i]);
461 	}
462 }
463 
464 /*
465  * qsort() helper function
466  */
467 static int
468 scmp(const void *s1, const void *s2)
469 {
470 	return strcmp(*(const char * const *)s1, *(const char * const *)s2);
471 }
472 
473 /*
474  * Output information about all available charmaps
475  *
476  * XXX this function is doing a task in hackish way, i.e. by scaning
477  *     list of locales, spliting their codeset part and building list of
478  *     them.
479  */
480 void
481 list_charmaps(void)
482 {
483 	size_t i;
484 	char *s, *cs;
485 	StringList *charmaps;
486 
487 	/* initialize StringList */
488 	charmaps = sl_init();
489 	if (charmaps == NULL)
490 		err(1, "could not allocate memory");
491 
492 	/* fetch locales list */
493 	init_locales_list();
494 
495 	/* split codesets and build their list */
496 	for (i = 0; i < locales->sl_cur; i++) {
497 		s = locales->sl_str[i];
498 		if ((cs = strchr(s, '.')) != NULL) {
499 			cs++;
500 			if (sl_find(charmaps, cs) == NULL)
501 				sl_add(charmaps, cs);
502 		}
503 	}
504 
505 	/* add US-ASCII, if not yet added */
506 	if (sl_find(charmaps, "US-ASCII") == NULL)
507 		sl_add(charmaps, strdup("US-ASCII"));
508 
509 	/* sort the list */
510 	qsort(charmaps->sl_str, charmaps->sl_cur, sizeof(char *), scmp);
511 
512 	/* print results */
513 	for (i = 0; i < charmaps->sl_cur; i++) {
514 		printf("%s\n", charmaps->sl_str[i]);
515 	}
516 }
517 
518 /*
519  * Retrieve sorted list of system locales (or user locales, if PATH_LOCALE
520  * environment variable is set)
521  */
522 void
523 init_locales_list(void)
524 {
525 	DIR *dirp;
526 	struct dirent *dp;
527 	size_t i;
528 	int bogus;
529 
530 	/* why call this function twice ? */
531 	if (locales != NULL)
532 		return;
533 
534 	/* initialize StringList */
535 	locales = sl_init();
536 	if (locales == NULL)
537 		err(1, "could not allocate memory");
538 
539 	/* get actual locales directory name */
540 	if (__detect_path_locale() != 0)
541 		err(1, "unable to find locales storage");
542 
543 	/* open locales directory */
544 	dirp = opendir(_PathLocale);
545 	if (dirp == NULL)
546 		err(1, "could not open directory '%s'", _PathLocale);
547 
548 	/* scan directory and store its contents except "." and ".." */
549 	while ((dp = readdir(dirp)) != NULL) {
550 		if (*(dp->d_name) == '.')
551 			continue;		/* exclude "." and ".." */
552 		for (bogus = i = 0; i < NBOGUS; i++)
553 			if (strncmp(dp->d_name, boguslocales[i],
554 			    strlen(boguslocales[i])) == 0)
555 				bogus = 1;
556 		if (!bogus)
557 			sl_add(locales, strdup(dp->d_name));
558 	}
559 	closedir(dirp);
560 
561 	/* make sure that 'POSIX' and 'C' locales are present in the list.
562 	 * POSIX 1003.1-2001 requires presence of 'POSIX' name only here, but
563 	 * we also list 'C' for constistency
564 	 */
565 	if (sl_find(locales, "POSIX") == NULL)
566 		sl_add(locales, strdup("POSIX"));
567 
568 	if (sl_find(locales, "C") == NULL)
569 		sl_add(locales, strdup("C"));
570 
571 	/* make output nicer, sort the list */
572 	qsort(locales->sl_str, locales->sl_cur, sizeof(char *), scmp);
573 }
574 
575 /*
576  * Show current locale status, depending on environment variables
577  */
578 void
579 showlocale(void)
580 {
581 	size_t	i;
582 	const char *lang, *vval, *eval;
583 
584 	setlocale(LC_ALL, "");
585 
586 	lang = getenv("LANG");
587 	if (lang == NULL) {
588 		lang = "";
589 	}
590 	printf("LANG=%s\n", lang);
591 	/* XXX: if LANG is null, then set it to "C" to get implied values? */
592 
593 	for (i = 0; i < NLCINFO; i++) {
594 		vval = setlocale(lcinfo[i].id, NULL);
595 		eval = getenv(lcinfo[i].name);
596 		if (eval != NULL && !strcmp(eval, vval)
597 				&& strcmp(lang, vval)) {
598 			/*
599 			 * Appropriate environment variable set, its value
600 			 * is valid and not overridden by LC_ALL
601 			 *
602 			 * XXX: possible side effect: if both LANG and
603 			 * overridden environment variable are set into same
604 			 * value, then it'll be assumed as 'implied'
605 			 */
606 			printf("%s=%s\n", lcinfo[i].name, vval);
607 		} else {
608 			printf("%s=\"%s\"\n", lcinfo[i].name, vval);
609 		}
610 	}
611 
612 	vval = getenv("LC_ALL");
613 	if (vval == NULL) {
614 		vval = "";
615 	}
616 	printf("LC_ALL=%s\n", vval);
617 }
618 
619 char *
620 format_grouping(char *binary)
621 {
622 	static char rval[64];
623 	const char *cp;
624 	size_t roff;
625 	int len;
626 
627 	/*
628 	 * XXX This check will need to be modified if/when localeconv() is
629 	 * fixed (PR172215).
630 	 */
631 	if (*binary == CHAR_MAX)
632 		return (binary);
633 
634 	rval[0] = '\0';
635 	roff = 0;
636 	for (cp = binary; *cp != '\0'; ++cp) {
637 #if CHAR_MIN != 0
638 		if (*cp < 0)
639 			break;		/* garbage input */
640 #endif
641 		len = snprintf(&rval[roff], sizeof(rval) - roff, "%u;", *cp);
642 		if (len < 0 || (unsigned)len >= sizeof(rval) - roff)
643 			break;		/* insufficient space for output */
644 		roff += len;
645 		if (*cp == CHAR_MAX)
646 			break;		/* special termination */
647 	}
648 
649 	/* Truncate at the last successfully snprintf()ed semicolon. */
650 	if (roff != 0)
651 		rval[roff - 1] = '\0';
652 
653 	return (&rval[0]);
654 }
655 
656 /*
657  * keyword value lookup helper for values accessible via localeconv()
658  */
659 char *
660 kwval_lconv(int id)
661 {
662 	struct lconv *lc;
663 	char *rval;
664 
665 	rval = NULL;
666 	lc = localeconv();
667 	switch (id) {
668 		case KW_GROUPING:
669 			rval = format_grouping(lc->grouping);
670 			break;
671 		case KW_INT_CURR_SYMBOL:
672 			rval = lc->int_curr_symbol;
673 			break;
674 		case KW_CURRENCY_SYMBOL:
675 			rval = lc->currency_symbol;
676 			break;
677 		case KW_MON_DECIMAL_POINT:
678 			rval = lc->mon_decimal_point;
679 			break;
680 		case KW_MON_THOUSANDS_SEP:
681 			rval = lc->mon_thousands_sep;
682 			break;
683 		case KW_MON_GROUPING:
684 			rval = format_grouping(lc->mon_grouping);
685 			break;
686 		case KW_POSITIVE_SIGN:
687 			rval = lc->positive_sign;
688 			break;
689 		case KW_NEGATIVE_SIGN:
690 			rval = lc->negative_sign;
691 			break;
692 		case KW_INT_FRAC_DIGITS:
693 			rval = &(lc->int_frac_digits);
694 			break;
695 		case KW_FRAC_DIGITS:
696 			rval = &(lc->frac_digits);
697 			break;
698 		case KW_P_CS_PRECEDES:
699 			rval = &(lc->p_cs_precedes);
700 			break;
701 		case KW_P_SEP_BY_SPACE:
702 			rval = &(lc->p_sep_by_space);
703 			break;
704 		case KW_N_CS_PRECEDES:
705 			rval = &(lc->n_cs_precedes);
706 			break;
707 		case KW_N_SEP_BY_SPACE:
708 			rval = &(lc->n_sep_by_space);
709 			break;
710 		case KW_P_SIGN_POSN:
711 			rval = &(lc->p_sign_posn);
712 			break;
713 		case KW_N_SIGN_POSN:
714 			rval = &(lc->n_sign_posn);
715 			break;
716 		case KW_INT_P_CS_PRECEDES:
717 			rval = &(lc->int_p_cs_precedes);
718 			break;
719 		case KW_INT_P_SEP_BY_SPACE:
720 			rval = &(lc->int_p_sep_by_space);
721 			break;
722 		case KW_INT_N_CS_PRECEDES:
723 			rval = &(lc->int_n_cs_precedes);
724 			break;
725 		case KW_INT_N_SEP_BY_SPACE:
726 			rval = &(lc->int_n_sep_by_space);
727 			break;
728 		case KW_INT_P_SIGN_POSN:
729 			rval = &(lc->int_p_sign_posn);
730 			break;
731 		case KW_INT_N_SIGN_POSN:
732 			rval = &(lc->int_n_sign_posn);
733 			break;
734 		default:
735 			break;
736 	}
737 	return (rval);
738 }
739 
740 /*
741  * keyword value lookup helper for LC_TIME keywords not accessible
742  * via nl_langinfo() or localeconv()
743  */
744 static char *
745 kwval_ltime(int id)
746 {
747 	char *rval;
748 	struct sbuf *kwsbuf;
749 	nl_item i, s_item = 0, e_item = 0;
750 
751 	switch (id) {
752 	case KW_TIME_DAY:
753 		s_item = DAY_1;
754 		e_item = DAY_7;
755 		break;
756 	case KW_TIME_ABDAY:
757 		s_item = ABDAY_1;
758 		e_item = ABDAY_7;
759 		break;
760 	case KW_TIME_MON:
761 		s_item = MON_1;
762 		e_item = MON_12;
763 		break;
764 	case KW_TIME_ABMON:
765 		s_item = ABMON_1;
766 		e_item = ABMON_12;
767 		break;
768 	case KW_TIME_AM_PM:
769 		if (asprintf(&rval, "%s;%s",
770 		    nl_langinfo(AM_STR),
771 		    nl_langinfo(PM_STR)) == -1)
772 			err(1, "asprintf");
773 		return (rval);
774 	}
775 
776 	kwsbuf = sbuf_new_auto();
777 	if (kwsbuf == NULL)
778 		err(1, "sbuf");
779 	for (i = s_item; i <= e_item; i++) {
780 		(void) sbuf_cat(kwsbuf, nl_langinfo(i));
781 		if (i != e_item)
782 			(void) sbuf_cat(kwsbuf, ";");
783 	}
784 	(void) sbuf_finish(kwsbuf);
785 	rval = strdup(sbuf_data(kwsbuf));
786 	if (rval == NULL)
787 		err(1, "strdup");
788 	sbuf_delete(kwsbuf);
789 	return (rval);
790 }
791 
792 /*
793  * keyword value and properties lookup
794  */
795 int
796 kwval_lookup(const char *kwname, char **kwval, int *cat, int *type, int *alloc)
797 {
798 	int	rval;
799 	size_t	i;
800 	static char nastr[3] = "-1";
801 
802 	rval = 0;
803 	*alloc = 0;
804 	for (i = 0; i < NKWINFO; i++) {
805 		if (strcasecmp(kwname, kwinfo[i].name) == 0) {
806 			rval = 1;
807 			*cat = kwinfo[i].catid;
808 			*type = kwinfo[i].type;
809 			switch (kwinfo[i].source) {
810 			case SRC_LINFO:
811 				*kwval = nl_langinfo(kwinfo[i].value_ref);
812 				break;
813 			case SRC_LCONV:
814 				*kwval = kwval_lconv(kwinfo[i].value_ref);
815 				/*
816 				 * XXX This check will need to be modified
817 				 * if/when localeconv() is fixed (PR172215).
818 				 */
819 				if (**kwval == CHAR_MAX) {
820 					if (*type == TYPE_NUM)
821 						*type = TYPE_UNQ;
822 					*kwval = nastr;
823 				}
824 				break;
825 			case SRC_LTIME:
826 				*kwval = kwval_ltime(kwinfo[i].value_ref);
827 				*alloc = 1;
828 				break;
829 			}
830 			break;
831 		}
832 	}
833 
834 	return (rval);
835 }
836 
837 /*
838  * Show details about requested keyword according to '-k' and/or '-c'
839  * command line options specified.
840  */
841 int
842 showdetails(const char *kw)
843 {
844 	int	type, cat, tmpval, alloc;
845 	char	*kwval;
846 
847 	if (kwval_lookup(kw, &kwval, &cat, &type, &alloc) == 0) {
848 		/* Invalid keyword specified */
849 		fprintf(stderr, "Unknown keyword: `%s'\n", kw);
850 		return (1);
851 	}
852 
853 	if (prt_categories) {
854 		if (prt_keywords)
855 			printf("%-20s ", lookup_localecat(cat));
856 		else
857 			printf("%-20s\t%s\n", kw, lookup_localecat(cat));
858 	}
859 
860 	if (prt_keywords) {
861 		switch (type) {
862 		case TYPE_NUM:
863 			tmpval = (char)*kwval;
864 			printf("%s=%d\n", kw, tmpval);
865 			break;
866 		case TYPE_STR:
867 			printf("%s=\"%s\"\n", kw, kwval);
868 			break;
869 		case TYPE_UNQ:
870 			printf("%s=%s\n", kw, kwval);
871 			break;
872 		}
873 	}
874 
875 	if (!prt_categories && !prt_keywords) {
876 		switch (type) {
877 		case TYPE_NUM:
878 			tmpval = (char)*kwval;
879 			printf("%d\n", tmpval);
880 			break;
881 		case TYPE_STR:
882 		case TYPE_UNQ:
883 			printf("%s\n", kwval);
884 			break;
885 		}
886 	}
887 
888 	if (alloc)
889 		free(kwval);
890 
891 	return (0);
892 }
893 
894 /*
895  * Convert locale category id into string
896  */
897 const char *
898 lookup_localecat(int cat)
899 {
900 	size_t	i;
901 
902 	for (i = 0; i < NLCINFO; i++)
903 		if (lcinfo[i].id == cat) {
904 			return (lcinfo[i].name);
905 		}
906 	return ("UNKNOWN");
907 }
908 
909 /*
910  * Show list of keywords
911  */
912 void
913 showkeywordslist(char *substring)
914 {
915 	size_t	i;
916 
917 #define	FMT "%-20s %-12s %-7s %-20s\n"
918 
919 	if (substring == NULL)
920 		printf("List of available keywords\n\n");
921 	else
922 		printf("List of available keywords starting with '%s'\n\n",
923 		    substring);
924 	printf(FMT, "Keyword", "Category", "Type", "Comment");
925 	printf("-------------------- ------------ ------- --------------------\n");
926 	for (i = 0; i < NKWINFO; i++) {
927 		if (substring != NULL) {
928 			if (strncmp(kwinfo[i].name, substring,
929 			    strlen(substring)) != 0)
930 				continue;
931 		}
932 		printf(FMT,
933 			kwinfo[i].name,
934 			lookup_localecat(kwinfo[i].catid),
935 			(kwinfo[i].type == TYPE_NUM) ? "number" : "string",
936 			kwinfo[i].comment);
937 	}
938 }
939