xref: /freebsd/usr.bin/locale/locale.c (revision f126890ac5386406dadf7c4cfa9566cbb56537c5)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
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 
29 /*
30  * XXX: implement missing era_* (LC_TIME) keywords (require libc &
31  *	nl_langinfo(3) extensions)
32  *
33  * XXX: correctly handle reserved 'charmap' keyword and '-m' option (require
34  *	localedef(1) implementation).  Currently it's handled via
35  *	nl_langinfo(CODESET).
36  */
37 
38 #include <sys/param.h>
39 #include <sys/types.h>
40 #include <sys/sbuf.h>
41 
42 #include <dirent.h>
43 #include <err.h>
44 #include <limits.h>
45 #include <locale.h>
46 #include <langinfo.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <stringlist.h>
51 #include <unistd.h>
52 #include "setlocale.h"
53 
54 /* Local prototypes */
55 char	*format_grouping(char *);
56 void	init_locales_list(void);
57 void	list_charmaps(void);
58 void	list_locales(void);
59 const char *lookup_localecat(int);
60 char	*kwval_lconv(int);
61 int	kwval_lookup(const char *, char **, int *, int *, int *);
62 int	showdetails(const char *);
63 void	showkeywordslist(char *substring);
64 void	showlocale(void);
65 void	usage(void);
66 
67 /* Global variables */
68 static StringList *locales = NULL;
69 
70 static int	all_locales = 0;
71 static int	all_charmaps = 0;
72 static int	prt_categories = 0;
73 static int	prt_keywords = 0;
74 
75 static const struct _lcinfo {
76 	const char	*name;
77 	int		id;
78 } lcinfo [] = {
79 	{ "LC_CTYPE",		LC_CTYPE },
80 	{ "LC_COLLATE",		LC_COLLATE },
81 	{ "LC_TIME",		LC_TIME },
82 	{ "LC_NUMERIC",		LC_NUMERIC },
83 	{ "LC_MONETARY",	LC_MONETARY },
84 	{ "LC_MESSAGES",	LC_MESSAGES }
85 };
86 #define	NLCINFO nitems(lcinfo)
87 
88 /* ids for values not referenced by nl_langinfo() */
89 enum {
90 	KW_GROUPING,
91 	KW_INT_CURR_SYMBOL,
92 	KW_CURRENCY_SYMBOL,
93 	KW_MON_DECIMAL_POINT,
94 	KW_MON_THOUSANDS_SEP,
95 	KW_MON_GROUPING,
96 	KW_POSITIVE_SIGN,
97 	KW_NEGATIVE_SIGN,
98 	KW_INT_FRAC_DIGITS,
99 	KW_FRAC_DIGITS,
100 	KW_P_CS_PRECEDES,
101 	KW_P_SEP_BY_SPACE,
102 	KW_N_CS_PRECEDES,
103 	KW_N_SEP_BY_SPACE,
104 	KW_P_SIGN_POSN,
105 	KW_N_SIGN_POSN,
106 	KW_INT_P_CS_PRECEDES,
107 	KW_INT_P_SEP_BY_SPACE,
108 	KW_INT_N_CS_PRECEDES,
109 	KW_INT_N_SEP_BY_SPACE,
110 	KW_INT_P_SIGN_POSN,
111 	KW_INT_N_SIGN_POSN,
112 	KW_TIME_DAY,
113 	KW_TIME_ABDAY,
114 	KW_TIME_MON,
115 	KW_TIME_ABMON,
116 	KW_TIME_AM_PM
117 };
118 
119 enum {
120 	TYPE_NUM,
121 	TYPE_STR,
122 	TYPE_UNQ
123 };
124 
125 enum {
126 	SRC_LINFO,
127 	SRC_LCONV,
128 	SRC_LTIME
129 };
130 
131 static const struct _kwinfo {
132 	const char	*name;
133 	int		type;
134 	int		catid;		/* LC_* */
135 	int		source;
136 	int		value_ref;
137 	const char	*comment;
138 } kwinfo [] = {
139 	{ "charmap",		TYPE_STR, LC_CTYPE,	SRC_LINFO,
140 	  CODESET, "" },					/* hack */
141 
142 	/* LC_MONETARY - POSIX */
143 	{ "int_curr_symbol",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
144 	  KW_INT_CURR_SYMBOL, "" },
145 	{ "currency_symbol",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
146 	  KW_CURRENCY_SYMBOL, "" },
147 	{ "mon_decimal_point",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
148 	  KW_MON_DECIMAL_POINT, "" },
149 	{ "mon_thousands_sep",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
150 	  KW_MON_THOUSANDS_SEP, "" },
151 	{ "mon_grouping",	TYPE_UNQ, LC_MONETARY,	SRC_LCONV,
152 	  KW_MON_GROUPING, "" },
153 	{ "positive_sign",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
154 	  KW_POSITIVE_SIGN, "" },
155 	{ "negative_sign",	TYPE_STR, LC_MONETARY,	SRC_LCONV,
156 	  KW_NEGATIVE_SIGN, "" },
157 	{ "int_frac_digits",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
158 	  KW_INT_FRAC_DIGITS, "" },
159 	{ "frac_digits",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
160 	  KW_FRAC_DIGITS, "" },
161 	{ "p_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
162 	  KW_P_CS_PRECEDES, "" },
163 	{ "p_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
164 	  KW_P_SEP_BY_SPACE, "" },
165 	{ "n_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
166 	  KW_N_CS_PRECEDES, "" },
167 	{ "n_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
168 	  KW_N_SEP_BY_SPACE, "" },
169 	{ "p_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
170 	  KW_P_SIGN_POSN, "" },
171 	{ "n_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
172 	  KW_N_SIGN_POSN, "" },
173 	{ "int_p_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
174 	  KW_INT_P_CS_PRECEDES, "" },
175 	{ "int_p_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
176 	  KW_INT_P_SEP_BY_SPACE, "" },
177 	{ "int_n_cs_precedes",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
178 	  KW_INT_N_CS_PRECEDES, "" },
179 	{ "int_n_sep_by_space",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
180 	  KW_INT_N_SEP_BY_SPACE, "" },
181 	{ "int_p_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
182 	  KW_INT_P_SIGN_POSN, "" },
183 	{ "int_n_sign_posn",	TYPE_NUM, LC_MONETARY,	SRC_LCONV,
184 	  KW_INT_N_SIGN_POSN, "" },
185 
186 	/* LC_NUMERIC - POSIX */
187 	{ "decimal_point",	TYPE_STR, LC_NUMERIC,	SRC_LINFO,
188 	  RADIXCHAR, "" },
189 	{ "thousands_sep",	TYPE_STR, LC_NUMERIC,	SRC_LINFO,
190 	  THOUSEP, "" },
191 	{ "grouping",		TYPE_UNQ, LC_NUMERIC,	SRC_LCONV,
192 	  KW_GROUPING, "" },
193 	/* LC_NUMERIC - local additions */
194 	{ "radixchar",		TYPE_STR, LC_NUMERIC,	SRC_LINFO,
195 	  RADIXCHAR, "Same as decimal_point (FreeBSD only)" },	/* compat */
196 	{ "thousep",		TYPE_STR, LC_NUMERIC,	SRC_LINFO,
197 	  THOUSEP, "Same as thousands_sep (FreeBSD only)" },	/* compat */
198 
199 	/* LC_TIME - POSIX */
200 	{ "abday",		TYPE_STR, LC_TIME,	SRC_LTIME,
201 	  KW_TIME_ABDAY, "" },
202 	{ "day",		TYPE_STR, LC_TIME,	SRC_LTIME,
203 	  KW_TIME_DAY, "" },
204 	{ "abmon",		TYPE_STR, LC_TIME,	SRC_LTIME,
205 	  KW_TIME_ABMON, "" },
206 	{ "mon",		TYPE_STR, LC_TIME,	SRC_LTIME,
207 	  KW_TIME_MON, "" },
208 	{ "d_t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
209 	  D_T_FMT, "" },
210 	{ "d_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
211 	  D_FMT, "" },
212 	{ "t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
213 	  T_FMT, "" },
214 	{ "am_pm",		TYPE_STR, LC_TIME,	SRC_LTIME,
215 	  KW_TIME_AM_PM, "" },
216 	{ "t_fmt_ampm",		TYPE_STR, LC_TIME,	SRC_LINFO,
217 	  T_FMT_AMPM, "" },
218 	{ "era",		TYPE_UNQ, LC_TIME,	SRC_LINFO,
219 	  ERA, "(unavailable)" },
220 	{ "era_d_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
221 	  ERA_D_FMT, "(unavailable)" },
222 	{ "era_d_t_fmt",	TYPE_STR, LC_TIME,	SRC_LINFO,
223 	  ERA_D_T_FMT, "(unavailable)" },
224 	{ "era_t_fmt",		TYPE_STR, LC_TIME,	SRC_LINFO,
225 	  ERA_T_FMT, "(unavailable)" },
226 	{ "alt_digits",		TYPE_UNQ, LC_TIME,	SRC_LINFO,
227 	  ALT_DIGITS, "" },
228 	/* LC_TIME - local additions */
229 	{ "abday_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
230 	  ABDAY_1, "(FreeBSD only)" },
231 	{ "abday_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
232 	  ABDAY_2, "(FreeBSD only)" },
233 	{ "abday_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
234 	  ABDAY_3, "(FreeBSD only)" },
235 	{ "abday_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
236 	  ABDAY_4, "(FreeBSD only)" },
237 	{ "abday_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
238 	  ABDAY_5, "(FreeBSD only)" },
239 	{ "abday_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
240 	  ABDAY_6, "(FreeBSD only)" },
241 	{ "abday_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
242 	  ABDAY_7, "(FreeBSD only)" },
243 	{ "day_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
244 	  DAY_1, "(FreeBSD only)" },
245 	{ "day_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
246 	  DAY_2, "(FreeBSD only)" },
247 	{ "day_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
248 	  DAY_3, "(FreeBSD only)" },
249 	{ "day_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
250 	  DAY_4, "(FreeBSD only)" },
251 	{ "day_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
252 	  DAY_5, "(FreeBSD only)" },
253 	{ "day_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
254 	  DAY_6, "(FreeBSD only)" },
255 	{ "day_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
256 	  DAY_7, "(FreeBSD only)" },
257 	{ "abmon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
258 	  ABMON_1, "(FreeBSD only)" },
259 	{ "abmon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
260 	  ABMON_2, "(FreeBSD only)" },
261 	{ "abmon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
262 	  ABMON_3, "(FreeBSD only)" },
263 	{ "abmon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
264 	  ABMON_4, "(FreeBSD only)" },
265 	{ "abmon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
266 	  ABMON_5, "(FreeBSD only)" },
267 	{ "abmon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
268 	  ABMON_6, "(FreeBSD only)" },
269 	{ "abmon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
270 	  ABMON_7, "(FreeBSD only)" },
271 	{ "abmon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
272 	  ABMON_8, "(FreeBSD only)" },
273 	{ "abmon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
274 	  ABMON_9, "(FreeBSD only)" },
275 	{ "abmon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
276 	  ABMON_10, "(FreeBSD only)" },
277 	{ "abmon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
278 	  ABMON_11, "(FreeBSD only)" },
279 	{ "abmon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
280 	  ABMON_12, "(FreeBSD only)" },
281 	{ "mon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
282 	  MON_1, "(FreeBSD only)" },
283 	{ "mon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
284 	  MON_2, "(FreeBSD only)" },
285 	{ "mon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
286 	  MON_3, "(FreeBSD only)" },
287 	{ "mon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
288 	  MON_4, "(FreeBSD only)" },
289 	{ "mon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
290 	  MON_5, "(FreeBSD only)" },
291 	{ "mon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
292 	  MON_6, "(FreeBSD only)" },
293 	{ "mon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
294 	  MON_7, "(FreeBSD only)" },
295 	{ "mon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
296 	  MON_8, "(FreeBSD only)" },
297 	{ "mon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
298 	  MON_9, "(FreeBSD only)" },
299 	{ "mon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
300 	  MON_10, "(FreeBSD only)" },
301 	{ "mon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
302 	  MON_11, "(FreeBSD only)" },
303 	{ "mon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
304 	  MON_12, "(FreeBSD only)" },
305 	{ "altmon_1",		TYPE_STR, LC_TIME,	SRC_LINFO,
306 	  ALTMON_1, "(FreeBSD only)" },
307 	{ "altmon_2",		TYPE_STR, LC_TIME,	SRC_LINFO,
308 	  ALTMON_2, "(FreeBSD only)" },
309 	{ "altmon_3",		TYPE_STR, LC_TIME,	SRC_LINFO,
310 	  ALTMON_3, "(FreeBSD only)" },
311 	{ "altmon_4",		TYPE_STR, LC_TIME,	SRC_LINFO,
312 	  ALTMON_4, "(FreeBSD only)" },
313 	{ "altmon_5",		TYPE_STR, LC_TIME,	SRC_LINFO,
314 	  ALTMON_5, "(FreeBSD only)" },
315 	{ "altmon_6",		TYPE_STR, LC_TIME,	SRC_LINFO,
316 	  ALTMON_6, "(FreeBSD only)" },
317 	{ "altmon_7",		TYPE_STR, LC_TIME,	SRC_LINFO,
318 	  ALTMON_7, "(FreeBSD only)" },
319 	{ "altmon_8",		TYPE_STR, LC_TIME,	SRC_LINFO,
320 	  ALTMON_8, "(FreeBSD only)" },
321 	{ "altmon_9",		TYPE_STR, LC_TIME,	SRC_LINFO,
322 	  ALTMON_9, "(FreeBSD only)" },
323 	{ "altmon_10",		TYPE_STR, LC_TIME,	SRC_LINFO,
324 	  ALTMON_10, "(FreeBSD only)" },
325 	{ "altmon_11",		TYPE_STR, LC_TIME,	SRC_LINFO,
326 	  ALTMON_11, "(FreeBSD only)" },
327 	{ "altmon_12",		TYPE_STR, LC_TIME,	SRC_LINFO,
328 	  ALTMON_12, "(FreeBSD only)" },
329 	{ "am_str",		TYPE_STR, LC_TIME,	SRC_LINFO,
330 	  AM_STR, "(FreeBSD only)" },
331 	{ "pm_str",		TYPE_STR, LC_TIME,	SRC_LINFO,
332 	  PM_STR, "(FreeBSD only)" },
333 	{ "d_md_order",		TYPE_STR, LC_TIME,	SRC_LINFO,
334 	  D_MD_ORDER, "(FreeBSD only)" },			/* local */
335 
336 	/* LC_MESSAGES - POSIX */
337 	{ "yesexpr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
338 	  YESEXPR, "" },
339 	{ "noexpr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
340 	  NOEXPR, "" },
341 	/* LC_MESSAGES - local additions */
342 	{ "yesstr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
343 	  YESSTR, "(POSIX legacy)" },				/* compat */
344 	{ "nostr",		TYPE_STR, LC_MESSAGES, SRC_LINFO,
345 	  NOSTR, "(POSIX legacy)" }				/* compat */
346 
347 };
348 #define	NKWINFO (nitems(kwinfo))
349 
350 static const char *boguslocales[] = { "UTF-8" };
351 #define	NBOGUS	(nitems(boguslocales))
352 
353 int
354 main(int argc, char *argv[])
355 {
356 	int	ch;
357 	int	tmp;
358 
359 	while ((ch = getopt(argc, argv, "ackms:")) != -1) {
360 		switch (ch) {
361 		case 'a':
362 			all_locales = 1;
363 			break;
364 		case 'c':
365 			prt_categories = 1;
366 			break;
367 		case 'k':
368 			prt_keywords = 1;
369 			break;
370 		case 'm':
371 			all_charmaps = 1;
372 			break;
373 		default:
374 			usage();
375 		}
376 	}
377 	argc -= optind;
378 	argv += optind;
379 
380 	/* validate arguments */
381 	if (all_locales && all_charmaps)
382 		usage();
383 	if ((all_locales || all_charmaps) && argc > 0)
384 		usage();
385 	if ((all_locales || all_charmaps) && (prt_categories || prt_keywords))
386 		usage();
387 
388 	/* process '-a' */
389 	if (all_locales) {
390 		list_locales();
391 		exit(0);
392 	}
393 
394 	/* process '-m' */
395 	if (all_charmaps) {
396 		list_charmaps();
397 		exit(0);
398 	}
399 
400 	/* check for special case '-k list' */
401 	tmp = 0;
402 	if (prt_keywords && argc > 0)
403 		while (tmp < argc)
404 			if (strcasecmp(argv[tmp++], "list") == 0) {
405 				showkeywordslist(argv[tmp]);
406 				exit(0);
407 			}
408 
409 	/* process '-c', '-k', or command line arguments. */
410 	if (prt_categories || prt_keywords || argc > 0) {
411 		if (prt_keywords || argc > 0)
412 			setlocale(LC_ALL, "");
413 		if (argc > 0) {
414 			while (argc > 0) {
415 				if (showdetails(*argv) != 0)
416 					exit(EXIT_FAILURE);
417 				argv++;
418 				argc--;
419 			}
420 		} else {
421 			uint i;
422 			for (i = 0; i < nitems(kwinfo); i++)
423 				showdetails(kwinfo[i].name);
424 		}
425 		exit(0);
426 	}
427 
428 	/* no arguments, show current locale state */
429 	showlocale();
430 
431 	return (0);
432 }
433 
434 void
435 usage(void)
436 {
437 	printf("Usage: locale [ -a | -m ]\n"
438 	       "       locale -k list [prefix]\n"
439 	       "       locale [ -ck ] [keyword ...]\n");
440 	exit(1);
441 }
442 
443 /*
444  * Output information about all available locales
445  *
446  * XXX actually output of this function does not guarantee that locale
447  *     is really available to application, since it can be broken or
448  *     inconsistent thus setlocale() will fail.  Maybe add '-V' function to
449  *     also validate these locales?
450  */
451 void
452 list_locales(void)
453 {
454 	size_t i;
455 
456 	init_locales_list();
457 	for (i = 0; i < locales->sl_cur; i++) {
458 		printf("%s\n", locales->sl_str[i]);
459 	}
460 }
461 
462 /*
463  * qsort() helper function
464  */
465 static int
466 scmp(const void *s1, const void *s2)
467 {
468 	return strcmp(*(const char * const *)s1, *(const char * const *)s2);
469 }
470 
471 /*
472  * Output information about all available charmaps
473  *
474  * XXX this function is doing a task in hackish way, i.e. by scaning
475  *     list of locales, spliting their codeset part and building list of
476  *     them.
477  */
478 void
479 list_charmaps(void)
480 {
481 	size_t i;
482 	char *s, *cs;
483 	StringList *charmaps;
484 
485 	/* initialize StringList */
486 	charmaps = sl_init();
487 	if (charmaps == NULL)
488 		err(1, "could not allocate memory");
489 
490 	/* fetch locales list */
491 	init_locales_list();
492 
493 	/* split codesets and build their list */
494 	for (i = 0; i < locales->sl_cur; i++) {
495 		s = locales->sl_str[i];
496 		if ((cs = strchr(s, '.')) != NULL) {
497 			cs++;
498 			if (sl_find(charmaps, cs) == NULL)
499 				sl_add(charmaps, cs);
500 		}
501 	}
502 
503 	/* add US-ASCII, if not yet added */
504 	if (sl_find(charmaps, "US-ASCII") == NULL)
505 		sl_add(charmaps, strdup("US-ASCII"));
506 
507 	/* sort the list */
508 	qsort(charmaps->sl_str, charmaps->sl_cur, sizeof(char *), scmp);
509 
510 	/* print results */
511 	for (i = 0; i < charmaps->sl_cur; i++) {
512 		printf("%s\n", charmaps->sl_str[i]);
513 	}
514 }
515 
516 /*
517  * Retrieve sorted list of system locales (or user locales, if PATH_LOCALE
518  * environment variable is set)
519  */
520 void
521 init_locales_list(void)
522 {
523 	DIR *dirp;
524 	struct dirent *dp;
525 	size_t i;
526 	int bogus;
527 
528 	/* why call this function twice ? */
529 	if (locales != NULL)
530 		return;
531 
532 	/* initialize StringList */
533 	locales = sl_init();
534 	if (locales == NULL)
535 		err(1, "could not allocate memory");
536 
537 	/* get actual locales directory name */
538 	if (__detect_path_locale() != 0)
539 		err(1, "unable to find locales storage");
540 
541 	/* open locales directory */
542 	dirp = opendir(_PathLocale);
543 	if (dirp == NULL)
544 		err(1, "could not open directory '%s'", _PathLocale);
545 
546 	/* scan directory and store its contents except "." and ".." */
547 	while ((dp = readdir(dirp)) != NULL) {
548 		if (*(dp->d_name) == '.')
549 			continue;		/* exclude "." and ".." */
550 		for (bogus = i = 0; i < NBOGUS; i++)
551 			if (strncmp(dp->d_name, boguslocales[i],
552 			    strlen(boguslocales[i])) == 0)
553 				bogus = 1;
554 		if (!bogus)
555 			sl_add(locales, strdup(dp->d_name));
556 	}
557 	closedir(dirp);
558 
559 	/* make sure that 'POSIX' and 'C' locales are present in the list.
560 	 * POSIX 1003.1-2001 requires presence of 'POSIX' name only here, but
561 	 * we also list 'C' for constistency
562 	 */
563 	if (sl_find(locales, "POSIX") == NULL)
564 		sl_add(locales, strdup("POSIX"));
565 
566 	if (sl_find(locales, "C") == NULL)
567 		sl_add(locales, strdup("C"));
568 
569 	/* make output nicer, sort the list */
570 	qsort(locales->sl_str, locales->sl_cur, sizeof(char *), scmp);
571 }
572 
573 /*
574  * Show current locale status, depending on environment variables
575  */
576 void
577 showlocale(void)
578 {
579 	size_t	i;
580 	const char *lang, *vval, *eval;
581 
582 	setlocale(LC_ALL, "");
583 
584 	lang = getenv("LANG");
585 	if (lang == NULL) {
586 		lang = "";
587 	}
588 	printf("LANG=%s\n", lang);
589 	/* XXX: if LANG is null, then set it to "C" to get implied values? */
590 
591 	for (i = 0; i < NLCINFO; i++) {
592 		vval = setlocale(lcinfo[i].id, NULL);
593 		eval = getenv(lcinfo[i].name);
594 		if (eval != NULL && !strcmp(eval, vval)
595 				&& strcmp(lang, vval)) {
596 			/*
597 			 * Appropriate environment variable set, its value
598 			 * is valid and not overridden by LC_ALL
599 			 *
600 			 * XXX: possible side effect: if both LANG and
601 			 * overridden environment variable are set into same
602 			 * value, then it'll be assumed as 'implied'
603 			 */
604 			printf("%s=%s\n", lcinfo[i].name, vval);
605 		} else {
606 			printf("%s=\"%s\"\n", lcinfo[i].name, vval);
607 		}
608 	}
609 
610 	vval = getenv("LC_ALL");
611 	if (vval == NULL) {
612 		vval = "";
613 	}
614 	printf("LC_ALL=%s\n", vval);
615 }
616 
617 char *
618 format_grouping(char *binary)
619 {
620 	static char rval[64];
621 	const char *cp;
622 	size_t roff;
623 	int len;
624 
625 	/*
626 	 * XXX This check will need to be modified if/when localeconv() is
627 	 * fixed (PR172215).
628 	 */
629 	if (*binary == CHAR_MAX)
630 		return (binary);
631 
632 	rval[0] = '\0';
633 	roff = 0;
634 	for (cp = binary; *cp != '\0'; ++cp) {
635 #if CHAR_MIN != 0
636 		if (*cp < 0)
637 			break;		/* garbage input */
638 #endif
639 		len = snprintf(&rval[roff], sizeof(rval) - roff, "%u;", *cp);
640 		if (len < 0 || (unsigned)len >= sizeof(rval) - roff)
641 			break;		/* insufficient space for output */
642 		roff += len;
643 		if (*cp == CHAR_MAX)
644 			break;		/* special termination */
645 	}
646 
647 	/* Truncate at the last successfully snprintf()ed semicolon. */
648 	if (roff != 0)
649 		rval[roff - 1] = '\0';
650 
651 	return (&rval[0]);
652 }
653 
654 /*
655  * keyword value lookup helper for values accessible via localeconv()
656  */
657 char *
658 kwval_lconv(int id)
659 {
660 	struct lconv *lc;
661 	char *rval;
662 
663 	rval = NULL;
664 	lc = localeconv();
665 	switch (id) {
666 		case KW_GROUPING:
667 			rval = format_grouping(lc->grouping);
668 			break;
669 		case KW_INT_CURR_SYMBOL:
670 			rval = lc->int_curr_symbol;
671 			break;
672 		case KW_CURRENCY_SYMBOL:
673 			rval = lc->currency_symbol;
674 			break;
675 		case KW_MON_DECIMAL_POINT:
676 			rval = lc->mon_decimal_point;
677 			break;
678 		case KW_MON_THOUSANDS_SEP:
679 			rval = lc->mon_thousands_sep;
680 			break;
681 		case KW_MON_GROUPING:
682 			rval = format_grouping(lc->mon_grouping);
683 			break;
684 		case KW_POSITIVE_SIGN:
685 			rval = lc->positive_sign;
686 			break;
687 		case KW_NEGATIVE_SIGN:
688 			rval = lc->negative_sign;
689 			break;
690 		case KW_INT_FRAC_DIGITS:
691 			rval = &(lc->int_frac_digits);
692 			break;
693 		case KW_FRAC_DIGITS:
694 			rval = &(lc->frac_digits);
695 			break;
696 		case KW_P_CS_PRECEDES:
697 			rval = &(lc->p_cs_precedes);
698 			break;
699 		case KW_P_SEP_BY_SPACE:
700 			rval = &(lc->p_sep_by_space);
701 			break;
702 		case KW_N_CS_PRECEDES:
703 			rval = &(lc->n_cs_precedes);
704 			break;
705 		case KW_N_SEP_BY_SPACE:
706 			rval = &(lc->n_sep_by_space);
707 			break;
708 		case KW_P_SIGN_POSN:
709 			rval = &(lc->p_sign_posn);
710 			break;
711 		case KW_N_SIGN_POSN:
712 			rval = &(lc->n_sign_posn);
713 			break;
714 		case KW_INT_P_CS_PRECEDES:
715 			rval = &(lc->int_p_cs_precedes);
716 			break;
717 		case KW_INT_P_SEP_BY_SPACE:
718 			rval = &(lc->int_p_sep_by_space);
719 			break;
720 		case KW_INT_N_CS_PRECEDES:
721 			rval = &(lc->int_n_cs_precedes);
722 			break;
723 		case KW_INT_N_SEP_BY_SPACE:
724 			rval = &(lc->int_n_sep_by_space);
725 			break;
726 		case KW_INT_P_SIGN_POSN:
727 			rval = &(lc->int_p_sign_posn);
728 			break;
729 		case KW_INT_N_SIGN_POSN:
730 			rval = &(lc->int_n_sign_posn);
731 			break;
732 		default:
733 			break;
734 	}
735 	return (rval);
736 }
737 
738 /*
739  * keyword value lookup helper for LC_TIME keywords not accessible
740  * via nl_langinfo() or localeconv()
741  */
742 static char *
743 kwval_ltime(int id)
744 {
745 	char *rval;
746 	struct sbuf *kwsbuf;
747 	nl_item i, s_item = 0, e_item = 0;
748 
749 	switch (id) {
750 	case KW_TIME_DAY:
751 		s_item = DAY_1;
752 		e_item = DAY_7;
753 		break;
754 	case KW_TIME_ABDAY:
755 		s_item = ABDAY_1;
756 		e_item = ABDAY_7;
757 		break;
758 	case KW_TIME_MON:
759 		s_item = MON_1;
760 		e_item = MON_12;
761 		break;
762 	case KW_TIME_ABMON:
763 		s_item = ABMON_1;
764 		e_item = ABMON_12;
765 		break;
766 	case KW_TIME_AM_PM:
767 		if (asprintf(&rval, "%s;%s",
768 		    nl_langinfo(AM_STR),
769 		    nl_langinfo(PM_STR)) == -1)
770 			err(1, "asprintf");
771 		return (rval);
772 	}
773 
774 	kwsbuf = sbuf_new_auto();
775 	if (kwsbuf == NULL)
776 		err(1, "sbuf");
777 	for (i = s_item; i <= e_item; i++) {
778 		(void) sbuf_cat(kwsbuf, nl_langinfo(i));
779 		if (i != e_item)
780 			(void) sbuf_cat(kwsbuf, ";");
781 	}
782 	(void) sbuf_finish(kwsbuf);
783 	rval = strdup(sbuf_data(kwsbuf));
784 	if (rval == NULL)
785 		err(1, "strdup");
786 	sbuf_delete(kwsbuf);
787 	return (rval);
788 }
789 
790 /*
791  * keyword value and properties lookup
792  */
793 int
794 kwval_lookup(const char *kwname, char **kwval, int *cat, int *type, int *alloc)
795 {
796 	int	rval;
797 	size_t	i;
798 	static char nastr[3] = "-1";
799 
800 	rval = 0;
801 	*alloc = 0;
802 	for (i = 0; i < NKWINFO; i++) {
803 		if (strcasecmp(kwname, kwinfo[i].name) == 0) {
804 			rval = 1;
805 			*cat = kwinfo[i].catid;
806 			*type = kwinfo[i].type;
807 			switch (kwinfo[i].source) {
808 			case SRC_LINFO:
809 				*kwval = nl_langinfo(kwinfo[i].value_ref);
810 				break;
811 			case SRC_LCONV:
812 				*kwval = kwval_lconv(kwinfo[i].value_ref);
813 				/*
814 				 * XXX This check will need to be modified
815 				 * if/when localeconv() is fixed (PR172215).
816 				 */
817 				if (**kwval == CHAR_MAX) {
818 					if (*type == TYPE_NUM)
819 						*type = TYPE_UNQ;
820 					*kwval = nastr;
821 				}
822 				break;
823 			case SRC_LTIME:
824 				*kwval = kwval_ltime(kwinfo[i].value_ref);
825 				*alloc = 1;
826 				break;
827 			}
828 			break;
829 		}
830 	}
831 
832 	return (rval);
833 }
834 
835 /*
836  * Show details about requested keyword according to '-k' and/or '-c'
837  * command line options specified.
838  */
839 int
840 showdetails(const char *kw)
841 {
842 	int	type, cat, tmpval, alloc;
843 	char	*kwval;
844 
845 	if (kwval_lookup(kw, &kwval, &cat, &type, &alloc) == 0) {
846 		/* Invalid keyword specified */
847 		fprintf(stderr, "Unknown keyword: `%s'\n", kw);
848 		return (1);
849 	}
850 
851 	if (prt_categories) {
852 		if (prt_keywords)
853 			printf("%-20s ", lookup_localecat(cat));
854 		else
855 			printf("%-20s\t%s\n", kw, lookup_localecat(cat));
856 	}
857 
858 	if (prt_keywords) {
859 		switch (type) {
860 		case TYPE_NUM:
861 			tmpval = (char)*kwval;
862 			printf("%s=%d\n", kw, tmpval);
863 			break;
864 		case TYPE_STR:
865 			printf("%s=\"%s\"\n", kw, kwval);
866 			break;
867 		case TYPE_UNQ:
868 			printf("%s=%s\n", kw, kwval);
869 			break;
870 		}
871 	}
872 
873 	if (!prt_categories && !prt_keywords) {
874 		switch (type) {
875 		case TYPE_NUM:
876 			tmpval = (char)*kwval;
877 			printf("%d\n", tmpval);
878 			break;
879 		case TYPE_STR:
880 		case TYPE_UNQ:
881 			printf("%s\n", kwval);
882 			break;
883 		}
884 	}
885 
886 	if (alloc)
887 		free(kwval);
888 
889 	return (0);
890 }
891 
892 /*
893  * Convert locale category id into string
894  */
895 const char *
896 lookup_localecat(int cat)
897 {
898 	size_t	i;
899 
900 	for (i = 0; i < NLCINFO; i++)
901 		if (lcinfo[i].id == cat) {
902 			return (lcinfo[i].name);
903 		}
904 	return ("UNKNOWN");
905 }
906 
907 /*
908  * Show list of keywords
909  */
910 void
911 showkeywordslist(char *substring)
912 {
913 	size_t	i;
914 
915 #define	FMT "%-20s %-12s %-7s %-20s\n"
916 
917 	if (substring == NULL)
918 		printf("List of available keywords\n\n");
919 	else
920 		printf("List of available keywords starting with '%s'\n\n",
921 		    substring);
922 	printf(FMT, "Keyword", "Category", "Type", "Comment");
923 	printf("-------------------- ------------ ------- --------------------\n");
924 	for (i = 0; i < NKWINFO; i++) {
925 		if (substring != NULL) {
926 			if (strncmp(kwinfo[i].name, substring,
927 			    strlen(substring)) != 0)
928 				continue;
929 		}
930 		printf(FMT,
931 			kwinfo[i].name,
932 			lookup_localecat(kwinfo[i].catid),
933 			(kwinfo[i].type == TYPE_NUM) ? "number" : "string",
934 			kwinfo[i].comment);
935 	}
936 }
937