1 /*-------------------------------------------------------------------------
4 * Functions for handling locale-related info
7 * Copyright (c) 1996-2007, PostgreSQL Global Development Group
11 * $PostgreSQL: pgsql/src/port/chklocale.c,v 1.4 2007/10/03 17:16:39 tgl Exp $
13 *-------------------------------------------------------------------------
19 #include "postgres_fe.h"
23 #ifdef HAVE_LANGINFO_H
27 #include "mb/pg_wchar.h"
31 * This table needs to recognize all the CODESET spellings for supported
32 * backend encodings, as well as frontend-only encodings where possible
33 * (the latter case is currently only needed for initdb to recognize
34 * error situations). On Windows, we rely on entries for codepage
37 * Note that we search the table with pg_strcasecmp(), so variant
38 * capitalizations don't need their own entries.
42 enum pg_enc pg_enc_code;
43 const char *system_enc_name;
46 static const struct encoding_match encoding_match_list[] = {
47 {PG_EUC_JP, "EUC-JP"},
49 {PG_EUC_JP, "IBM-eucJP"},
50 {PG_EUC_JP, "sdeckanji"},
51 {PG_EUC_JP, "CP20932"},
53 {PG_EUC_CN, "EUC-CN"},
55 {PG_EUC_CN, "IBM-eucCN"},
56 {PG_EUC_CN, "GB2312"},
57 {PG_EUC_CN, "dechanzi"},
58 {PG_EUC_CN, "CP20936"},
60 {PG_EUC_KR, "EUC-KR"},
62 {PG_EUC_KR, "IBM-eucKR"},
63 {PG_EUC_KR, "deckorean"},
65 {PG_EUC_KR, "CP51949"}, /* or 20949 ? */
67 {PG_EUC_TW, "EUC-TW"},
69 {PG_EUC_TW, "IBM-eucTW"},
70 {PG_EUC_TW, "cns11643"},
71 /* No codepage for EUC-TW ? */
76 {PG_LATIN1, "ISO-8859-1"},
77 {PG_LATIN1, "ISO8859-1"},
78 {PG_LATIN1, "iso88591"},
80 {PG_LATIN2, "ISO-8859-2"},
81 {PG_LATIN2, "ISO8859-2"},
82 {PG_LATIN2, "iso88592"},
84 {PG_LATIN3, "ISO-8859-3"},
85 {PG_LATIN3, "ISO8859-3"},
86 {PG_LATIN3, "iso88593"},
88 {PG_LATIN4, "ISO-8859-4"},
89 {PG_LATIN4, "ISO8859-4"},
90 {PG_LATIN4, "iso88594"},
92 {PG_LATIN5, "ISO-8859-9"},
93 {PG_LATIN5, "ISO8859-9"},
94 {PG_LATIN5, "iso88599"},
96 {PG_LATIN6, "ISO-8859-10"},
97 {PG_LATIN6, "ISO8859-10"},
98 {PG_LATIN6, "iso885910"},
100 {PG_LATIN7, "ISO-8859-13"},
101 {PG_LATIN7, "ISO8859-13"},
102 {PG_LATIN7, "iso885913"},
104 {PG_LATIN8, "ISO-8859-14"},
105 {PG_LATIN8, "ISO8859-14"},
106 {PG_LATIN8, "iso885914"},
108 {PG_LATIN9, "ISO-8859-15"},
109 {PG_LATIN9, "ISO8859-15"},
110 {PG_LATIN9, "iso885915"},
112 {PG_LATIN10, "ISO-8859-16"},
113 {PG_LATIN10, "ISO8859-16"},
114 {PG_LATIN10, "iso885916"},
116 {PG_KOI8R, "KOI8-R"},
117 {PG_KOI8R, "CP20866"},
119 {PG_WIN1252, "CP1252"},
120 {PG_WIN1253, "CP1253"},
121 {PG_WIN1254, "CP1254"},
122 {PG_WIN1255, "CP1255"},
123 {PG_WIN1256, "CP1256"},
124 {PG_WIN1257, "CP1257"},
125 {PG_WIN1258, "CP1258"},
129 {PG_WIN1251, "CP1251"},
130 {PG_WIN866, "CP866"},
132 {PG_ISO_8859_5, "ISO-8859-5"},
133 {PG_ISO_8859_5, "ISO8859-5"},
134 {PG_ISO_8859_5, "iso88595"},
136 {PG_ISO_8859_6, "ISO-8859-6"},
137 {PG_ISO_8859_6, "ISO8859-6"},
138 {PG_ISO_8859_6, "iso88596"},
140 {PG_ISO_8859_7, "ISO-8859-7"},
141 {PG_ISO_8859_7, "ISO8859-7"},
142 {PG_ISO_8859_7, "iso88597"},
144 {PG_ISO_8859_8, "ISO-8859-8"},
145 {PG_ISO_8859_8, "ISO8859-8"},
146 {PG_ISO_8859_8, "iso88598"},
153 {PG_BIG5, "BIG5HKSCS"},
162 {PG_JOHAB, "CP1361"},
164 {PG_GB18030, "GB18030"},
165 {PG_GB18030, "CP54936"},
167 {PG_SHIFT_JIS_2004, "SJIS_2004"},
169 {PG_SQL_ASCII, NULL} /* end marker */
174 * On Windows, use CP<codepage number> instead of the nl_langinfo() result
177 win32_langinfo(const char *ctype)
184 * Locale format on Win32 is <Language>_<Country>.<CodePage> .
185 * For example, English_USA.1252.
187 codepage = strrchr(ctype, '.');
191 ln = strlen(codepage);
193 sprintf(r, "CP%s", codepage);
199 #if (defined(HAVE_LANGINFO_H) && defined(CODESET)) || defined(WIN32)
202 * Given a setting for LC_CTYPE, return the Postgres ID of the associated
203 * encoding, if we can determine it.
205 * Pass in NULL to get the encoding for the current locale setting.
207 * If the result is PG_SQL_ASCII, callers should treat it as being compatible
208 * with any desired encoding. We return this if the locale is C/POSIX or we
209 * can't determine the encoding.
212 pg_get_encoding_from_locale(const char *ctype)
217 /* Get the CODESET property, and also LC_CTYPE if not passed in */
223 save = setlocale(LC_CTYPE, NULL);
225 return PG_SQL_ASCII; /* setlocale() broken? */
226 /* must copy result, or it might change after setlocale */
229 return PG_SQL_ASCII; /* out of memory; unlikely */
231 name = setlocale(LC_CTYPE, ctype);
235 return PG_SQL_ASCII; /* bogus ctype passed in? */
239 sys = nl_langinfo(CODESET);
243 sys = win32_langinfo(name);
246 setlocale(LC_CTYPE, save);
252 ctype = setlocale(LC_CTYPE, NULL);
254 return PG_SQL_ASCII; /* setlocale() broken? */
256 sys = nl_langinfo(CODESET);
260 sys = win32_langinfo(ctype);
265 return PG_SQL_ASCII; /* out of memory; unlikely */
267 /* If locale is C or POSIX, we can allow all encodings */
268 if (pg_strcasecmp(ctype, "C") == 0 || pg_strcasecmp(ctype, "POSIX") == 0)
274 /* Check the table */
275 for (i = 0; encoding_match_list[i].system_enc_name; i++)
277 if (pg_strcasecmp(sys, encoding_match_list[i].system_enc_name) == 0)
280 return encoding_match_list[i].pg_enc_code;
284 /* Special-case kluges for particular platforms go here */
288 * Current OS X has many locales that report an empty string for CODESET,
289 * but they all seem to actually use UTF-8.
291 if (strlen(sys) == 0)
299 * We print a warning if we got a CODESET string but couldn't recognize
300 * it. This means we need another entry in the table.
303 fprintf(stderr, _("could not determine encoding for locale \"%s\": codeset is \"%s\""),
305 /* keep newline separate so there's only one translatable string */
309 (errmsg("could not determine encoding for locale \"%s\": codeset is \"%s\"",
311 errdetail("Please report this to <pgsql-bugs@postgresql.org>.")));
318 #else /* (HAVE_LANGINFO_H && CODESET) || WIN32 */
321 * stub if no platform support
324 pg_get_encoding_from_locale(const char *ctype)
329 #endif /* (HAVE_LANGINFO_H && CODESET) || WIN32 */