granicus.if.org Git - postgresql/blob - src/port/chklocale.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  * chklocale.c
   4  *              Functions for handling locale-related info
   5  *
   6  *
   7  * Copyright (c) 1996-2009, PostgreSQL Global Development Group
   8  *
   9  *
  10  * IDENTIFICATION
  11  *        $PostgreSQL: pgsql/src/port/chklocale.c,v 1.10 2009/01/01 17:24:04 momjian Exp $
  12  *
  13  *-------------------------------------------------------------------------
  14  */
  15
  16 #ifndef FRONTEND
  17 #include "postgres.h"
  18 #else
  19 #include "postgres_fe.h"
  20 #endif
  21
  22 #include <locale.h>
  23 #ifdef HAVE_LANGINFO_H
  24 #include <langinfo.h>
  25 #endif
  26
  27 #include "mb/pg_wchar.h"
  28
  29
  30 /*
  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
  35  * numbers (CPnnn).
  36  *
  37  * Note that we search the table with pg_strcasecmp(), so variant
  38  * capitalizations don't need their own entries.
  39  */
  40 struct encoding_match
  41 {
  42         enum pg_enc pg_enc_code;
  43         const char *system_enc_name;
  44 };
  45
  46 static const struct encoding_match encoding_match_list[] = {
  47         {PG_EUC_JP, "EUC-JP"},
  48         {PG_EUC_JP, "eucJP"},
  49         {PG_EUC_JP, "IBM-eucJP"},
  50         {PG_EUC_JP, "sdeckanji"},
  51         {PG_EUC_JP, "CP20932"},
  52
  53         {PG_EUC_CN, "EUC-CN"},
  54         {PG_EUC_CN, "eucCN"},
  55         {PG_EUC_CN, "IBM-eucCN"},
  56         {PG_EUC_CN, "GB2312"},
  57         {PG_EUC_CN, "dechanzi"},
  58         {PG_EUC_CN, "CP20936"},
  59
  60         {PG_EUC_KR, "EUC-KR"},
  61         {PG_EUC_KR, "eucKR"},
  62         {PG_EUC_KR, "IBM-eucKR"},
  63         {PG_EUC_KR, "deckorean"},
  64         {PG_EUC_KR, "5601"},
  65         {PG_EUC_KR, "CP51949"},         /* or 20949 ? */
  66
  67         {PG_EUC_TW, "EUC-TW"},
  68         {PG_EUC_TW, "eucTW"},
  69         {PG_EUC_TW, "IBM-eucTW"},
  70         {PG_EUC_TW, "cns11643"},
  71         /* No codepage for EUC-TW ? */
  72
  73         {PG_UTF8, "UTF-8"},
  74         {PG_UTF8, "utf8"},
  75         {PG_UTF8, "CP65001"},
  76
  77         {PG_LATIN1, "ISO-8859-1"},
  78         {PG_LATIN1, "ISO8859-1"},
  79         {PG_LATIN1, "iso88591"},
  80         {PG_LATIN1, "CP28591"},
  81
  82         {PG_LATIN2, "ISO-8859-2"},
  83         {PG_LATIN2, "ISO8859-2"},
  84         {PG_LATIN2, "iso88592"},
  85         {PG_LATIN2, "CP28592"},
  86
  87         {PG_LATIN3, "ISO-8859-3"},
  88         {PG_LATIN3, "ISO8859-3"},
  89         {PG_LATIN3, "iso88593"},
  90         {PG_LATIN3, "CP28593"},
  91
  92         {PG_LATIN4, "ISO-8859-4"},
  93         {PG_LATIN4, "ISO8859-4"},
  94         {PG_LATIN4, "iso88594"},
  95         {PG_LATIN4, "CP28594"},
  96
  97         {PG_LATIN5, "ISO-8859-9"},
  98         {PG_LATIN5, "ISO8859-9"},
  99         {PG_LATIN5, "iso88599"},
 100         {PG_LATIN5, "CP28599"},
 101
 102         {PG_LATIN6, "ISO-8859-10"},
 103         {PG_LATIN6, "ISO8859-10"},
 104         {PG_LATIN6, "iso885910"},
 105
 106         {PG_LATIN7, "ISO-8859-13"},
 107         {PG_LATIN7, "ISO8859-13"},
 108         {PG_LATIN7, "iso885913"},
 109
 110         {PG_LATIN8, "ISO-8859-14"},
 111         {PG_LATIN8, "ISO8859-14"},
 112         {PG_LATIN8, "iso885914"},
 113
 114         {PG_LATIN9, "ISO-8859-15"},
 115         {PG_LATIN9, "ISO8859-15"},
 116         {PG_LATIN9, "iso885915"},
 117         {PG_LATIN9, "CP28605"},
 118
 119         {PG_LATIN10, "ISO-8859-16"},
 120         {PG_LATIN10, "ISO8859-16"},
 121         {PG_LATIN10, "iso885916"},
 122
 123         {PG_KOI8R, "KOI8-R"},
 124         {PG_KOI8R, "CP20866"},
 125
 126         {PG_WIN866, "CP866"},
 127         {PG_WIN874, "CP874"},
 128         {PG_WIN1250, "CP1250"},
 129         {PG_WIN1251, "CP1251"},
 130         {PG_WIN1251, "ansi-1251"},
 131         {PG_WIN1252, "CP1252"},
 132         {PG_WIN1253, "CP1253"},
 133         {PG_WIN1254, "CP1254"},
 134         {PG_WIN1255, "CP1255"},
 135         {PG_WIN1256, "CP1256"},
 136         {PG_WIN1257, "CP1257"},
 137         {PG_WIN1258, "CP1258"},
 138
 139         {PG_ISO_8859_5, "ISO-8859-5"},
 140         {PG_ISO_8859_5, "ISO8859-5"},
 141         {PG_ISO_8859_5, "iso88595"},
 142         {PG_ISO_8859_5, "CP28595"},
 143
 144         {PG_ISO_8859_6, "ISO-8859-6"},
 145         {PG_ISO_8859_6, "ISO8859-6"},
 146         {PG_ISO_8859_6, "iso88596"},
 147         {PG_ISO_8859_6, "CP28596"},
 148
 149         {PG_ISO_8859_7, "ISO-8859-7"},
 150         {PG_ISO_8859_7, "ISO8859-7"},
 151         {PG_ISO_8859_7, "iso88597"},
 152         {PG_ISO_8859_7, "CP28597"},
 153
 154         {PG_ISO_8859_8, "ISO-8859-8"},
 155         {PG_ISO_8859_8, "ISO8859-8"},
 156         {PG_ISO_8859_8, "iso88598"},
 157         {PG_ISO_8859_8, "CP28598"},
 158
 159         {PG_SJIS, "SJIS"},
 160         {PG_SJIS, "PCK"},
 161         {PG_SJIS, "CP932"},
 162
 163         {PG_BIG5, "BIG5"},
 164         {PG_BIG5, "BIG5HKSCS"},
 165         {PG_BIG5, "Big5-HKSCS"},
 166         {PG_BIG5, "CP950"},
 167
 168         {PG_GBK, "GBK"},
 169         {PG_GBK, "CP936"},
 170
 171         {PG_UHC, "UHC"},
 172
 173         {PG_JOHAB, "JOHAB"},
 174         {PG_JOHAB, "CP1361"},
 175
 176         {PG_GB18030, "GB18030"},
 177         {PG_GB18030, "CP54936"},
 178
 179         {PG_SHIFT_JIS_2004, "SJIS_2004"},
 180
 181         {PG_SQL_ASCII, NULL}            /* end marker */
 182 };
 183
 184 #ifdef WIN32
 185 /*
 186  * On Windows, use CP<codepage number> instead of the nl_langinfo() result
 187  */
 188 static char *
 189 win32_langinfo(const char *ctype)
 190 {
 191         char       *r;
 192         char       *codepage;
 193         int                     ln;
 194
 195         /*
 196          * Locale format on Win32 is <Language>_<Country>.<CodePage> . For
 197          * example, English_USA.1252.
 198          */
 199         codepage = strrchr(ctype, '.');
 200         if (!codepage)
 201                 return NULL;
 202         codepage++;
 203         ln = strlen(codepage);
 204         r = malloc(ln + 3);
 205         sprintf(r, "CP%s", codepage);
 206
 207         return r;
 208 }
 209 #endif   /* WIN32 */
 210
 211 #if (defined(HAVE_LANGINFO_H) && defined(CODESET)) || defined(WIN32)
 212
 213 /*
 214  * Given a setting for LC_CTYPE, return the Postgres ID of the associated
 215  * encoding, if we can determine it.
 216  *
 217  * Pass in NULL to get the encoding for the current locale setting.
 218  *
 219  * If the result is PG_SQL_ASCII, callers should treat it as being compatible
 220  * with any desired encoding.  We return this if the locale is C/POSIX or we
 221  * can't determine the encoding.
 222  */
 223 int
 224 pg_get_encoding_from_locale(const char *ctype)
 225 {
 226         char       *sys;
 227         int                     i;
 228
 229         /* Get the CODESET property, and also LC_CTYPE if not passed in */
 230         if (ctype)
 231         {
 232                 char       *save;
 233                 char       *name;
 234
 235                 save = setlocale(LC_CTYPE, NULL);
 236                 if (!save)
 237                         return PG_SQL_ASCII;    /* setlocale() broken? */
 238                 /* must copy result, or it might change after setlocale */
 239                 save = strdup(save);
 240                 if (!save)
 241                         return PG_SQL_ASCII;    /* out of memory; unlikely */
 242
 243                 name = setlocale(LC_CTYPE, ctype);
 244                 if (!name)
 245                 {
 246                         free(save);
 247                         return PG_SQL_ASCII;    /* bogus ctype passed in? */
 248                 }
 249
 250 #ifndef WIN32
 251                 sys = nl_langinfo(CODESET);
 252                 if (sys)
 253                         sys = strdup(sys);
 254 #else
 255                 sys = win32_langinfo(name);
 256 #endif
 257
 258                 setlocale(LC_CTYPE, save);
 259                 free(save);
 260         }
 261         else
 262         {
 263                 /* much easier... */
 264                 ctype = setlocale(LC_CTYPE, NULL);
 265                 if (!ctype)
 266                         return PG_SQL_ASCII;    /* setlocale() broken? */
 267 #ifndef WIN32
 268                 sys = nl_langinfo(CODESET);
 269                 if (sys)
 270                         sys = strdup(sys);
 271 #else
 272                 sys = win32_langinfo(ctype);
 273 #endif
 274         }
 275
 276         if (!sys)
 277                 return PG_SQL_ASCII;    /* out of memory; unlikely */
 278
 279         /* If locale is C or POSIX, we can allow all encodings */
 280         if (pg_strcasecmp(ctype, "C") == 0 || pg_strcasecmp(ctype, "POSIX") == 0)
 281         {
 282                 free(sys);
 283                 return PG_SQL_ASCII;
 284         }
 285
 286         /* Check the table */
 287         for (i = 0; encoding_match_list[i].system_enc_name; i++)
 288         {
 289                 if (pg_strcasecmp(sys, encoding_match_list[i].system_enc_name) == 0)
 290                 {
 291                         free(sys);
 292                         return encoding_match_list[i].pg_enc_code;
 293                 }
 294         }
 295
 296         /* Special-case kluges for particular platforms go here */
 297
 298 #ifdef __darwin__
 299
 300         /*
 301          * Current OS X has many locales that report an empty string for CODESET,
 302          * but they all seem to actually use UTF-8.
 303          */
 304         if (strlen(sys) == 0)
 305         {
 306                 free(sys);
 307                 return PG_UTF8;
 308         }
 309 #endif
 310
 311         /*
 312          * We print a warning if we got a CODESET string but couldn't recognize
 313          * it.  This means we need another entry in the table.
 314          */
 315 #ifdef FRONTEND
 316         fprintf(stderr, _("could not determine encoding for locale \"%s\": codeset is \"%s\""),
 317                         ctype, sys);
 318         /* keep newline separate so there's only one translatable string */
 319         fputc('\n', stderr);
 320 #else
 321         ereport(WARNING,
 322                         (errmsg("could not determine encoding for locale \"%s\": codeset is \"%s\"",
 323                                         ctype, sys),
 324                    errdetail("Please report this to <pgsql-bugs@postgresql.org>.")));
 325 #endif
 326
 327         free(sys);
 328         return PG_SQL_ASCII;
 329 }
 330 #else                                                   /* (HAVE_LANGINFO_H && CODESET) || WIN32 */
 331
 332 /*
 333  * stub if no platform support
 334  */
 335 int
 336 pg_get_encoding_from_locale(const char *ctype)
 337 {
 338         return PG_SQL_ASCII;
 339 }
 340
 341 #endif   /* (HAVE_LANGINFO_H && CODESET) || WIN32 */