PostgreSQL Source Code  git master
chklocale.c
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * chklocale.c
4  * Functions for handling locale-related info
5  *
6  *
7  * Copyright (c) 1996-2024, PostgreSQL Global Development Group
8  *
9  *
10  * IDENTIFICATION
11  * src/port/chklocale.c
12  *
13  *-------------------------------------------------------------------------
14  */
15 
16 #ifndef FRONTEND
17 #include "postgres.h"
18 #else
19 #include "postgres_fe.h"
20 #endif
21 
22 #ifdef HAVE_LANGINFO_H
23 #include <langinfo.h>
24 #endif
25 
26 #include "mb/pg_wchar.h"
27 
28 
29 /*
30  * This table needs to recognize all the CODESET spellings for supported
31  * backend encodings, as well as frontend-only encodings where possible
32  * (the latter case is currently only needed for initdb to recognize
33  * error situations). On Windows, we rely on entries for codepage
34  * numbers (CPnnn).
35  *
36  * Note that we search the table with pg_strcasecmp(), so variant
37  * capitalizations don't need their own entries.
38  */
40 {
41  enum pg_enc pg_enc_code;
42  const char *system_enc_name;
43 };
44 
45 static const struct encoding_match encoding_match_list[] = {
46  {PG_EUC_JP, "EUC-JP"},
47  {PG_EUC_JP, "eucJP"},
48  {PG_EUC_JP, "IBM-eucJP"},
49  {PG_EUC_JP, "sdeckanji"},
50  {PG_EUC_JP, "CP20932"},
51 
52  {PG_EUC_CN, "EUC-CN"},
53  {PG_EUC_CN, "eucCN"},
54  {PG_EUC_CN, "IBM-eucCN"},
55  {PG_EUC_CN, "GB2312"},
56  {PG_EUC_CN, "dechanzi"},
57  {PG_EUC_CN, "CP20936"},
58 
59  {PG_EUC_KR, "EUC-KR"},
60  {PG_EUC_KR, "eucKR"},
61  {PG_EUC_KR, "IBM-eucKR"},
62  {PG_EUC_KR, "deckorean"},
63  {PG_EUC_KR, "5601"},
64  {PG_EUC_KR, "CP51949"},
65 
66  {PG_EUC_TW, "EUC-TW"},
67  {PG_EUC_TW, "eucTW"},
68  {PG_EUC_TW, "IBM-eucTW"},
69  {PG_EUC_TW, "cns11643"},
70  /* No codepage for EUC-TW ? */
71 
72  {PG_UTF8, "UTF-8"},
73  {PG_UTF8, "utf8"},
74  {PG_UTF8, "CP65001"},
75 
76  {PG_LATIN1, "ISO-8859-1"},
77  {PG_LATIN1, "ISO8859-1"},
78  {PG_LATIN1, "iso88591"},
79  {PG_LATIN1, "CP28591"},
80 
81  {PG_LATIN2, "ISO-8859-2"},
82  {PG_LATIN2, "ISO8859-2"},
83  {PG_LATIN2, "iso88592"},
84  {PG_LATIN2, "CP28592"},
85 
86  {PG_LATIN3, "ISO-8859-3"},
87  {PG_LATIN3, "ISO8859-3"},
88  {PG_LATIN3, "iso88593"},
89  {PG_LATIN3, "CP28593"},
90 
91  {PG_LATIN4, "ISO-8859-4"},
92  {PG_LATIN4, "ISO8859-4"},
93  {PG_LATIN4, "iso88594"},
94  {PG_LATIN4, "CP28594"},
95 
96  {PG_LATIN5, "ISO-8859-9"},
97  {PG_LATIN5, "ISO8859-9"},
98  {PG_LATIN5, "iso88599"},
99  {PG_LATIN5, "CP28599"},
100 
101  {PG_LATIN6, "ISO-8859-10"},
102  {PG_LATIN6, "ISO8859-10"},
103  {PG_LATIN6, "iso885910"},
104 
105  {PG_LATIN7, "ISO-8859-13"},
106  {PG_LATIN7, "ISO8859-13"},
107  {PG_LATIN7, "iso885913"},
108 
109  {PG_LATIN8, "ISO-8859-14"},
110  {PG_LATIN8, "ISO8859-14"},
111  {PG_LATIN8, "iso885914"},
112 
113  {PG_LATIN9, "ISO-8859-15"},
114  {PG_LATIN9, "ISO8859-15"},
115  {PG_LATIN9, "iso885915"},
116  {PG_LATIN9, "CP28605"},
117 
118  {PG_LATIN10, "ISO-8859-16"},
119  {PG_LATIN10, "ISO8859-16"},
120  {PG_LATIN10, "iso885916"},
121 
122  {PG_KOI8R, "KOI8-R"},
123  {PG_KOI8R, "CP20866"},
124 
125  {PG_KOI8U, "KOI8-U"},
126  {PG_KOI8U, "CP21866"},
127 
128  {PG_WIN866, "CP866"},
129  {PG_WIN874, "CP874"},
130  {PG_WIN1250, "CP1250"},
131  {PG_WIN1251, "CP1251"},
132  {PG_WIN1251, "ansi-1251"},
133  {PG_WIN1252, "CP1252"},
134  {PG_WIN1253, "CP1253"},
135  {PG_WIN1254, "CP1254"},
136  {PG_WIN1255, "CP1255"},
137  {PG_WIN1256, "CP1256"},
138  {PG_WIN1257, "CP1257"},
139  {PG_WIN1258, "CP1258"},
140 
141  {PG_ISO_8859_5, "ISO-8859-5"},
142  {PG_ISO_8859_5, "ISO8859-5"},
143  {PG_ISO_8859_5, "iso88595"},
144  {PG_ISO_8859_5, "CP28595"},
145 
146  {PG_ISO_8859_6, "ISO-8859-6"},
147  {PG_ISO_8859_6, "ISO8859-6"},
148  {PG_ISO_8859_6, "iso88596"},
149  {PG_ISO_8859_6, "CP28596"},
150 
151  {PG_ISO_8859_7, "ISO-8859-7"},
152  {PG_ISO_8859_7, "ISO8859-7"},
153  {PG_ISO_8859_7, "iso88597"},
154  {PG_ISO_8859_7, "CP28597"},
155 
156  {PG_ISO_8859_8, "ISO-8859-8"},
157  {PG_ISO_8859_8, "ISO8859-8"},
158  {PG_ISO_8859_8, "iso88598"},
159  {PG_ISO_8859_8, "CP28598"},
160 
161  {PG_SJIS, "SJIS"},
162  {PG_SJIS, "PCK"},
163  {PG_SJIS, "CP932"},
164  {PG_SJIS, "SHIFT_JIS"},
165 
166  {PG_BIG5, "BIG5"},
167  {PG_BIG5, "BIG5HKSCS"},
168  {PG_BIG5, "Big5-HKSCS"},
169  {PG_BIG5, "CP950"},
170 
171  {PG_GBK, "GBK"},
172  {PG_GBK, "CP936"},
173 
174  {PG_UHC, "UHC"},
175  {PG_UHC, "CP949"},
176 
177  {PG_JOHAB, "JOHAB"},
178  {PG_JOHAB, "CP1361"},
179 
180  {PG_GB18030, "GB18030"},
181  {PG_GB18030, "CP54936"},
182 
183  {PG_SHIFT_JIS_2004, "SJIS_2004"},
184 
185  {PG_SQL_ASCII, "US-ASCII"},
186 
187  {PG_SQL_ASCII, NULL} /* end marker */
188 };
189 
190 #ifdef WIN32
191 /*
192  * On Windows, use CP<code page number> instead of the nl_langinfo() result
193  *
194  * This routine uses GetLocaleInfoEx() to parse short locale names like
195  * "de-DE", "fr-FR", etc. If those cannot be parsed correctly process falls
196  * back to the pre-VS-2010 manual parsing done with using
197  * <Language>_<Country>.<CodePage> as a base.
198  *
199  * Returns a malloc()'d string for the caller to free.
200  */
201 static char *
202 win32_langinfo(const char *ctype)
203 {
204  char *r = NULL;
205  char *codepage;
206 
207 #if defined(_MSC_VER)
208  uint32 cp;
209  WCHAR wctype[LOCALE_NAME_MAX_LENGTH];
210 
211  memset(wctype, 0, sizeof(wctype));
212  MultiByteToWideChar(CP_ACP, 0, ctype, -1, wctype, LOCALE_NAME_MAX_LENGTH);
213 
214  if (GetLocaleInfoEx(wctype,
215  LOCALE_IDEFAULTANSICODEPAGE | LOCALE_RETURN_NUMBER,
216  (LPWSTR) &cp, sizeof(cp) / sizeof(WCHAR)) > 0)
217  {
218  r = malloc(16); /* excess */
219  if (r != NULL)
220  {
221  /*
222  * If the return value is CP_ACP that means no ANSI code page is
223  * available, so only Unicode can be used for the locale.
224  */
225  if (cp == CP_ACP)
226  strcpy(r, "utf8");
227  else
228  sprintf(r, "CP%u", cp);
229  }
230  }
231  else
232 #endif
233  {
234  /*
235  * Locale format on Win32 is <Language>_<Country>.<CodePage>. For
236  * example, English_United States.1252. If we see digits after the
237  * last dot, assume it's a codepage number. Otherwise, we might be
238  * dealing with a Unix-style locale string; Windows' setlocale() will
239  * take those even though GetLocaleInfoEx() won't, so we end up here.
240  * In that case, just return what's after the last dot and hope we can
241  * find it in our table.
242  */
243  codepage = strrchr(ctype, '.');
244  if (codepage != NULL)
245  {
246  size_t ln;
247 
248  codepage++;
249  ln = strlen(codepage);
250  r = malloc(ln + 3);
251  if (r != NULL)
252  {
253  if (strspn(codepage, "0123456789") == ln)
254  sprintf(r, "CP%s", codepage);
255  else
256  strcpy(r, codepage);
257  }
258  }
259  }
260 
261  return r;
262 }
263 
264 #ifndef FRONTEND
265 /*
266  * Given a Windows code page identifier, find the corresponding PostgreSQL
267  * encoding. Issue a warning and return -1 if none found.
268  */
269 int
270 pg_codepage_to_encoding(UINT cp)
271 {
272  char sys[16];
273  int i;
274 
275  sprintf(sys, "CP%u", cp);
276 
277  /* Check the table */
278  for (i = 0; encoding_match_list[i].system_enc_name; i++)
281 
283  (errmsg("could not determine encoding for codeset \"%s\"", sys)));
284 
285  return -1;
286 }
287 #endif
288 #endif /* WIN32 */
289 
290 #if (defined(HAVE_LANGINFO_H) && defined(CODESET)) || defined(WIN32)
291 
292 /*
293  * Given a setting for LC_CTYPE, return the Postgres ID of the associated
294  * encoding, if we can determine it. Return -1 if we can't determine it.
295  *
296  * Pass in NULL to get the encoding for the current locale setting.
297  * Pass "" to get the encoding selected by the server's environment.
298  *
299  * If the result is PG_SQL_ASCII, callers should treat it as being compatible
300  * with any desired encoding.
301  *
302  * If running in the backend and write_message is false, this function must
303  * cope with the possibility that elog() and palloc() are not yet usable.
304  */
305 int
306 pg_get_encoding_from_locale(const char *ctype, bool write_message)
307 {
308  char *sys;
309  int i;
310 
311  /* Get the CODESET property, and also LC_CTYPE if not passed in */
312  if (ctype)
313  {
314  char *save;
315  char *name;
316 
317  /* If locale is C or POSIX, we can allow all encodings */
318  if (pg_strcasecmp(ctype, "C") == 0 ||
319  pg_strcasecmp(ctype, "POSIX") == 0)
320  return PG_SQL_ASCII;
321 
322  save = setlocale(LC_CTYPE, NULL);
323  if (!save)
324  return -1; /* setlocale() broken? */
325  /* must copy result, or it might change after setlocale */
326  save = strdup(save);
327  if (!save)
328  return -1; /* out of memory; unlikely */
329 
330  name = setlocale(LC_CTYPE, ctype);
331  if (!name)
332  {
333  free(save);
334  return -1; /* bogus ctype passed in? */
335  }
336 
337 #ifndef WIN32
338  sys = nl_langinfo(CODESET);
339  if (sys)
340  sys = strdup(sys);
341 #else
342  sys = win32_langinfo(name);
343 #endif
344 
345  setlocale(LC_CTYPE, save);
346  free(save);
347  }
348  else
349  {
350  /* much easier... */
351  ctype = setlocale(LC_CTYPE, NULL);
352  if (!ctype)
353  return -1; /* setlocale() broken? */
354 
355  /* If locale is C or POSIX, we can allow all encodings */
356  if (pg_strcasecmp(ctype, "C") == 0 ||
357  pg_strcasecmp(ctype, "POSIX") == 0)
358  return PG_SQL_ASCII;
359 
360 #ifndef WIN32
361  sys = nl_langinfo(CODESET);
362  if (sys)
363  sys = strdup(sys);
364 #else
365  sys = win32_langinfo(ctype);
366 #endif
367  }
368 
369  if (!sys)
370  return -1; /* out of memory; unlikely */
371 
372  /* Check the table */
373  for (i = 0; encoding_match_list[i].system_enc_name; i++)
374  {
376  {
377  free(sys);
379  }
380  }
381 
382  /* Special-case kluges for particular platforms go here */
383 
384 #ifdef __darwin__
385 
386  /*
387  * Current macOS has many locales that report an empty string for CODESET,
388  * but they all seem to actually use UTF-8.
389  */
390  if (strlen(sys) == 0)
391  {
392  free(sys);
393  return PG_UTF8;
394  }
395 #endif
396 
397  /*
398  * We print a warning if we got a CODESET string but couldn't recognize
399  * it. This means we need another entry in the table.
400  */
401  if (write_message)
402  {
403 #ifdef FRONTEND
404  fprintf(stderr, _("could not determine encoding for locale \"%s\": codeset is \"%s\""),
405  ctype, sys);
406  /* keep newline separate so there's only one translatable string */
407  fputc('\n', stderr);
408 #else
410  (errmsg("could not determine encoding for locale \"%s\": codeset is \"%s\"",
411  ctype, sys)));
412 #endif
413  }
414 
415  free(sys);
416  return -1;
417 }
418 #else /* (HAVE_LANGINFO_H && CODESET) || WIN32 */
419 
420 /*
421  * stub if no multi-language platform support
422  *
423  * Note: we could return -1 here, but that would have the effect of
424  * forcing users to specify an encoding to initdb on such platforms.
425  * It seems better to silently default to SQL_ASCII.
426  */
427 int
428 pg_get_encoding_from_locale(const char *ctype, bool write_message)
429 {
430  return PG_SQL_ASCII;
431 }
432 
433 #endif /* (HAVE_LANGINFO_H && CODESET) || WIN32 */
unsigned int uint32
Definition: c.h:493
static const struct encoding_match encoding_match_list[]
Definition: chklocale.c:45
int pg_get_encoding_from_locale(const char *ctype, bool write_message)
Definition: chklocale.c:428
int errmsg(const char *fmt,...)
Definition: elog.c:1072
#define _(x)
Definition: elog.c:90
#define WARNING
Definition: elog.h:36
#define ereport(elevel,...)
Definition: elog.h:149
#define free(a)
Definition: header.h:65
#define malloc(a)
Definition: header.h:50
int i
Definition: isn.c:73
pg_enc
Definition: pg_wchar.h:225
@ PG_WIN1254
Definition: pg_wchar.h:257
@ PG_LATIN4
Definition: pg_wchar.h:237
@ PG_LATIN9
Definition: pg_wchar.h:242
@ PG_JOHAB
Definition: pg_wchar.h:269
@ PG_GB18030
Definition: pg_wchar.h:268
@ PG_SQL_ASCII
Definition: pg_wchar.h:226
@ PG_KOI8R
Definition: pg_wchar.h:248
@ PG_ISO_8859_6
Definition: pg_wchar.h:252
@ PG_WIN1253
Definition: pg_wchar.h:256
@ PG_KOI8U
Definition: pg_wchar.h:260
@ PG_LATIN6
Definition: pg_wchar.h:239
@ PG_LATIN5
Definition: pg_wchar.h:238
@ PG_EUC_CN
Definition: pg_wchar.h:228
@ PG_UHC
Definition: pg_wchar.h:267
@ PG_LATIN2
Definition: pg_wchar.h:235
@ PG_ISO_8859_5
Definition: pg_wchar.h:251
@ PG_LATIN10
Definition: pg_wchar.h:243
@ PG_WIN1250
Definition: pg_wchar.h:255
@ PG_ISO_8859_7
Definition: pg_wchar.h:253
@ PG_SJIS
Definition: pg_wchar.h:264
@ PG_LATIN8
Definition: pg_wchar.h:241
@ PG_EUC_JP
Definition: pg_wchar.h:227
@ PG_GBK
Definition: pg_wchar.h:266
@ PG_LATIN3
Definition: pg_wchar.h:236
@ PG_WIN1256
Definition: pg_wchar.h:244
@ PG_LATIN1
Definition: pg_wchar.h:234
@ PG_EUC_TW
Definition: pg_wchar.h:230
@ PG_WIN1258
Definition: pg_wchar.h:245
@ PG_SHIFT_JIS_2004
Definition: pg_wchar.h:270
@ PG_WIN1252
Definition: pg_wchar.h:250
@ PG_LATIN7
Definition: pg_wchar.h:240
@ PG_UTF8
Definition: pg_wchar.h:232
@ PG_WIN1255
Definition: pg_wchar.h:258
@ PG_WIN1257
Definition: pg_wchar.h:259
@ PG_WIN1251
Definition: pg_wchar.h:249
@ PG_EUC_KR
Definition: pg_wchar.h:229
@ PG_WIN866
Definition: pg_wchar.h:246
@ PG_ISO_8859_8
Definition: pg_wchar.h:254
@ PG_WIN874
Definition: pg_wchar.h:247
@ PG_BIG5
Definition: pg_wchar.h:265
int pg_strcasecmp(const char *s1, const char *s2)
Definition: pgstrcasecmp.c:36
#define sprintf
Definition: port.h:240
#define fprintf
Definition: port.h:242
enum pg_enc pg_enc_code
Definition: chklocale.c:41
const char * system_enc_name
Definition: chklocale.c:42
const char * name
#define setlocale(a, b)
Definition: win32_port.h:467