@@ -1030,3 +1030,176 @@ pg_newlocale_from_collation(Oid collid)
10301030
10311031 return cache_entry -> locale ;
10321032}
1033+
1034+
1035+ /*
1036+ * These functions convert from/to libc's wchar_t, *not* pg_wchar_t.
1037+ * Therefore we keep them here rather than with the mbutils code.
1038+ */
1039+
1040+ #ifdef USE_WIDE_UPPER_LOWER
1041+
1042+ /*
1043+ * wchar2char --- convert wide characters to multibyte format
1044+ *
1045+ * This has the same API as the standard wcstombs_l() function; in particular,
1046+ * tolen is the maximum number of bytes to store at *to, and *from must be
1047+ * zero-terminated. The output will be zero-terminated iff there is room.
1048+ */
1049+ size_t
1050+ wchar2char (char * to , const wchar_t * from , size_t tolen , pg_locale_t locale )
1051+ {
1052+ size_t result ;
1053+
1054+ if (tolen == 0 )
1055+ return 0 ;
1056+
1057+ #ifdef WIN32
1058+
1059+ /*
1060+ * On Windows, the "Unicode" locales assume UTF16 not UTF8 encoding, and
1061+ * for some reason mbstowcs and wcstombs won't do this for us, so we use
1062+ * MultiByteToWideChar().
1063+ */
1064+ if (GetDatabaseEncoding () == PG_UTF8 )
1065+ {
1066+ result = WideCharToMultiByte (CP_UTF8 , 0 , from , -1 , to , tolen ,
1067+ NULL , NULL );
1068+ /* A zero return is failure */
1069+ if (result <= 0 )
1070+ result = -1 ;
1071+ else
1072+ {
1073+ Assert (result <= tolen );
1074+ /* Microsoft counts the zero terminator in the result */
1075+ result -- ;
1076+ }
1077+ }
1078+ else
1079+ #endif /* WIN32 */
1080+ if (locale == (pg_locale_t ) 0 )
1081+ {
1082+ /* Use wcstombs directly for the default locale */
1083+ result = wcstombs (to , from , tolen );
1084+ }
1085+ else
1086+ {
1087+ #ifdef HAVE_LOCALE_T
1088+ #ifdef HAVE_WCSTOMBS_L
1089+ /* Use wcstombs_l for nondefault locales */
1090+ result = wcstombs_l (to , from , tolen , locale );
1091+ #else /* !HAVE_WCSTOMBS_L */
1092+ /* We have to temporarily set the locale as current ... ugh */
1093+ locale_t save_locale = uselocale (locale );
1094+
1095+ result = wcstombs (to , from , tolen );
1096+
1097+ uselocale (save_locale );
1098+ #endif /* HAVE_WCSTOMBS_L */
1099+ #else /* !HAVE_LOCALE_T */
1100+ /* Can't have locale != 0 without HAVE_LOCALE_T */
1101+ elog (ERROR , "wcstombs_l is not available" );
1102+ result = 0 ; /* keep compiler quiet */
1103+ #endif /* HAVE_LOCALE_T */
1104+ }
1105+
1106+ return result ;
1107+ }
1108+
1109+ /*
1110+ * char2wchar --- convert multibyte characters to wide characters
1111+ *
1112+ * This has almost the API of mbstowcs_l(), except that *from need not be
1113+ * null-terminated; instead, the number of input bytes is specified as
1114+ * fromlen. Also, we ereport() rather than returning -1 for invalid
1115+ * input encoding. tolen is the maximum number of wchar_t's to store at *to.
1116+ * The output will be zero-terminated iff there is room.
1117+ */
1118+ size_t
1119+ char2wchar (wchar_t * to , size_t tolen , const char * from , size_t fromlen ,
1120+ pg_locale_t locale )
1121+ {
1122+ size_t result ;
1123+
1124+ if (tolen == 0 )
1125+ return 0 ;
1126+
1127+ #ifdef WIN32
1128+ /* See WIN32 "Unicode" comment above */
1129+ if (GetDatabaseEncoding () == PG_UTF8 )
1130+ {
1131+ /* Win32 API does not work for zero-length input */
1132+ if (fromlen == 0 )
1133+ result = 0 ;
1134+ else
1135+ {
1136+ result = MultiByteToWideChar (CP_UTF8 , 0 , from , fromlen , to , tolen - 1 );
1137+ /* A zero return is failure */
1138+ if (result == 0 )
1139+ result = -1 ;
1140+ }
1141+
1142+ if (result != -1 )
1143+ {
1144+ Assert (result < tolen );
1145+ /* Append trailing null wchar (MultiByteToWideChar() does not) */
1146+ to [result ] = 0 ;
1147+ }
1148+ }
1149+ else
1150+ #endif /* WIN32 */
1151+ {
1152+ /* mbstowcs requires ending '\0' */
1153+ char * str = pnstrdup (from , fromlen );
1154+
1155+ if (locale == (pg_locale_t ) 0 )
1156+ {
1157+ /* Use mbstowcs directly for the default locale */
1158+ result = mbstowcs (to , str , tolen );
1159+ }
1160+ else
1161+ {
1162+ #ifdef HAVE_LOCALE_T
1163+ #ifdef HAVE_WCSTOMBS_L
1164+ /* Use mbstowcs_l for nondefault locales */
1165+ result = mbstowcs_l (to , str , tolen , locale );
1166+ #else /* !HAVE_WCSTOMBS_L */
1167+ /* We have to temporarily set the locale as current ... ugh */
1168+ locale_t save_locale = uselocale (locale );
1169+
1170+ result = mbstowcs (to , str , tolen );
1171+
1172+ uselocale (save_locale );
1173+ #endif /* HAVE_WCSTOMBS_L */
1174+ #else /* !HAVE_LOCALE_T */
1175+ /* Can't have locale != 0 without HAVE_LOCALE_T */
1176+ elog (ERROR , "mbstowcs_l is not available" );
1177+ result = 0 ; /* keep compiler quiet */
1178+ #endif /* HAVE_LOCALE_T */
1179+ }
1180+
1181+ pfree (str );
1182+ }
1183+
1184+ if (result == -1 )
1185+ {
1186+ /*
1187+ * Invalid multibyte character encountered. We try to give a useful
1188+ * error message by letting pg_verifymbstr check the string. But it's
1189+ * possible that the string is OK to us, and not OK to mbstowcs ---
1190+ * this suggests that the LC_CTYPE locale is different from the
1191+ * database encoding. Give a generic error message if verifymbstr
1192+ * can't find anything wrong.
1193+ */
1194+ pg_verifymbstr (from , fromlen , false); /* might not return */
1195+ /* but if it does ... */
1196+ ereport (ERROR ,
1197+ (errcode (ERRCODE_CHARACTER_NOT_IN_REPERTOIRE ),
1198+ errmsg ("invalid multibyte character for locale" ),
1199+ errhint ("The server's LC_CTYPE locale is probably incompatible with the database encoding." )));
1200+ }
1201+
1202+ return result ;
1203+ }
1204+
1205+ #endif /* USE_WIDE_UPPER_LOWER */
0 commit comments