| 1 | diff -Naur postgresql-9.0.6-orig/configure.in postgresql-9.0.6/configure.in |
| 2 | --- configure.in 2012-02-03 11:56:32.000000000 +1100 |
| 3 | +++ configure.in 2012-02-03 12:02:54.000000000 +1100 |
| 4 | @@ -678,6 +678,16 @@ |
| 5 | |
| 6 | |
| 7 | # |
| 8 | +# ICU |
| 9 | +# |
| 10 | +AC_MSG_CHECKING([whether to build with ICU support]) |
| 11 | +PGAC_ARG_BOOL(with, icu, no, [ --with-icu build with ICU support], |
| 12 | + [AC_DEFINE([USE_ICU], 1, [Define to build with ICU support. (--with-icu)])]) |
| 13 | +AC_MSG_RESULT([$with_icu]) |
| 14 | +AC_SUBST(with_icu) |
| 15 | + |
| 16 | + |
| 17 | +# |
| 18 | # Readline |
| 19 | # |
| 20 | PGAC_ARG_BOOL(with, readline, yes, |
| 21 | @@ -936,6 +946,31 @@ |
| 22 | fi |
| 23 | fi |
| 24 | |
| 25 | +if test "$with_icu" = yes ; then |
| 26 | + AC_CHECK_LIB(icui18n, ucol_open_48, [], [ |
| 27 | + AC_CHECK_LIB(icui18n, ucol_open_44, [], [ |
| 28 | + AC_CHECK_LIB(icui18n, ucol_open_43, [], [ |
| 29 | + AC_CHECK_LIB(icui18n, ucol_open_3_8, [], [ |
| 30 | + AC_CHECK_LIB(icui18n, ucol_open_3_6, [], [ |
| 31 | + AC_CHECK_LIB(icui18n, ucol_open_3_4, [], [AC_MSG_ERROR([library 'icui18n' is required for ICU])]) |
| 32 | + ]) |
| 33 | + ]) |
| 34 | + ]) |
| 35 | + ]) |
| 36 | + ]) |
| 37 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_48, [], [ |
| 38 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_44, [], [ |
| 39 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_43, [], [ |
| 40 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_3_8, [], [ |
| 41 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_3_6, [], [ |
| 42 | + AC_CHECK_LIB(icuuc, ucnv_fromUChars_3_4, [], [AC_MSG_ERROR([library 'icuuc' is required for ICU])]) |
| 43 | + ]) |
| 44 | + ]) |
| 45 | + ]) |
| 46 | + ]) |
| 47 | + ]) |
| 48 | +fi |
| 49 | + |
| 50 | if test "$with_pam" = yes ; then |
| 51 | AC_CHECK_LIB(pam, pam_start, [], [AC_MSG_ERROR([library 'pam' is required for PAM])]) |
| 52 | fi |
| 53 | @@ -1042,6 +1077,10 @@ |
| 54 | AC_CHECK_FUNCS([ERR_set_mark]) |
| 55 | fi |
| 56 | |
| 57 | +if test "$with_icu" = yes ; then |
| 58 | + AC_CHECK_HEADER(unicode/utypes.h, [], [AC_MSG_ERROR([header file <unicode/utypes.h> is required for ICU])]) |
| 59 | +fi |
| 60 | + |
| 61 | if test "$with_pam" = yes ; then |
| 62 | AC_CHECK_HEADERS(security/pam_appl.h, [], |
| 63 | [AC_CHECK_HEADERS(pam/pam_appl.h, [], |
| 64 | diff -Naur postgresql-9.0.6-orig/src/backend/utils/adt/formatting.c postgresql-9.0.6/src/backend/utils/adt/formatting.c |
| 65 | --- src/backend/utils/adt/formatting.c 2012-02-03 11:56:34.000000000 +1100 |
| 66 | +++ src/backend/utils/adt/formatting.c 2012-02-03 12:02:54.000000000 +1100 |
| 67 | @@ -91,6 +91,12 @@ |
| 68 | #include "utils/numeric.h" |
| 69 | #include "utils/pg_locale.h" |
| 70 | |
| 71 | +#ifdef USE_ICU |
| 72 | +#include <unicode/utypes.h> /* Basic ICU data types */ |
| 73 | +#include <unicode/ucnv.h> /* C Converter API */ |
| 74 | +#include <unicode/ustring.h> |
| 75 | +#endif /* USE_ICU */ |
| 76 | + |
| 77 | /* ---------- |
| 78 | * Routines type |
| 79 | * ---------- |
| 80 | @@ -941,6 +947,12 @@ |
| 81 | } NUMProc; |
| 82 | |
| 83 | |
| 84 | +#ifdef USE_ICU |
| 85 | +static UConverter *conv = NULL; |
| 86 | +#define STACKBUFLEN 1024 / sizeof(UChar) |
| 87 | +#endif /* USE_ICU */ |
| 88 | + |
| 89 | + |
| 90 | /* ---------- |
| 91 | * Functions |
| 92 | * ---------- |
| 93 | @@ -1477,6 +1489,83 @@ |
| 94 | if (!buff) |
| 95 | return NULL; |
| 96 | |
| 97 | +#ifdef USE_ICU |
| 98 | + /* use ICU only when max encoding length > one */ |
| 99 | + if (pg_database_encoding_max_length() > 1) |
| 100 | + { |
| 101 | + UChar sourcebuf[STACKBUFLEN], destbuf[STACKBUFLEN]; |
| 102 | + UChar *source, *dest; |
| 103 | + int buflen; |
| 104 | + size_t result_size, usize; |
| 105 | + UErrorCode status = U_ZERO_ERROR; |
| 106 | + |
| 107 | + if (conv == NULL) |
| 108 | + { |
| 109 | + conv = ucnv_open(NULL, &status); |
| 110 | + if (U_FAILURE(status)) |
| 111 | + { |
| 112 | + ereport(ERROR, |
| 113 | + (errcode(status), |
| 114 | + errmsg("ICU error: oracle_compat.c, could not get converter for \"%s\"", ucnv_getDefaultName()))); |
| 115 | + } |
| 116 | + } |
| 117 | + |
| 118 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 119 | + { |
| 120 | + buflen = (nbytes + 1) * sizeof(UChar); |
| 121 | + source = palloc(buflen); |
| 122 | + dest = palloc(buflen); |
| 123 | + } |
| 124 | + else |
| 125 | + { |
| 126 | + buflen = STACKBUFLEN; |
| 127 | + source = sourcebuf; |
| 128 | + dest = destbuf; |
| 129 | + } |
| 130 | + // convert to UTF-16 |
| 131 | + ucnv_toUChars(conv, source, buflen, buff, nbytes, &status); |
| 132 | + if (U_FAILURE(status)) |
| 133 | + { |
| 134 | + ereport(ERROR, |
| 135 | + (errcode(status), |
| 136 | + errmsg("ICU error: Could not convert string"))); |
| 137 | + } |
| 138 | + |
| 139 | + // run desired function |
| 140 | + buflen = u_strToLower(dest, buflen, source, -1, NULL, &status); |
| 141 | + if (U_FAILURE(status)) |
| 142 | + { |
| 143 | + ereport(ERROR, |
| 144 | + (errcode(status), |
| 145 | + errmsg("ICU error: Could not modify case"))); |
| 146 | + } |
| 147 | + |
| 148 | + // and convert modified utf-16 string back to text |
| 149 | + result_size = UCNV_GET_MAX_BYTES_FOR_STRING(buflen, ucnv_getMaxCharSize(conv)); |
| 150 | + result = palloc(result_size); |
| 151 | + |
| 152 | + usize = ucnv_fromUChars(conv, result, result_size, |
| 153 | + dest, buflen, &status); |
| 154 | + |
| 155 | + if (U_FAILURE(status)) |
| 156 | + { |
| 157 | + /* Invalid multibyte character encountered ... shouldn't happen */ |
| 158 | + ereport(ERROR, |
| 159 | + (errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE), |
| 160 | + errmsg("ICU: invalid multibyte character for locale"))); |
| 161 | + } |
| 162 | + |
| 163 | + Assert(usize <= (size_t) (buflen * sizeof(UChar))); |
| 164 | + |
| 165 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 166 | + { |
| 167 | + pfree(source); |
| 168 | + pfree(dest); |
| 169 | + } |
| 170 | + return result; |
| 171 | + } |
| 172 | + else |
| 173 | +#else |
| 174 | #ifdef USE_WIDE_UPPER_LOWER |
| 175 | if (pg_database_encoding_max_length() > 1 && !lc_ctype_is_c()) |
| 176 | { |
| 177 | @@ -1507,6 +1596,7 @@ |
| 178 | } |
| 179 | else |
| 180 | #endif /* USE_WIDE_UPPER_LOWER */ |
| 181 | +#endif /* USE_ICU */ |
| 182 | { |
| 183 | char *p; |
| 184 | |
| 185 | @@ -1533,6 +1623,83 @@ |
| 186 | if (!buff) |
| 187 | return NULL; |
| 188 | |
| 189 | +#ifdef USE_ICU |
| 190 | + /* use ICU only when max encoding length > one */ |
| 191 | + if (pg_database_encoding_max_length() > 1) |
| 192 | + { |
| 193 | + UChar sourcebuf[STACKBUFLEN], destbuf[STACKBUFLEN]; |
| 194 | + UChar *source, *dest; |
| 195 | + int buflen; |
| 196 | + size_t result_size, usize; |
| 197 | + UErrorCode status = U_ZERO_ERROR; |
| 198 | + |
| 199 | + if (conv == NULL) |
| 200 | + { |
| 201 | + conv = ucnv_open(NULL, &status); |
| 202 | + if (U_FAILURE(status)) |
| 203 | + { |
| 204 | + ereport(ERROR, |
| 205 | + (errcode(status), |
| 206 | + errmsg("ICU error: oracle_compat.c, could not get converter for \"%s\"", ucnv_getDefaultName()))); |
| 207 | + } |
| 208 | + } |
| 209 | + |
| 210 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 211 | + { |
| 212 | + buflen = (nbytes + 1) * sizeof(UChar); |
| 213 | + source = palloc(buflen); |
| 214 | + dest = palloc(buflen); |
| 215 | + } |
| 216 | + else |
| 217 | + { |
| 218 | + buflen = STACKBUFLEN; |
| 219 | + source = sourcebuf; |
| 220 | + dest = destbuf; |
| 221 | + } |
| 222 | + // convert to UTF-16 |
| 223 | + ucnv_toUChars(conv, source, buflen, buff, nbytes, &status); |
| 224 | + if (U_FAILURE(status)) |
| 225 | + { |
| 226 | + ereport(ERROR, |
| 227 | + (errcode(status), |
| 228 | + errmsg("ICU error: Could not convert string"))); |
| 229 | + } |
| 230 | + |
| 231 | + // run desired function |
| 232 | + buflen = u_strToUpper(dest, buflen, source, -1, NULL, &status); |
| 233 | + if (U_FAILURE(status)) |
| 234 | + { |
| 235 | + ereport(ERROR, |
| 236 | + (errcode(status), |
| 237 | + errmsg("ICU error: Could not modify case"))); |
| 238 | + } |
| 239 | + |
| 240 | + // and convert modified utf-16 string back to text |
| 241 | + result_size = UCNV_GET_MAX_BYTES_FOR_STRING(buflen, ucnv_getMaxCharSize(conv)); |
| 242 | + result = palloc(result_size); |
| 243 | + |
| 244 | + usize = ucnv_fromUChars(conv, result, result_size, |
| 245 | + dest, buflen, &status); |
| 246 | + |
| 247 | + if (U_FAILURE(status)) |
| 248 | + { |
| 249 | + /* Invalid multibyte character encountered ... shouldn't happen */ |
| 250 | + ereport(ERROR, |
| 251 | + (errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE), |
| 252 | + errmsg("ICU: invalid multibyte character for locale"))); |
| 253 | + } |
| 254 | + |
| 255 | + Assert(usize <= (size_t) (buflen * sizeof(UChar))); |
| 256 | + |
| 257 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 258 | + { |
| 259 | + pfree(source); |
| 260 | + pfree(dest); |
| 261 | + } |
| 262 | + return result; |
| 263 | + } |
| 264 | + else |
| 265 | +#else |
| 266 | #ifdef USE_WIDE_UPPER_LOWER |
| 267 | if (pg_database_encoding_max_length() > 1 && !lc_ctype_is_c()) |
| 268 | { |
| 269 | @@ -1563,6 +1730,7 @@ |
| 270 | } |
| 271 | else |
| 272 | #endif /* USE_WIDE_UPPER_LOWER */ |
| 273 | +#endif /* USE_ICU */ |
| 274 | { |
| 275 | char *p; |
| 276 | |
| 277 | @@ -1590,6 +1758,83 @@ |
| 278 | if (!buff) |
| 279 | return NULL; |
| 280 | |
| 281 | +#ifdef USE_ICU |
| 282 | + /* use ICU only when max encoding length > one */ |
| 283 | + if (pg_database_encoding_max_length() > 1) |
| 284 | + { |
| 285 | + UChar sourcebuf[STACKBUFLEN], destbuf[STACKBUFLEN]; |
| 286 | + UChar *source, *dest; |
| 287 | + int buflen; |
| 288 | + size_t result_size, usize; |
| 289 | + UErrorCode status = U_ZERO_ERROR; |
| 290 | + |
| 291 | + if (conv == NULL) |
| 292 | + { |
| 293 | + conv = ucnv_open(NULL, &status); |
| 294 | + if (U_FAILURE(status)) |
| 295 | + { |
| 296 | + ereport(ERROR, |
| 297 | + (errcode(status), |
| 298 | + errmsg("ICU error: oracle_compat.c, could not get converter for \"%s\"", ucnv_getDefaultName()))); |
| 299 | + } |
| 300 | + } |
| 301 | + |
| 302 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 303 | + { |
| 304 | + buflen = (nbytes + 1) * sizeof(UChar); |
| 305 | + source = palloc(buflen); |
| 306 | + dest = palloc(buflen); |
| 307 | + } |
| 308 | + else |
| 309 | + { |
| 310 | + buflen = STACKBUFLEN; |
| 311 | + source = sourcebuf; |
| 312 | + dest = destbuf; |
| 313 | + } |
| 314 | + // convert to UTF-16 |
| 315 | + ucnv_toUChars(conv, source, buflen, buff, nbytes, &status); |
| 316 | + if (U_FAILURE(status)) |
| 317 | + { |
| 318 | + ereport(ERROR, |
| 319 | + (errcode(status), |
| 320 | + errmsg("ICU error: Could not convert string"))); |
| 321 | + } |
| 322 | + |
| 323 | + // run desired function |
| 324 | + buflen = u_strToTitle(dest, buflen, source, -1, NULL, NULL, &status); |
| 325 | + if (U_FAILURE(status)) |
| 326 | + { |
| 327 | + ereport(ERROR, |
| 328 | + (errcode(status), |
| 329 | + errmsg("ICU error: Could not modify case"))); |
| 330 | + } |
| 331 | + |
| 332 | + // and convert modified utf-16 string back to text |
| 333 | + result_size = UCNV_GET_MAX_BYTES_FOR_STRING(buflen, ucnv_getMaxCharSize(conv)); |
| 334 | + result = palloc(result_size); |
| 335 | + |
| 336 | + usize = ucnv_fromUChars(conv, result, result_size, |
| 337 | + dest, buflen, &status); |
| 338 | + |
| 339 | + if (U_FAILURE(status)) |
| 340 | + { |
| 341 | + /* Invalid multibyte character encountered ... shouldn't happen */ |
| 342 | + ereport(ERROR, |
| 343 | + (errcode(ERRCODE_CHARACTER_NOT_IN_REPERTOIRE), |
| 344 | + errmsg("ICU: invalid multibyte character for locale"))); |
| 345 | + } |
| 346 | + |
| 347 | + Assert(usize <= (size_t) (buflen * sizeof(UChar))); |
| 348 | + |
| 349 | + if (nbytes >= STACKBUFLEN / sizeof(UChar)) |
| 350 | + { |
| 351 | + pfree(source); |
| 352 | + pfree(dest); |
| 353 | + } |
| 354 | + return result; |
| 355 | + } |
| 356 | + else |
| 357 | +#else |
| 358 | #ifdef USE_WIDE_UPPER_LOWER |
| 359 | if (pg_database_encoding_max_length() > 1 && !lc_ctype_is_c()) |
| 360 | { |
| 361 | @@ -1626,6 +1871,7 @@ |
| 362 | } |
| 363 | else |
| 364 | #endif /* USE_WIDE_UPPER_LOWER */ |
| 365 | +#endif /* USE_ICU */ |
| 366 | { |
| 367 | char *p; |
| 368 | |
| 369 | diff -Naur postgresql-9.0.6-orig/src/backend/utils/adt/varlena.c postgresql-9.0.6/src/backend/utils/adt/varlena.c |
| 370 | --- src/backend/utils/adt/varlena.c 2012-02-03 11:56:34.000000000 +1100 |
| 371 | +++ src/backend/utils/adt/varlena.c 2012-02-03 12:02:54.000000000 +1100 |
| 372 | @@ -28,6 +28,15 @@ |
| 373 | #include "utils/lsyscache.h" |
| 374 | #include "utils/pg_locale.h" |
| 375 | |
| 376 | +#ifdef USE_ICU |
| 377 | +#include <unicode/utypes.h> /* Basic ICU data types */ |
| 378 | +#include <unicode/ucnv.h> /* C Converter API */ |
| 379 | +#include <unicode/ucol.h> |
| 380 | +#include <unicode/uloc.h> |
| 381 | +#include "unicode/uiter.h" |
| 382 | +#define USTACKBUFLEN STACKBUFLEN / sizeof(UChar) |
| 383 | +#endif /* USE_ICU */ |
| 384 | + |
| 385 | |
| 386 | /* GUC variable */ |
| 387 | int bytea_output = BYTEA_OUTPUT_HEX; |
| 388 | @@ -1287,6 +1296,134 @@ |
| 389 | { |
| 390 | #define STACKBUFLEN 1024 |
| 391 | |
| 392 | +#ifdef USE_ICU |
| 393 | + |
| 394 | + if (pg_database_encoding_max_length() > 1) |
| 395 | + { |
| 396 | + static UCollator * collator = NULL; |
| 397 | + UErrorCode status = U_ZERO_ERROR; |
| 398 | + |
| 399 | + /* We keep a static collator "forever", since it is hard |
| 400 | + * coded into the database cluster at initdb time |
| 401 | + * anyway. Create it first time we get here. */ |
| 402 | + if (collator == NULL) |
| 403 | + { |
| 404 | + /* Expect LC_COLLATE to be set to something that ICU |
| 405 | + * will understand. This is quite probable, since ICU |
| 406 | + * does a lot of heuristics with this argument. I'd |
| 407 | + * rather set this in xlog.c, but it seems ICU forgets |
| 408 | + * it??? */ |
| 409 | + uloc_setDefault(setlocale(LC_COLLATE, NULL), &status); |
| 410 | + if(U_FAILURE(status)) |
| 411 | + { |
| 412 | + ereport(WARNING, |
| 413 | + (errcode(status), |
| 414 | + errmsg("ICU Error: varlena.c, could not set default lc_collate"))); |
| 415 | + } |
| 416 | + collator = ucol_open(NULL, &status); |
| 417 | + if (U_FAILURE(status)) |
| 418 | + { |
| 419 | + ereport(WARNING, |
| 420 | + (errcode(status), |
| 421 | + errmsg("ICU Error: varlena.c, could not open collator"))); |
| 422 | + } |
| 423 | + } |
| 424 | + |
| 425 | + if (GetDatabaseEncoding() == PG_UTF8) |
| 426 | + { |
| 427 | + UCharIterator sIter, tIter; |
| 428 | + uiter_setUTF8(&sIter, arg1, len1); |
| 429 | + uiter_setUTF8(&tIter, arg2, len2); |
| 430 | + result = ucol_strcollIter(collator, &sIter, &tIter, &status); |
| 431 | + if (U_FAILURE(status)) |
| 432 | + { |
| 433 | + ereport(WARNING, |
| 434 | + (errcode(status), |
| 435 | + errmsg("ICU Error: varlena.c, could not collate"))); |
| 436 | + } |
| 437 | + } |
| 438 | + else { |
| 439 | + /* We keep a static converter "forever". |
| 440 | + * Create it first time we get here. */ |
| 441 | + static UConverter * conv = NULL; |
| 442 | + if (conv == NULL) |
| 443 | + { |
| 444 | + conv = ucnv_open(NULL, &status); |
| 445 | + if (U_FAILURE(status) || conv == NULL) |
| 446 | + { |
| 447 | + ereport(ERROR, |
| 448 | + (errcode(status), |
| 449 | + errmsg("ICU error: varlena.c, could not get converter for \"%s\"", ucnv_getDefaultName()))); |
| 450 | + } |
| 451 | + } |
| 452 | + |
| 453 | + UChar a1buf[USTACKBUFLEN], |
| 454 | + a2buf[USTACKBUFLEN]; |
| 455 | + int a1len = USTACKBUFLEN, |
| 456 | + a2len = USTACKBUFLEN; |
| 457 | + UChar *a1p, |
| 458 | + *a2p; |
| 459 | + if (len1 >= USTACKBUFLEN / sizeof(UChar)) |
| 460 | + { |
| 461 | + a1len = (len1 + 1) * sizeof(UChar); |
| 462 | + a1p = (UChar *) palloc(a1len); |
| 463 | + } |
| 464 | + else |
| 465 | + a1p = a1buf; |
| 466 | + |
| 467 | + if (len2 >= USTACKBUFLEN / sizeof(UChar)) |
| 468 | + { |
| 469 | + a2len = (len2 + 1) * sizeof(UChar); |
| 470 | + a2p = (UChar *) palloc(a2len); |
| 471 | + } |
| 472 | + else |
| 473 | + a2p = a2buf; |
| 474 | + |
| 475 | + ucnv_toUChars(conv, a1p, a1len, arg1, len1, &status); |
| 476 | + if(U_FAILURE(status)) |
| 477 | + { |
| 478 | + ereport(WARNING, |
| 479 | + (errcode(status), |
| 480 | + errmsg("ICU Error: varlena.c, could not convert to UChars"))); |
| 481 | + } |
| 482 | + ucnv_toUChars(conv, a2p, a2len, arg2, len2, &status); |
| 483 | + if(U_FAILURE(status)) |
| 484 | + { |
| 485 | + ereport(WARNING, |
| 486 | + (errcode(status), |
| 487 | + errmsg("ICU Error: varlena.c, could not convert to UChars"))); |
| 488 | + } |
| 489 | + |
| 490 | + result = ucol_strcoll(collator, a1p, -1, a2p, -1); |
| 491 | + if(U_FAILURE(status)) |
| 492 | + { |
| 493 | + ereport(WARNING, |
| 494 | + (errcode(status), |
| 495 | + errmsg("ICU Error: varlena.c, could not collate"))); |
| 496 | + } |
| 497 | + if (len1 * sizeof(UChar) >= USTACKBUFLEN) |
| 498 | + pfree(a1p); |
| 499 | + if (len2 * sizeof(UChar) >= USTACKBUFLEN) |
| 500 | + pfree(a2p); |
| 501 | + } |
| 502 | + /* |
| 503 | + * In some locales wcscoll() can claim that nonidentical strings |
| 504 | + * are equal. Believing that this might be so also for ICU, and |
| 505 | + * believing that would be bad news for a number of |
| 506 | + * reasons, we follow Perl's lead and sort "equal" strings |
| 507 | + * according to strcmp (on the byte representation). |
| 508 | + */ |
| 509 | + if (result == 0) |
| 510 | + { |
| 511 | + result = strncmp(arg1, arg2, Min(len1, len2)); |
| 512 | + if ((result == 0) && (len1 != len2)) |
| 513 | + result = (len1 < len2) ? -1 : 1; |
| 514 | + } |
| 515 | + |
| 516 | + return result; |
| 517 | + } |
| 518 | +#endif /* USE_ICU */ |
| 519 | + |
| 520 | char a1buf[STACKBUFLEN]; |
| 521 | char a2buf[STACKBUFLEN]; |
| 522 | char *a1p, |
| 523 | diff -Naur postgresql-9.0.6-orig/src/backend/utils/mb/encnames.c postgresql-9.0.6/src/backend/utils/mb/encnames.c |
| 524 | --- src/backend/utils/mb/encnames.c 2012-02-03 11:56:34.000000000 +1100 |
| 525 | +++ src/backend/utils/mb/encnames.c 2012-02-03 12:02:54.000000000 +1100 |
| 526 | @@ -394,6 +394,118 @@ |
| 527 | }; |
| 528 | |
| 529 | |
| 530 | +#ifdef USE_ICU |
| 531 | +/* |
| 532 | + * Try to map most internal character encodings to the proper and |
| 533 | + * preferred IANA string. Use this in mbutils.c to feed ICU info about |
| 534 | + * the database's character encoding. |
| 535 | + * |
| 536 | + * Palle Girgensohn, 2005 |
| 537 | + */ |
| 538 | + |
| 539 | +pg_enc2name pg_enc2iananame_tbl[] = |
| 540 | +{ |
| 541 | + { |
| 542 | + "US-ASCII", PG_SQL_ASCII |
| 543 | + }, |
| 544 | + { |
| 545 | + "EUC-JP", PG_EUC_JP |
| 546 | + }, |
| 547 | + { |
| 548 | + "GB2312", PG_EUC_CN |
| 549 | + }, |
| 550 | + { |
| 551 | + "EUC-KR", PG_EUC_KR |
| 552 | + }, |
| 553 | + { |
| 554 | + "ISO-2022-CN", PG_EUC_TW |
| 555 | + }, |
| 556 | + { |
| 557 | + "KS_C_5601-1987", PG_JOHAB /* either KS_C_5601-1987 or ISO-2022-KR ??? */ |
| 558 | + }, |
| 559 | + { |
| 560 | + "UTF-8", PG_UTF8 |
| 561 | + }, |
| 562 | + { |
| 563 | + "MULE_INTERNAL", PG_MULE_INTERNAL /* is not for real */ |
| 564 | + }, |
| 565 | + { |
| 566 | + "ISO-8859-1", PG_LATIN1 |
| 567 | + }, |
| 568 | + { |
| 569 | + "ISO-8859-2", PG_LATIN2 |
| 570 | + }, |
| 571 | + { |
| 572 | + "ISO-8859-3", PG_LATIN3 |
| 573 | + }, |
| 574 | + { |
| 575 | + "ISO-8859-4", PG_LATIN4 |
| 576 | + }, |
| 577 | + { |
| 578 | + "ISO-8859-9", PG_LATIN5 |
| 579 | + }, |
| 580 | + { |
| 581 | + "ISO-8859-10", PG_LATIN6 |
| 582 | + }, |
| 583 | + { |
| 584 | + "ISO-8859-13", PG_LATIN7 |
| 585 | + }, |
| 586 | + { |
| 587 | + "ISO-8859-14", PG_LATIN8 |
| 588 | + }, |
| 589 | + { |
| 590 | + "ISO-8859-15", PG_LATIN9 |
| 591 | + }, |
| 592 | + { |
| 593 | + "ISO-8859-16", PG_LATIN10 |
| 594 | + }, |
| 595 | + { |
| 596 | + "windows-1256", PG_WIN1256 |
| 597 | + }, |
| 598 | + { |
| 599 | + "windows-874", PG_WIN874 |
| 600 | + }, |
| 601 | + { |
| 602 | + "KOI8-R", PG_KOI8R |
| 603 | + }, |
| 604 | + { |
| 605 | + "windows-1251", PG_WIN1251 |
| 606 | + }, |
| 607 | + { |
| 608 | + "ISO-8859-5", PG_ISO_8859_5 |
| 609 | + }, |
| 610 | + { |
| 611 | + "ISO-8859-6", PG_ISO_8859_6 |
| 612 | + }, |
| 613 | + { |
| 614 | + "ISO-8859-7", PG_ISO_8859_7 |
| 615 | + }, |
| 616 | + { |
| 617 | + "ISO-8859-8", PG_ISO_8859_8 |
| 618 | + }, |
| 619 | + { |
| 620 | + "windows-1250", PG_WIN1250 |
| 621 | + }, |
| 622 | + { |
| 623 | + "Shift_JIS", PG_SJIS |
| 624 | + }, |
| 625 | + { |
| 626 | + "Big5", PG_BIG5 |
| 627 | + }, |
| 628 | + { |
| 629 | + "GBK", PG_GBK |
| 630 | + }, |
| 631 | + { |
| 632 | + "cp949", PG_UHC |
| 633 | + }, |
| 634 | + { |
| 635 | + "GB18030", PG_GB18030 |
| 636 | + } |
| 637 | +}; |
| 638 | +#endif /* USE_ICU */ |
| 639 | + |
| 640 | + |
| 641 | + |
| 642 | /* ---------- |
| 643 | * Encoding checks, for error returns -1 else encoding id |
| 644 | * ---------- |
| 645 | diff -Naur postgresql-9.0.6-orig/src/backend/utils/mb/mbutils.c postgresql-9.0.6/src/backend/utils/mb/mbutils.c |
| 646 | --- src/backend/utils/mb/mbutils.c 2012-02-03 11:56:34.000000000 +1100 |
| 647 | +++ src/backend/utils/mb/mbutils.c 2012-02-03 12:02:54.000000000 +1100 |
| 648 | @@ -15,6 +15,9 @@ |
| 649 | #include "utils/memutils.h" |
| 650 | #include "utils/pg_locale.h" |
| 651 | #include "utils/syscache.h" |
| 652 | +#ifdef USE_ICU |
| 653 | +#include <unicode/ucnv.h> |
| 654 | +#endif /* USE_ICU */ |
| 655 | |
| 656 | /* |
| 657 | * When converting strings between different encodings, we assume that space |
| 658 | @@ -908,6 +911,9 @@ |
| 659 | |
| 660 | DatabaseEncoding = &pg_enc2name_tbl[encoding]; |
| 661 | Assert(DatabaseEncoding->encoding == encoding); |
| 662 | +#ifdef USE_ICU |
| 663 | + ucnv_setDefaultName((&pg_enc2iananame_tbl[encoding])->name); |
| 664 | +#endif |
| 665 | } |
| 666 | |
| 667 | /* |
| 668 | diff -Naur postgresql-9.0.6-orig/src/include/mb/pg_wchar.h postgresql-9.0.6/src/include/mb/pg_wchar.h |
| 669 | --- src/include/mb/pg_wchar.h 2012-02-03 11:56:34.000000000 +1100 |
| 670 | +++ src/include/mb/pg_wchar.h 2012-02-03 12:02:54.000000000 +1100 |
| 671 | @@ -263,6 +263,9 @@ |
| 672 | } pg_enc2name; |
| 673 | |
| 674 | extern pg_enc2name pg_enc2name_tbl[]; |
| 675 | +#ifdef USE_ICU |
| 676 | +extern pg_enc2name pg_enc2iananame_tbl[]; |
| 677 | +#endif |
| 678 | |
| 679 | /* |
| 680 | * Encoding names for gettext |
| 681 | diff -Naur postgresql-9.0.6-orig/src/include/pg_config.h.in postgresql-9.0.6/src/include/pg_config.h.in |
| 682 | --- src/include/pg_config.h.in 2012-02-03 11:56:34.000000000 +1100 |
| 683 | +++ src/include/pg_config.h.in 2012-02-03 12:02:54.000000000 +1100 |
| 684 | @@ -290,6 +290,12 @@ |
| 685 | /* Define to 1 if you have the `eay32' library (-leay32). */ |
| 686 | #undef HAVE_LIBEAY32 |
| 687 | |
| 688 | +/* Define to 1 if you have the `icui18n' library (-licui18n). */ |
| 689 | +#undef HAVE_LIBICUI18N |
| 690 | + |
| 691 | +/* Define to 1 if you have the `icuuc' library (-licuuc). */ |
| 692 | +#undef HAVE_LIBICUUC |
| 693 | + |
| 694 | /* Define to 1 if you have the `ldap' library (-lldap). */ |
| 695 | #undef HAVE_LIBLDAP |
| 696 | |
| 697 | @@ -747,6 +753,9 @@ |
| 698 | /* Define to 1 to build with Bonjour support. (--with-bonjour) */ |
| 699 | #undef USE_BONJOUR |
| 700 | |
| 701 | +/* Define to build with ICU support. (--with-icu) */ |
| 702 | +#undef USE_ICU |
| 703 | + |
| 704 | /* Define to 1 if you want float4 values to be passed by value. |
| 705 | (--enable-float4-byval) */ |
| 706 | #undef USE_FLOAT4_BYVAL |