1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/putil.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,2315 @@
1.4 +/*
1.5 +******************************************************************************
1.6 +*
1.7 +* Copyright (C) 1997-2013, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +******************************************************************************
1.11 +*
1.12 +* FILE NAME : putil.c (previously putil.cpp and ptypes.cpp)
1.13 +*
1.14 +* Date Name Description
1.15 +* 04/14/97 aliu Creation.
1.16 +* 04/24/97 aliu Added getDefaultDataDirectory() and
1.17 +* getDefaultLocaleID().
1.18 +* 04/28/97 aliu Rewritten to assume Unix and apply general methods
1.19 +* for assumed case. Non-UNIX platforms must be
1.20 +* special-cased. Rewrote numeric methods dealing
1.21 +* with NaN and Infinity to be platform independent
1.22 +* over all IEEE 754 platforms.
1.23 +* 05/13/97 aliu Restored sign of timezone
1.24 +* (semantics are hours West of GMT)
1.25 +* 06/16/98 erm Added IEEE_754 stuff, cleaned up isInfinite, isNan,
1.26 +* nextDouble..
1.27 +* 07/22/98 stephen Added remainder, max, min, trunc
1.28 +* 08/13/98 stephen Added isNegativeInfinity, isPositiveInfinity
1.29 +* 08/24/98 stephen Added longBitsFromDouble
1.30 +* 09/08/98 stephen Minor changes for Mac Port
1.31 +* 03/02/99 stephen Removed openFile(). Added AS400 support.
1.32 +* Fixed EBCDIC tables
1.33 +* 04/15/99 stephen Converted to C.
1.34 +* 06/28/99 stephen Removed mutex locking in u_isBigEndian().
1.35 +* 08/04/99 jeffrey R. Added OS/2 changes
1.36 +* 11/15/99 helena Integrated S/390 IEEE support.
1.37 +* 04/26/01 Barry N. OS/400 support for uprv_getDefaultLocaleID
1.38 +* 08/15/01 Steven H. OS/400 support for uprv_getDefaultCodepage
1.39 +* 01/03/08 Steven L. Fake Time Support
1.40 +******************************************************************************
1.41 +*/
1.42 +
1.43 +// Defines _XOPEN_SOURCE for access to POSIX functions.
1.44 +// Must be before any other #includes.
1.45 +#include "uposixdefs.h"
1.46 +
1.47 +/* include ICU headers */
1.48 +#include "unicode/utypes.h"
1.49 +#include "unicode/putil.h"
1.50 +#include "unicode/ustring.h"
1.51 +#include "putilimp.h"
1.52 +#include "uassert.h"
1.53 +#include "umutex.h"
1.54 +#include "cmemory.h"
1.55 +#include "cstring.h"
1.56 +#include "locmap.h"
1.57 +#include "ucln_cmn.h"
1.58 +
1.59 +/* Include standard headers. */
1.60 +#include <stdio.h>
1.61 +#include <stdlib.h>
1.62 +#include <string.h>
1.63 +#include <math.h>
1.64 +#include <locale.h>
1.65 +#include <float.h>
1.66 +
1.67 +#ifndef U_COMMON_IMPLEMENTATION
1.68 +#error U_COMMON_IMPLEMENTATION not set - must be set for all ICU source files in common/ - see http://userguide.icu-project.org/howtouseicu
1.69 +#endif
1.70 +
1.71 +
1.72 +/* include system headers */
1.73 +#if U_PLATFORM_USES_ONLY_WIN32_API
1.74 + /*
1.75 + * TODO: U_PLATFORM_USES_ONLY_WIN32_API includes MinGW.
1.76 + * Should Cygwin be included as well (U_PLATFORM_HAS_WIN32_API)
1.77 + * to use native APIs as much as possible?
1.78 + */
1.79 +# define WIN32_LEAN_AND_MEAN
1.80 +# define VC_EXTRALEAN
1.81 +# define NOUSER
1.82 +# define NOSERVICE
1.83 +# define NOIME
1.84 +# define NOMCX
1.85 +# include <windows.h>
1.86 +# include "wintz.h"
1.87 +#elif U_PLATFORM == U_PF_OS400
1.88 +# include <float.h>
1.89 +# include <qusec.h> /* error code structure */
1.90 +# include <qusrjobi.h>
1.91 +# include <qliept.h> /* EPT_CALL macro - this include must be after all other "QSYSINCs" */
1.92 +# include <mih/testptr.h> /* For uprv_maximumPtr */
1.93 +#elif U_PLATFORM == U_PF_CLASSIC_MACOS
1.94 +# include <Files.h>
1.95 +# include <IntlResources.h>
1.96 +# include <Script.h>
1.97 +# include <Folders.h>
1.98 +# include <MacTypes.h>
1.99 +# include <TextUtils.h>
1.100 +# define ICU_NO_USER_DATA_OVERRIDE 1
1.101 +#elif U_PLATFORM == U_PF_OS390
1.102 +# include "unicode/ucnv.h" /* Needed for UCNV_SWAP_LFNL_OPTION_STRING */
1.103 +#elif U_PLATFORM_IS_DARWIN_BASED || U_PLATFORM_IS_LINUX_BASED || U_PLATFORM == U_PF_BSD || U_PLATFORM == U_PF_SOLARIS
1.104 +# include <limits.h>
1.105 +# include <unistd.h>
1.106 +# if U_PLATFORM == U_PF_SOLARIS
1.107 +# ifndef _XPG4_2
1.108 +# define _XPG4_2
1.109 +# endif
1.110 +# endif
1.111 +#elif U_PLATFORM == U_PF_QNX
1.112 +# include <sys/neutrino.h>
1.113 +#endif
1.114 +
1.115 +#if (U_PF_MINGW <= U_PLATFORM && U_PLATFORM <= U_PF_CYGWIN) && defined(__STRICT_ANSI__)
1.116 +/* tzset isn't defined in strict ANSI on Cygwin and MinGW. */
1.117 +#undef __STRICT_ANSI__
1.118 +#endif
1.119 +
1.120 +/*
1.121 + * Cygwin with GCC requires inclusion of time.h after the above disabling strict asci mode statement.
1.122 + */
1.123 +#include <time.h>
1.124 +
1.125 +#if !U_PLATFORM_USES_ONLY_WIN32_API
1.126 +#include <sys/time.h>
1.127 +#endif
1.128 +
1.129 +/*
1.130 + * Only include langinfo.h if we have a way to get the codeset. If we later
1.131 + * depend on more feature, we can test on U_HAVE_NL_LANGINFO.
1.132 + *
1.133 + */
1.134 +
1.135 +#if U_HAVE_NL_LANGINFO_CODESET
1.136 +#include <langinfo.h>
1.137 +#endif
1.138 +
1.139 +/**
1.140 + * Simple things (presence of functions, etc) should just go in configure.in and be added to
1.141 + * icucfg.h via autoheader.
1.142 + */
1.143 +#if U_PLATFORM_IMPLEMENTS_POSIX
1.144 +# if U_PLATFORM == U_PF_OS400
1.145 +# define HAVE_DLFCN_H 0
1.146 +# define HAVE_DLOPEN 0
1.147 +# else
1.148 +# ifndef HAVE_DLFCN_H
1.149 +# define HAVE_DLFCN_H 1
1.150 +# endif
1.151 +# ifndef HAVE_DLOPEN
1.152 +# define HAVE_DLOPEN 1
1.153 +# endif
1.154 +# endif
1.155 +# ifndef HAVE_GETTIMEOFDAY
1.156 +# define HAVE_GETTIMEOFDAY 1
1.157 +# endif
1.158 +#else
1.159 +# define HAVE_DLFCN_H 0
1.160 +# define HAVE_DLOPEN 0
1.161 +# define HAVE_GETTIMEOFDAY 0
1.162 +#endif
1.163 +
1.164 +#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
1.165 +
1.166 +/* Define the extension for data files, again... */
1.167 +#define DATA_TYPE "dat"
1.168 +
1.169 +/* Leave this copyright notice here! */
1.170 +static const char copyright[] = U_COPYRIGHT_STRING;
1.171 +
1.172 +/* floating point implementations ------------------------------------------- */
1.173 +
1.174 +/* We return QNAN rather than SNAN*/
1.175 +#define SIGN 0x80000000U
1.176 +
1.177 +/* Make it easy to define certain types of constants */
1.178 +typedef union {
1.179 + int64_t i64; /* This must be defined first in order to allow the initialization to work. This is a C89 feature. */
1.180 + double d64;
1.181 +} BitPatternConversion;
1.182 +static const BitPatternConversion gNan = { (int64_t) INT64_C(0x7FF8000000000000) };
1.183 +static const BitPatternConversion gInf = { (int64_t) INT64_C(0x7FF0000000000000) };
1.184 +
1.185 +/*---------------------------------------------------------------------------
1.186 + Platform utilities
1.187 + Our general strategy is to assume we're on a POSIX platform. Platforms which
1.188 + are non-POSIX must declare themselves so. The default POSIX implementation
1.189 + will sometimes work for non-POSIX platforms as well (e.g., the NaN-related
1.190 + functions).
1.191 + ---------------------------------------------------------------------------*/
1.192 +
1.193 +#if U_PLATFORM_USES_ONLY_WIN32_API || U_PLATFORM == U_PF_CLASSIC_MACOS || U_PLATFORM == U_PF_OS400
1.194 +# undef U_POSIX_LOCALE
1.195 +#else
1.196 +# define U_POSIX_LOCALE 1
1.197 +#endif
1.198 +
1.199 +/*
1.200 + WARNING! u_topNBytesOfDouble and u_bottomNBytesOfDouble
1.201 + can't be properly optimized by the gcc compiler sometimes (i.e. gcc 3.2).
1.202 +*/
1.203 +#if !IEEE_754
1.204 +static char*
1.205 +u_topNBytesOfDouble(double* d, int n)
1.206 +{
1.207 +#if U_IS_BIG_ENDIAN
1.208 + return (char*)d;
1.209 +#else
1.210 + return (char*)(d + 1) - n;
1.211 +#endif
1.212 +}
1.213 +
1.214 +static char*
1.215 +u_bottomNBytesOfDouble(double* d, int n)
1.216 +{
1.217 +#if U_IS_BIG_ENDIAN
1.218 + return (char*)(d + 1) - n;
1.219 +#else
1.220 + return (char*)d;
1.221 +#endif
1.222 +}
1.223 +#endif /* !IEEE_754 */
1.224 +
1.225 +#if IEEE_754
1.226 +static UBool
1.227 +u_signBit(double d) {
1.228 + uint8_t hiByte;
1.229 +#if U_IS_BIG_ENDIAN
1.230 + hiByte = *(uint8_t *)&d;
1.231 +#else
1.232 + hiByte = *(((uint8_t *)&d) + sizeof(double) - 1);
1.233 +#endif
1.234 + return (hiByte & 0x80) != 0;
1.235 +}
1.236 +#endif
1.237 +
1.238 +
1.239 +
1.240 +#if defined (U_DEBUG_FAKETIME)
1.241 +/* Override the clock to test things without having to move the system clock.
1.242 + * Assumes POSIX gettimeofday() will function
1.243 + */
1.244 +UDate fakeClock_t0 = 0; /** Time to start the clock from **/
1.245 +UDate fakeClock_dt = 0; /** Offset (fake time - real time) **/
1.246 +UBool fakeClock_set = FALSE; /** True if fake clock has spun up **/
1.247 +static UMutex fakeClockMutex = U_MUTEX_INTIALIZER;
1.248 +
1.249 +static UDate getUTCtime_real() {
1.250 + struct timeval posixTime;
1.251 + gettimeofday(&posixTime, NULL);
1.252 + return (UDate)(((int64_t)posixTime.tv_sec * U_MILLIS_PER_SECOND) + (posixTime.tv_usec/1000));
1.253 +}
1.254 +
1.255 +static UDate getUTCtime_fake() {
1.256 + umtx_lock(&fakeClockMutex);
1.257 + if(!fakeClock_set) {
1.258 + UDate real = getUTCtime_real();
1.259 + const char *fake_start = getenv("U_FAKETIME_START");
1.260 + if((fake_start!=NULL) && (fake_start[0]!=0)) {
1.261 + sscanf(fake_start,"%lf",&fakeClock_t0);
1.262 + fakeClock_dt = fakeClock_t0 - real;
1.263 + fprintf(stderr,"U_DEBUG_FAKETIME was set at compile time, so the ICU clock will start at a preset value\n"
1.264 + "env variable U_FAKETIME_START=%.0f (%s) for an offset of %.0f ms from the current time %.0f\n",
1.265 + fakeClock_t0, fake_start, fakeClock_dt, real);
1.266 + } else {
1.267 + fakeClock_dt = 0;
1.268 + fprintf(stderr,"U_DEBUG_FAKETIME was set at compile time, but U_FAKETIME_START was not set.\n"
1.269 + "Set U_FAKETIME_START to the number of milliseconds since 1/1/1970 to set the ICU clock.\n");
1.270 + }
1.271 + fakeClock_set = TRUE;
1.272 + }
1.273 + umtx_unlock(&fakeClockMutex);
1.274 +
1.275 + return getUTCtime_real() + fakeClock_dt;
1.276 +}
1.277 +#endif
1.278 +
1.279 +#if U_PLATFORM_USES_ONLY_WIN32_API
1.280 +typedef union {
1.281 + int64_t int64;
1.282 + FILETIME fileTime;
1.283 +} FileTimeConversion; /* This is like a ULARGE_INTEGER */
1.284 +
1.285 +/* Number of 100 nanoseconds from 1/1/1601 to 1/1/1970 */
1.286 +#define EPOCH_BIAS INT64_C(116444736000000000)
1.287 +#define HECTONANOSECOND_PER_MILLISECOND 10000
1.288 +
1.289 +#endif
1.290 +
1.291 +/*---------------------------------------------------------------------------
1.292 + Universal Implementations
1.293 + These are designed to work on all platforms. Try these, and if they
1.294 + don't work on your platform, then special case your platform with new
1.295 + implementations.
1.296 +---------------------------------------------------------------------------*/
1.297 +
1.298 +U_CAPI UDate U_EXPORT2
1.299 +uprv_getUTCtime()
1.300 +{
1.301 +#if defined(U_DEBUG_FAKETIME)
1.302 + return getUTCtime_fake(); /* Hook for overriding the clock */
1.303 +#else
1.304 + return uprv_getRawUTCtime();
1.305 +#endif
1.306 +}
1.307 +
1.308 +/* Return UTC (GMT) time measured in milliseconds since 0:00 on 1/1/70.*/
1.309 +U_CAPI UDate U_EXPORT2
1.310 +uprv_getRawUTCtime()
1.311 +{
1.312 +#if U_PLATFORM == U_PF_CLASSIC_MACOS
1.313 + time_t t, t1, t2;
1.314 + struct tm tmrec;
1.315 +
1.316 + uprv_memset( &tmrec, 0, sizeof(tmrec) );
1.317 + tmrec.tm_year = 70;
1.318 + tmrec.tm_mon = 0;
1.319 + tmrec.tm_mday = 1;
1.320 + t1 = mktime(&tmrec); /* seconds of 1/1/1970*/
1.321 +
1.322 + time(&t);
1.323 + uprv_memcpy( &tmrec, gmtime(&t), sizeof(tmrec) );
1.324 + t2 = mktime(&tmrec); /* seconds of current GMT*/
1.325 + return (UDate)(t2 - t1) * U_MILLIS_PER_SECOND; /* GMT (or UTC) in seconds since 1970*/
1.326 +#elif U_PLATFORM_USES_ONLY_WIN32_API
1.327 +
1.328 + FileTimeConversion winTime;
1.329 + GetSystemTimeAsFileTime(&winTime.fileTime);
1.330 + return (UDate)((winTime.int64 - EPOCH_BIAS) / HECTONANOSECOND_PER_MILLISECOND);
1.331 +#else
1.332 +
1.333 +#if HAVE_GETTIMEOFDAY
1.334 + struct timeval posixTime;
1.335 + gettimeofday(&posixTime, NULL);
1.336 + return (UDate)(((int64_t)posixTime.tv_sec * U_MILLIS_PER_SECOND) + (posixTime.tv_usec/1000));
1.337 +#else
1.338 + time_t epochtime;
1.339 + time(&epochtime);
1.340 + return (UDate)epochtime * U_MILLIS_PER_SECOND;
1.341 +#endif
1.342 +
1.343 +#endif
1.344 +}
1.345 +
1.346 +/*-----------------------------------------------------------------------------
1.347 + IEEE 754
1.348 + These methods detect and return NaN and infinity values for doubles
1.349 + conforming to IEEE 754. Platforms which support this standard include X86,
1.350 + Mac 680x0, Mac PowerPC, AIX RS/6000, and most others.
1.351 + If this doesn't work on your platform, you have non-IEEE floating-point, and
1.352 + will need to code your own versions. A naive implementation is to return 0.0
1.353 + for getNaN and getInfinity, and false for isNaN and isInfinite.
1.354 + ---------------------------------------------------------------------------*/
1.355 +
1.356 +U_CAPI UBool U_EXPORT2
1.357 +uprv_isNaN(double number)
1.358 +{
1.359 +#if IEEE_754
1.360 + BitPatternConversion convertedNumber;
1.361 + convertedNumber.d64 = number;
1.362 + /* Infinity is 0x7FF0000000000000U. Anything greater than that is a NaN */
1.363 + return (UBool)((convertedNumber.i64 & U_INT64_MAX) > gInf.i64);
1.364 +
1.365 +#elif U_PLATFORM == U_PF_OS390
1.366 + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number,
1.367 + sizeof(uint32_t));
1.368 + uint32_t lowBits = *(uint32_t*)u_bottomNBytesOfDouble(&number,
1.369 + sizeof(uint32_t));
1.370 +
1.371 + return ((highBits & 0x7F080000L) == 0x7F080000L) &&
1.372 + (lowBits == 0x00000000L);
1.373 +
1.374 +#else
1.375 + /* If your platform doesn't support IEEE 754 but *does* have an NaN value,*/
1.376 + /* you'll need to replace this default implementation with what's correct*/
1.377 + /* for your platform.*/
1.378 + return number != number;
1.379 +#endif
1.380 +}
1.381 +
1.382 +U_CAPI UBool U_EXPORT2
1.383 +uprv_isInfinite(double number)
1.384 +{
1.385 +#if IEEE_754
1.386 + BitPatternConversion convertedNumber;
1.387 + convertedNumber.d64 = number;
1.388 + /* Infinity is exactly 0x7FF0000000000000U. */
1.389 + return (UBool)((convertedNumber.i64 & U_INT64_MAX) == gInf.i64);
1.390 +#elif U_PLATFORM == U_PF_OS390
1.391 + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number,
1.392 + sizeof(uint32_t));
1.393 + uint32_t lowBits = *(uint32_t*)u_bottomNBytesOfDouble(&number,
1.394 + sizeof(uint32_t));
1.395 +
1.396 + return ((highBits & ~SIGN) == 0x70FF0000L) && (lowBits == 0x00000000L);
1.397 +
1.398 +#else
1.399 + /* If your platform doesn't support IEEE 754 but *does* have an infinity*/
1.400 + /* value, you'll need to replace this default implementation with what's*/
1.401 + /* correct for your platform.*/
1.402 + return number == (2.0 * number);
1.403 +#endif
1.404 +}
1.405 +
1.406 +U_CAPI UBool U_EXPORT2
1.407 +uprv_isPositiveInfinity(double number)
1.408 +{
1.409 +#if IEEE_754 || U_PLATFORM == U_PF_OS390
1.410 + return (UBool)(number > 0 && uprv_isInfinite(number));
1.411 +#else
1.412 + return uprv_isInfinite(number);
1.413 +#endif
1.414 +}
1.415 +
1.416 +U_CAPI UBool U_EXPORT2
1.417 +uprv_isNegativeInfinity(double number)
1.418 +{
1.419 +#if IEEE_754 || U_PLATFORM == U_PF_OS390
1.420 + return (UBool)(number < 0 && uprv_isInfinite(number));
1.421 +
1.422 +#else
1.423 + uint32_t highBits = *(uint32_t*)u_topNBytesOfDouble(&number,
1.424 + sizeof(uint32_t));
1.425 + return((highBits & SIGN) && uprv_isInfinite(number));
1.426 +
1.427 +#endif
1.428 +}
1.429 +
1.430 +U_CAPI double U_EXPORT2
1.431 +uprv_getNaN()
1.432 +{
1.433 +#if IEEE_754 || U_PLATFORM == U_PF_OS390
1.434 + return gNan.d64;
1.435 +#else
1.436 + /* If your platform doesn't support IEEE 754 but *does* have an NaN value,*/
1.437 + /* you'll need to replace this default implementation with what's correct*/
1.438 + /* for your platform.*/
1.439 + return 0.0;
1.440 +#endif
1.441 +}
1.442 +
1.443 +U_CAPI double U_EXPORT2
1.444 +uprv_getInfinity()
1.445 +{
1.446 +#if IEEE_754 || U_PLATFORM == U_PF_OS390
1.447 + return gInf.d64;
1.448 +#else
1.449 + /* If your platform doesn't support IEEE 754 but *does* have an infinity*/
1.450 + /* value, you'll need to replace this default implementation with what's*/
1.451 + /* correct for your platform.*/
1.452 + return 0.0;
1.453 +#endif
1.454 +}
1.455 +
1.456 +U_CAPI double U_EXPORT2
1.457 +uprv_floor(double x)
1.458 +{
1.459 + return floor(x);
1.460 +}
1.461 +
1.462 +U_CAPI double U_EXPORT2
1.463 +uprv_ceil(double x)
1.464 +{
1.465 + return ceil(x);
1.466 +}
1.467 +
1.468 +U_CAPI double U_EXPORT2
1.469 +uprv_round(double x)
1.470 +{
1.471 + return uprv_floor(x + 0.5);
1.472 +}
1.473 +
1.474 +U_CAPI double U_EXPORT2
1.475 +uprv_fabs(double x)
1.476 +{
1.477 + return fabs(x);
1.478 +}
1.479 +
1.480 +U_CAPI double U_EXPORT2
1.481 +uprv_modf(double x, double* y)
1.482 +{
1.483 + return modf(x, y);
1.484 +}
1.485 +
1.486 +U_CAPI double U_EXPORT2
1.487 +uprv_fmod(double x, double y)
1.488 +{
1.489 + return fmod(x, y);
1.490 +}
1.491 +
1.492 +U_CAPI double U_EXPORT2
1.493 +uprv_pow(double x, double y)
1.494 +{
1.495 + /* This is declared as "double pow(double x, double y)" */
1.496 + return pow(x, y);
1.497 +}
1.498 +
1.499 +U_CAPI double U_EXPORT2
1.500 +uprv_pow10(int32_t x)
1.501 +{
1.502 + return pow(10.0, (double)x);
1.503 +}
1.504 +
1.505 +U_CAPI double U_EXPORT2
1.506 +uprv_fmax(double x, double y)
1.507 +{
1.508 +#if IEEE_754
1.509 + /* first handle NaN*/
1.510 + if(uprv_isNaN(x) || uprv_isNaN(y))
1.511 + return uprv_getNaN();
1.512 +
1.513 + /* check for -0 and 0*/
1.514 + if(x == 0.0 && y == 0.0 && u_signBit(x))
1.515 + return y;
1.516 +
1.517 +#endif
1.518 +
1.519 + /* this should work for all flt point w/o NaN and Inf special cases */
1.520 + return (x > y ? x : y);
1.521 +}
1.522 +
1.523 +U_CAPI double U_EXPORT2
1.524 +uprv_fmin(double x, double y)
1.525 +{
1.526 +#if IEEE_754
1.527 + /* first handle NaN*/
1.528 + if(uprv_isNaN(x) || uprv_isNaN(y))
1.529 + return uprv_getNaN();
1.530 +
1.531 + /* check for -0 and 0*/
1.532 + if(x == 0.0 && y == 0.0 && u_signBit(y))
1.533 + return y;
1.534 +
1.535 +#endif
1.536 +
1.537 + /* this should work for all flt point w/o NaN and Inf special cases */
1.538 + return (x > y ? y : x);
1.539 +}
1.540 +
1.541 +/**
1.542 + * Truncates the given double.
1.543 + * trunc(3.3) = 3.0, trunc (-3.3) = -3.0
1.544 + * This is different than calling floor() or ceil():
1.545 + * floor(3.3) = 3, floor(-3.3) = -4
1.546 + * ceil(3.3) = 4, ceil(-3.3) = -3
1.547 + */
1.548 +U_CAPI double U_EXPORT2
1.549 +uprv_trunc(double d)
1.550 +{
1.551 +#if IEEE_754
1.552 + /* handle error cases*/
1.553 + if(uprv_isNaN(d))
1.554 + return uprv_getNaN();
1.555 + if(uprv_isInfinite(d))
1.556 + return uprv_getInfinity();
1.557 +
1.558 + if(u_signBit(d)) /* Signbit() picks up -0.0; d<0 does not. */
1.559 + return ceil(d);
1.560 + else
1.561 + return floor(d);
1.562 +
1.563 +#else
1.564 + return d >= 0 ? floor(d) : ceil(d);
1.565 +
1.566 +#endif
1.567 +}
1.568 +
1.569 +/**
1.570 + * Return the largest positive number that can be represented by an integer
1.571 + * type of arbitrary bit length.
1.572 + */
1.573 +U_CAPI double U_EXPORT2
1.574 +uprv_maxMantissa(void)
1.575 +{
1.576 + return pow(2.0, DBL_MANT_DIG + 1.0) - 1.0;
1.577 +}
1.578 +
1.579 +U_CAPI double U_EXPORT2
1.580 +uprv_log(double d)
1.581 +{
1.582 + return log(d);
1.583 +}
1.584 +
1.585 +U_CAPI void * U_EXPORT2
1.586 +uprv_maximumPtr(void * base)
1.587 +{
1.588 +#if U_PLATFORM == U_PF_OS400
1.589 + /*
1.590 + * With the provided function we should never be out of range of a given segment
1.591 + * (a traditional/typical segment that is). Our segments have 5 bytes for the
1.592 + * id and 3 bytes for the offset. The key is that the casting takes care of
1.593 + * only retrieving the offset portion minus x1000. Hence, the smallest offset
1.594 + * seen in a program is x001000 and when casted to an int would be 0.
1.595 + * That's why we can only add 0xffefff. Otherwise, we would exceed the segment.
1.596 + *
1.597 + * Currently, 16MB is the current addressing limitation on i5/OS if the activation is
1.598 + * non-TERASPACE. If it is TERASPACE it is 2GB - 4k(header information).
1.599 + * This function determines the activation based on the pointer that is passed in and
1.600 + * calculates the appropriate maximum available size for
1.601 + * each pointer type (TERASPACE and non-TERASPACE)
1.602 + *
1.603 + * Unlike other operating systems, the pointer model isn't determined at
1.604 + * compile time on i5/OS.
1.605 + */
1.606 + if ((base != NULL) && (_TESTPTR(base, _C_TERASPACE_CHECK))) {
1.607 + /* if it is a TERASPACE pointer the max is 2GB - 4k */
1.608 + return ((void *)(((char *)base)-((uint32_t)(base))+((uint32_t)0x7fffefff)));
1.609 + }
1.610 + /* otherwise 16MB since NULL ptr is not checkable or the ptr is not TERASPACE */
1.611 + return ((void *)(((char *)base)-((uint32_t)(base))+((uint32_t)0xffefff)));
1.612 +
1.613 +#else
1.614 + return U_MAX_PTR(base);
1.615 +#endif
1.616 +}
1.617 +
1.618 +/*---------------------------------------------------------------------------
1.619 + Platform-specific Implementations
1.620 + Try these, and if they don't work on your platform, then special case your
1.621 + platform with new implementations.
1.622 + ---------------------------------------------------------------------------*/
1.623 +
1.624 +/* Generic time zone layer -------------------------------------------------- */
1.625 +
1.626 +/* Time zone utilities */
1.627 +U_CAPI void U_EXPORT2
1.628 +uprv_tzset()
1.629 +{
1.630 +#if defined(U_TZSET)
1.631 + U_TZSET();
1.632 +#else
1.633 + /* no initialization*/
1.634 +#endif
1.635 +}
1.636 +
1.637 +U_CAPI int32_t U_EXPORT2
1.638 +uprv_timezone()
1.639 +{
1.640 +#ifdef U_TIMEZONE
1.641 + return U_TIMEZONE;
1.642 +#else
1.643 + time_t t, t1, t2;
1.644 + struct tm tmrec;
1.645 + int32_t tdiff = 0;
1.646 +
1.647 + time(&t);
1.648 + uprv_memcpy( &tmrec, localtime(&t), sizeof(tmrec) );
1.649 +#if U_PLATFORM != U_PF_IPHONE
1.650 + UBool dst_checked = (tmrec.tm_isdst != 0); /* daylight savings time is checked*/
1.651 +#endif
1.652 + t1 = mktime(&tmrec); /* local time in seconds*/
1.653 + uprv_memcpy( &tmrec, gmtime(&t), sizeof(tmrec) );
1.654 + t2 = mktime(&tmrec); /* GMT (or UTC) in seconds*/
1.655 + tdiff = t2 - t1;
1.656 +
1.657 +#if U_PLATFORM != U_PF_IPHONE
1.658 + /* imitate NT behaviour, which returns same timezone offset to GMT for
1.659 + winter and summer.
1.660 + This does not work on all platforms. For instance, on glibc on Linux
1.661 + and on Mac OS 10.5, tdiff calculated above remains the same
1.662 + regardless of whether DST is in effect or not. iOS is another
1.663 + platform where this does not work. Linux + glibc and Mac OS 10.5
1.664 + have U_TIMEZONE defined so that this code is not reached.
1.665 + */
1.666 + if (dst_checked)
1.667 + tdiff += 3600;
1.668 +#endif
1.669 + return tdiff;
1.670 +#endif
1.671 +}
1.672 +
1.673 +/* Note that U_TZNAME does *not* have to be tzname, but if it is,
1.674 + some platforms need to have it declared here. */
1.675 +
1.676 +#if defined(U_TZNAME) && (U_PLATFORM == U_PF_IRIX || U_PLATFORM_IS_DARWIN_BASED || (U_PLATFORM == U_PF_CYGWIN && !U_PLATFORM_USES_ONLY_WIN32_API))
1.677 +/* RS6000 and others reject char **tzname. */
1.678 +extern U_IMPORT char *U_TZNAME[];
1.679 +#endif
1.680 +
1.681 +#if !UCONFIG_NO_FILE_IO && ((U_PLATFORM_IS_DARWIN_BASED && (U_PLATFORM != U_PF_IPHONE || defined(U_TIMEZONE))) || U_PLATFORM_IS_LINUX_BASED || U_PLATFORM == U_PF_BSD || U_PLATFORM == U_PF_SOLARIS)
1.682 +/* These platforms are likely to use Olson timezone IDs. */
1.683 +#define CHECK_LOCALTIME_LINK 1
1.684 +#if U_PLATFORM_IS_DARWIN_BASED
1.685 +#include <tzfile.h>
1.686 +#define TZZONEINFO (TZDIR "/")
1.687 +#elif U_PLATFORM == U_PF_SOLARIS
1.688 +#define TZDEFAULT "/etc/localtime"
1.689 +#define TZZONEINFO "/usr/share/lib/zoneinfo/"
1.690 +#define TZZONEINFO2 "../usr/share/lib/zoneinfo/"
1.691 +#define TZ_ENV_CHECK "localtime"
1.692 +#else
1.693 +#define TZDEFAULT "/etc/localtime"
1.694 +#define TZZONEINFO "/usr/share/zoneinfo/"
1.695 +#endif
1.696 +#if U_HAVE_DIRENT_H
1.697 +#define TZFILE_SKIP "posixrules" /* tz file to skip when searching. */
1.698 +/* Some Linux distributions have 'localtime' in /usr/share/zoneinfo
1.699 + symlinked to /etc/localtime, which makes searchForTZFile return
1.700 + 'localtime' when it's the first match. */
1.701 +#define TZFILE_SKIP2 "localtime"
1.702 +#define SEARCH_TZFILE
1.703 +#include <dirent.h> /* Needed to search through system timezone files */
1.704 +#endif
1.705 +static char gTimeZoneBuffer[PATH_MAX];
1.706 +static char *gTimeZoneBufferPtr = NULL;
1.707 +#endif
1.708 +
1.709 +#if !U_PLATFORM_USES_ONLY_WIN32_API
1.710 +#define isNonDigit(ch) (ch < '0' || '9' < ch)
1.711 +static UBool isValidOlsonID(const char *id) {
1.712 + int32_t idx = 0;
1.713 +
1.714 + /* Determine if this is something like Iceland (Olson ID)
1.715 + or AST4ADT (non-Olson ID) */
1.716 + while (id[idx] && isNonDigit(id[idx]) && id[idx] != ',') {
1.717 + idx++;
1.718 + }
1.719 +
1.720 + /* If we went through the whole string, then it might be okay.
1.721 + The timezone is sometimes set to "CST-7CDT", "CST6CDT5,J129,J131/19:30",
1.722 + "GRNLNDST3GRNLNDDT" or similar, so we cannot use it.
1.723 + The rest of the time it could be an Olson ID. George */
1.724 + return (UBool)(id[idx] == 0
1.725 + || uprv_strcmp(id, "PST8PDT") == 0
1.726 + || uprv_strcmp(id, "MST7MDT") == 0
1.727 + || uprv_strcmp(id, "CST6CDT") == 0
1.728 + || uprv_strcmp(id, "EST5EDT") == 0);
1.729 +}
1.730 +
1.731 +/* On some Unix-like OS, 'posix' subdirectory in
1.732 + /usr/share/zoneinfo replicates the top-level contents. 'right'
1.733 + subdirectory has the same set of files, but individual files
1.734 + are different from those in the top-level directory or 'posix'
1.735 + because 'right' has files for TAI (Int'l Atomic Time) while 'posix'
1.736 + has files for UTC.
1.737 + When the first match for /etc/localtime is in either of them
1.738 + (usually in posix because 'right' has different file contents),
1.739 + or TZ environment variable points to one of them, createTimeZone
1.740 + fails because, say, 'posix/America/New_York' is not an Olson
1.741 + timezone id ('America/New_York' is). So, we have to skip
1.742 + 'posix/' and 'right/' at the beginning. */
1.743 +static void skipZoneIDPrefix(const char** id) {
1.744 + if (uprv_strncmp(*id, "posix/", 6) == 0
1.745 + || uprv_strncmp(*id, "right/", 6) == 0)
1.746 + {
1.747 + *id += 6;
1.748 + }
1.749 +}
1.750 +#endif
1.751 +
1.752 +#if defined(U_TZNAME) && !U_PLATFORM_USES_ONLY_WIN32_API
1.753 +
1.754 +#define CONVERT_HOURS_TO_SECONDS(offset) (int32_t)(offset*3600)
1.755 +typedef struct OffsetZoneMapping {
1.756 + int32_t offsetSeconds;
1.757 + int32_t daylightType; /* 0=U_DAYLIGHT_NONE, 1=daylight in June-U_DAYLIGHT_JUNE, 2=daylight in December=U_DAYLIGHT_DECEMBER*/
1.758 + const char *stdID;
1.759 + const char *dstID;
1.760 + const char *olsonID;
1.761 +} OffsetZoneMapping;
1.762 +
1.763 +enum { U_DAYLIGHT_NONE=0,U_DAYLIGHT_JUNE=1,U_DAYLIGHT_DECEMBER=2 };
1.764 +
1.765 +/*
1.766 +This list tries to disambiguate a set of abbreviated timezone IDs and offsets
1.767 +and maps it to an Olson ID.
1.768 +Before adding anything to this list, take a look at
1.769 +icu/source/tools/tzcode/tz.alias
1.770 +Sometimes no daylight savings (0) is important to define due to aliases.
1.771 +This list can be tested with icu/source/test/compat/tzone.pl
1.772 +More values could be added to daylightType to increase precision.
1.773 +*/
1.774 +static const struct OffsetZoneMapping OFFSET_ZONE_MAPPINGS[] = {
1.775 + {-45900, 2, "CHAST", "CHADT", "Pacific/Chatham"},
1.776 + {-43200, 1, "PETT", "PETST", "Asia/Kamchatka"},
1.777 + {-43200, 2, "NZST", "NZDT", "Pacific/Auckland"},
1.778 + {-43200, 1, "ANAT", "ANAST", "Asia/Anadyr"},
1.779 + {-39600, 1, "MAGT", "MAGST", "Asia/Magadan"},
1.780 + {-37800, 2, "LHST", "LHST", "Australia/Lord_Howe"},
1.781 + {-36000, 2, "EST", "EST", "Australia/Sydney"},
1.782 + {-36000, 1, "SAKT", "SAKST", "Asia/Sakhalin"},
1.783 + {-36000, 1, "VLAT", "VLAST", "Asia/Vladivostok"},
1.784 + {-34200, 2, "CST", "CST", "Australia/South"},
1.785 + {-32400, 1, "YAKT", "YAKST", "Asia/Yakutsk"},
1.786 + {-32400, 1, "CHOT", "CHOST", "Asia/Choibalsan"},
1.787 + {-31500, 2, "CWST", "CWST", "Australia/Eucla"},
1.788 + {-28800, 1, "IRKT", "IRKST", "Asia/Irkutsk"},
1.789 + {-28800, 1, "ULAT", "ULAST", "Asia/Ulaanbaatar"},
1.790 + {-28800, 2, "WST", "WST", "Australia/West"},
1.791 + {-25200, 1, "HOVT", "HOVST", "Asia/Hovd"},
1.792 + {-25200, 1, "KRAT", "KRAST", "Asia/Krasnoyarsk"},
1.793 + {-21600, 1, "NOVT", "NOVST", "Asia/Novosibirsk"},
1.794 + {-21600, 1, "OMST", "OMSST", "Asia/Omsk"},
1.795 + {-18000, 1, "YEKT", "YEKST", "Asia/Yekaterinburg"},
1.796 + {-14400, 1, "SAMT", "SAMST", "Europe/Samara"},
1.797 + {-14400, 1, "AMT", "AMST", "Asia/Yerevan"},
1.798 + {-14400, 1, "AZT", "AZST", "Asia/Baku"},
1.799 + {-10800, 1, "AST", "ADT", "Asia/Baghdad"},
1.800 + {-10800, 1, "MSK", "MSD", "Europe/Moscow"},
1.801 + {-10800, 1, "VOLT", "VOLST", "Europe/Volgograd"},
1.802 + {-7200, 0, "EET", "CEST", "Africa/Tripoli"},
1.803 + {-7200, 1, "EET", "EEST", "Europe/Athens"}, /* Conflicts with Africa/Cairo */
1.804 + {-7200, 1, "IST", "IDT", "Asia/Jerusalem"},
1.805 + {-3600, 0, "CET", "WEST", "Africa/Algiers"},
1.806 + {-3600, 2, "WAT", "WAST", "Africa/Windhoek"},
1.807 + {0, 1, "GMT", "IST", "Europe/Dublin"},
1.808 + {0, 1, "GMT", "BST", "Europe/London"},
1.809 + {0, 0, "WET", "WEST", "Africa/Casablanca"},
1.810 + {0, 0, "WET", "WET", "Africa/El_Aaiun"},
1.811 + {3600, 1, "AZOT", "AZOST", "Atlantic/Azores"},
1.812 + {3600, 1, "EGT", "EGST", "America/Scoresbysund"},
1.813 + {10800, 1, "PMST", "PMDT", "America/Miquelon"},
1.814 + {10800, 2, "UYT", "UYST", "America/Montevideo"},
1.815 + {10800, 1, "WGT", "WGST", "America/Godthab"},
1.816 + {10800, 2, "BRT", "BRST", "Brazil/East"},
1.817 + {12600, 1, "NST", "NDT", "America/St_Johns"},
1.818 + {14400, 1, "AST", "ADT", "Canada/Atlantic"},
1.819 + {14400, 2, "AMT", "AMST", "America/Cuiaba"},
1.820 + {14400, 2, "CLT", "CLST", "Chile/Continental"},
1.821 + {14400, 2, "FKT", "FKST", "Atlantic/Stanley"},
1.822 + {14400, 2, "PYT", "PYST", "America/Asuncion"},
1.823 + {18000, 1, "CST", "CDT", "America/Havana"},
1.824 + {18000, 1, "EST", "EDT", "US/Eastern"}, /* Conflicts with America/Grand_Turk */
1.825 + {21600, 2, "EAST", "EASST", "Chile/EasterIsland"},
1.826 + {21600, 0, "CST", "MDT", "Canada/Saskatchewan"},
1.827 + {21600, 0, "CST", "CDT", "America/Guatemala"},
1.828 + {21600, 1, "CST", "CDT", "US/Central"}, /* Conflicts with Mexico/General */
1.829 + {25200, 1, "MST", "MDT", "US/Mountain"}, /* Conflicts with Mexico/BajaSur */
1.830 + {28800, 0, "PST", "PST", "Pacific/Pitcairn"},
1.831 + {28800, 1, "PST", "PDT", "US/Pacific"}, /* Conflicts with Mexico/BajaNorte */
1.832 + {32400, 1, "AKST", "AKDT", "US/Alaska"},
1.833 + {36000, 1, "HAST", "HADT", "US/Aleutian"}
1.834 +};
1.835 +
1.836 +/*#define DEBUG_TZNAME*/
1.837 +
1.838 +static const char* remapShortTimeZone(const char *stdID, const char *dstID, int32_t daylightType, int32_t offset)
1.839 +{
1.840 + int32_t idx;
1.841 +#ifdef DEBUG_TZNAME
1.842 + fprintf(stderr, "TZ=%s std=%s dst=%s daylight=%d offset=%d\n", getenv("TZ"), stdID, dstID, daylightType, offset);
1.843 +#endif
1.844 + for (idx = 0; idx < LENGTHOF(OFFSET_ZONE_MAPPINGS); idx++)
1.845 + {
1.846 + if (offset == OFFSET_ZONE_MAPPINGS[idx].offsetSeconds
1.847 + && daylightType == OFFSET_ZONE_MAPPINGS[idx].daylightType
1.848 + && strcmp(OFFSET_ZONE_MAPPINGS[idx].stdID, stdID) == 0
1.849 + && strcmp(OFFSET_ZONE_MAPPINGS[idx].dstID, dstID) == 0)
1.850 + {
1.851 + return OFFSET_ZONE_MAPPINGS[idx].olsonID;
1.852 + }
1.853 + }
1.854 + return NULL;
1.855 +}
1.856 +#endif
1.857 +
1.858 +#ifdef SEARCH_TZFILE
1.859 +#define MAX_PATH_SIZE PATH_MAX /* Set the limit for the size of the path. */
1.860 +#define MAX_READ_SIZE 512
1.861 +
1.862 +typedef struct DefaultTZInfo {
1.863 + char* defaultTZBuffer;
1.864 + int64_t defaultTZFileSize;
1.865 + FILE* defaultTZFilePtr;
1.866 + UBool defaultTZstatus;
1.867 + int32_t defaultTZPosition;
1.868 +} DefaultTZInfo;
1.869 +
1.870 +/*
1.871 + * This method compares the two files given to see if they are a match.
1.872 + * It is currently use to compare two TZ files.
1.873 + */
1.874 +static UBool compareBinaryFiles(const char* defaultTZFileName, const char* TZFileName, DefaultTZInfo* tzInfo) {
1.875 + FILE* file;
1.876 + int64_t sizeFile;
1.877 + int64_t sizeFileLeft;
1.878 + int32_t sizeFileRead;
1.879 + int32_t sizeFileToRead;
1.880 + char bufferFile[MAX_READ_SIZE];
1.881 + UBool result = TRUE;
1.882 +
1.883 + if (tzInfo->defaultTZFilePtr == NULL) {
1.884 + tzInfo->defaultTZFilePtr = fopen(defaultTZFileName, "r");
1.885 + }
1.886 + file = fopen(TZFileName, "r");
1.887 +
1.888 + tzInfo->defaultTZPosition = 0; /* reset position to begin search */
1.889 +
1.890 + if (file != NULL && tzInfo->defaultTZFilePtr != NULL) {
1.891 + /* First check that the file size are equal. */
1.892 + if (tzInfo->defaultTZFileSize == 0) {
1.893 + fseek(tzInfo->defaultTZFilePtr, 0, SEEK_END);
1.894 + tzInfo->defaultTZFileSize = ftell(tzInfo->defaultTZFilePtr);
1.895 + }
1.896 + fseek(file, 0, SEEK_END);
1.897 + sizeFile = ftell(file);
1.898 + sizeFileLeft = sizeFile;
1.899 +
1.900 + if (sizeFile != tzInfo->defaultTZFileSize) {
1.901 + result = FALSE;
1.902 + } else {
1.903 + /* Store the data from the files in seperate buffers and
1.904 + * compare each byte to determine equality.
1.905 + */
1.906 + if (tzInfo->defaultTZBuffer == NULL) {
1.907 + rewind(tzInfo->defaultTZFilePtr);
1.908 + tzInfo->defaultTZBuffer = (char*)uprv_malloc(sizeof(char) * tzInfo->defaultTZFileSize);
1.909 + sizeFileRead = fread(tzInfo->defaultTZBuffer, 1, tzInfo->defaultTZFileSize, tzInfo->defaultTZFilePtr);
1.910 + }
1.911 + rewind(file);
1.912 + while(sizeFileLeft > 0) {
1.913 + uprv_memset(bufferFile, 0, MAX_READ_SIZE);
1.914 + sizeFileToRead = sizeFileLeft < MAX_READ_SIZE ? sizeFileLeft : MAX_READ_SIZE;
1.915 +
1.916 + sizeFileRead = fread(bufferFile, 1, sizeFileToRead, file);
1.917 + if (memcmp(tzInfo->defaultTZBuffer + tzInfo->defaultTZPosition, bufferFile, sizeFileRead) != 0) {
1.918 + result = FALSE;
1.919 + break;
1.920 + }
1.921 + sizeFileLeft -= sizeFileRead;
1.922 + tzInfo->defaultTZPosition += sizeFileRead;
1.923 + }
1.924 + }
1.925 + } else {
1.926 + result = FALSE;
1.927 + }
1.928 +
1.929 + if (file != NULL) {
1.930 + fclose(file);
1.931 + }
1.932 +
1.933 + return result;
1.934 +}
1.935 +/*
1.936 + * This method recursively traverses the directory given for a matching TZ file and returns the first match.
1.937 + */
1.938 +/* dirent also lists two entries: "." and ".." that we can safely ignore. */
1.939 +#define SKIP1 "."
1.940 +#define SKIP2 ".."
1.941 +static char SEARCH_TZFILE_RESULT[MAX_PATH_SIZE] = "";
1.942 +static char* searchForTZFile(const char* path, DefaultTZInfo* tzInfo) {
1.943 + char curpath[MAX_PATH_SIZE];
1.944 + DIR* dirp = opendir(path);
1.945 + DIR* subDirp = NULL;
1.946 + struct dirent* dirEntry = NULL;
1.947 +
1.948 + char* result = NULL;
1.949 + if (dirp == NULL) {
1.950 + return result;
1.951 + }
1.952 +
1.953 + /* Save the current path */
1.954 + uprv_memset(curpath, 0, MAX_PATH_SIZE);
1.955 + uprv_strcpy(curpath, path);
1.956 +
1.957 + /* Check each entry in the directory. */
1.958 + while((dirEntry = readdir(dirp)) != NULL) {
1.959 + const char* dirName = dirEntry->d_name;
1.960 + if (uprv_strcmp(dirName, SKIP1) != 0 && uprv_strcmp(dirName, SKIP2) != 0) {
1.961 + /* Create a newpath with the new entry to test each entry in the directory. */
1.962 + char newpath[MAX_PATH_SIZE];
1.963 + uprv_strcpy(newpath, curpath);
1.964 + uprv_strcat(newpath, dirName);
1.965 +
1.966 + if ((subDirp = opendir(newpath)) != NULL) {
1.967 + /* If this new path is a directory, make a recursive call with the newpath. */
1.968 + closedir(subDirp);
1.969 + uprv_strcat(newpath, "/");
1.970 + result = searchForTZFile(newpath, tzInfo);
1.971 + /*
1.972 + Have to get out here. Otherwise, we'd keep looking
1.973 + and return the first match in the top-level directory
1.974 + if there's a match in the top-level. If not, this function
1.975 + would return NULL and set gTimeZoneBufferPtr to NULL in initDefault().
1.976 + It worked without this in most cases because we have a fallback of calling
1.977 + localtime_r to figure out the default timezone.
1.978 + */
1.979 + if (result != NULL)
1.980 + break;
1.981 + } else if (uprv_strcmp(TZFILE_SKIP, dirName) != 0 && uprv_strcmp(TZFILE_SKIP2, dirName) != 0) {
1.982 + if(compareBinaryFiles(TZDEFAULT, newpath, tzInfo)) {
1.983 + const char* zoneid = newpath + (sizeof(TZZONEINFO)) - 1;
1.984 + skipZoneIDPrefix(&zoneid);
1.985 + uprv_strcpy(SEARCH_TZFILE_RESULT, zoneid);
1.986 + result = SEARCH_TZFILE_RESULT;
1.987 + /* Get out after the first one found. */
1.988 + break;
1.989 + }
1.990 + }
1.991 + }
1.992 + }
1.993 + closedir(dirp);
1.994 + return result;
1.995 +}
1.996 +#endif
1.997 +U_CAPI const char* U_EXPORT2
1.998 +uprv_tzname(int n)
1.999 +{
1.1000 + const char *tzid = NULL;
1.1001 +#if U_PLATFORM_USES_ONLY_WIN32_API
1.1002 + tzid = uprv_detectWindowsTimeZone();
1.1003 +
1.1004 + if (tzid != NULL) {
1.1005 + return tzid;
1.1006 + }
1.1007 +#else
1.1008 +
1.1009 +/*#if U_PLATFORM_IS_DARWIN_BASED
1.1010 + int ret;
1.1011 +
1.1012 + tzid = getenv("TZFILE");
1.1013 + if (tzid != NULL) {
1.1014 + return tzid;
1.1015 + }
1.1016 +#endif*/
1.1017 +
1.1018 +/* This code can be temporarily disabled to test tzname resolution later on. */
1.1019 +#ifndef DEBUG_TZNAME
1.1020 + tzid = getenv("TZ");
1.1021 + if (tzid != NULL && isValidOlsonID(tzid)
1.1022 +#if U_PLATFORM == U_PF_SOLARIS
1.1023 + /* When TZ equals localtime on Solaris, check the /etc/localtime file. */
1.1024 + && uprv_strcmp(tzid, TZ_ENV_CHECK) != 0
1.1025 +#endif
1.1026 + ) {
1.1027 + /* This might be a good Olson ID. */
1.1028 + skipZoneIDPrefix(&tzid);
1.1029 + return tzid;
1.1030 + }
1.1031 + /* else U_TZNAME will give a better result. */
1.1032 +#endif
1.1033 +
1.1034 +#if defined(CHECK_LOCALTIME_LINK) && !defined(DEBUG_SKIP_LOCALTIME_LINK)
1.1035 + /* Caller must handle threading issues */
1.1036 + if (gTimeZoneBufferPtr == NULL) {
1.1037 + /*
1.1038 + This is a trick to look at the name of the link to get the Olson ID
1.1039 + because the tzfile contents is underspecified.
1.1040 + This isn't guaranteed to work because it may not be a symlink.
1.1041 + */
1.1042 + int32_t ret = (int32_t)readlink(TZDEFAULT, gTimeZoneBuffer, sizeof(gTimeZoneBuffer));
1.1043 + if (0 < ret) {
1.1044 + int32_t tzZoneInfoLen = uprv_strlen(TZZONEINFO);
1.1045 + gTimeZoneBuffer[ret] = 0;
1.1046 + if (uprv_strncmp(gTimeZoneBuffer, TZZONEINFO, tzZoneInfoLen) == 0
1.1047 + && isValidOlsonID(gTimeZoneBuffer + tzZoneInfoLen))
1.1048 + {
1.1049 + return (gTimeZoneBufferPtr = gTimeZoneBuffer + tzZoneInfoLen);
1.1050 + }
1.1051 +#if U_PLATFORM == U_PF_SOLARIS
1.1052 + else
1.1053 + {
1.1054 + tzZoneInfoLen = uprv_strlen(TZZONEINFO2);
1.1055 + if (uprv_strncmp(gTimeZoneBuffer, TZZONEINFO2, tzZoneInfoLen) == 0
1.1056 + && isValidOlsonID(gTimeZoneBuffer + tzZoneInfoLen))
1.1057 + {
1.1058 + return (gTimeZoneBufferPtr = gTimeZoneBuffer + tzZoneInfoLen);
1.1059 + }
1.1060 + }
1.1061 +#endif
1.1062 + } else {
1.1063 +#if defined(SEARCH_TZFILE)
1.1064 + DefaultTZInfo* tzInfo = (DefaultTZInfo*)uprv_malloc(sizeof(DefaultTZInfo));
1.1065 + if (tzInfo != NULL) {
1.1066 + tzInfo->defaultTZBuffer = NULL;
1.1067 + tzInfo->defaultTZFileSize = 0;
1.1068 + tzInfo->defaultTZFilePtr = NULL;
1.1069 + tzInfo->defaultTZstatus = FALSE;
1.1070 + tzInfo->defaultTZPosition = 0;
1.1071 +
1.1072 + gTimeZoneBufferPtr = searchForTZFile(TZZONEINFO, tzInfo);
1.1073 +
1.1074 + /* Free previously allocated memory */
1.1075 + if (tzInfo->defaultTZBuffer != NULL) {
1.1076 + uprv_free(tzInfo->defaultTZBuffer);
1.1077 + }
1.1078 + if (tzInfo->defaultTZFilePtr != NULL) {
1.1079 + fclose(tzInfo->defaultTZFilePtr);
1.1080 + }
1.1081 + uprv_free(tzInfo);
1.1082 + }
1.1083 +
1.1084 + if (gTimeZoneBufferPtr != NULL && isValidOlsonID(gTimeZoneBufferPtr)) {
1.1085 + return gTimeZoneBufferPtr;
1.1086 + }
1.1087 +#endif
1.1088 + }
1.1089 + }
1.1090 + else {
1.1091 + return gTimeZoneBufferPtr;
1.1092 + }
1.1093 +#endif
1.1094 +#endif
1.1095 +
1.1096 +#ifdef U_TZNAME
1.1097 +#if U_PLATFORM_USES_ONLY_WIN32_API
1.1098 + /* The return value is free'd in timezone.cpp on Windows because
1.1099 + * the other code path returns a pointer to a heap location. */
1.1100 + return uprv_strdup(U_TZNAME[n]);
1.1101 +#else
1.1102 + /*
1.1103 + U_TZNAME is usually a non-unique abbreviation, which isn't normally usable.
1.1104 + So we remap the abbreviation to an olson ID.
1.1105 +
1.1106 + Since Windows exposes a little more timezone information,
1.1107 + we normally don't use this code on Windows because
1.1108 + uprv_detectWindowsTimeZone should have already given the correct answer.
1.1109 + */
1.1110 + {
1.1111 + struct tm juneSol, decemberSol;
1.1112 + int daylightType;
1.1113 + static const time_t juneSolstice=1182478260; /*2007-06-21 18:11 UT*/
1.1114 + static const time_t decemberSolstice=1198332540; /*2007-12-22 06:09 UT*/
1.1115 +
1.1116 + /* This probing will tell us when daylight savings occurs. */
1.1117 + localtime_r(&juneSolstice, &juneSol);
1.1118 + localtime_r(&decemberSolstice, &decemberSol);
1.1119 + if(decemberSol.tm_isdst > 0) {
1.1120 + daylightType = U_DAYLIGHT_DECEMBER;
1.1121 + } else if(juneSol.tm_isdst > 0) {
1.1122 + daylightType = U_DAYLIGHT_JUNE;
1.1123 + } else {
1.1124 + daylightType = U_DAYLIGHT_NONE;
1.1125 + }
1.1126 + tzid = remapShortTimeZone(U_TZNAME[0], U_TZNAME[1], daylightType, uprv_timezone());
1.1127 + if (tzid != NULL) {
1.1128 + return tzid;
1.1129 + }
1.1130 + }
1.1131 + return U_TZNAME[n];
1.1132 +#endif
1.1133 +#else
1.1134 + return "";
1.1135 +#endif
1.1136 +}
1.1137 +
1.1138 +/* Get and set the ICU data directory --------------------------------------- */
1.1139 +
1.1140 +static char *gDataDirectory = NULL;
1.1141 +#if U_POSIX_LOCALE || U_PLATFORM_USES_ONLY_WIN32_API
1.1142 + static char *gCorrectedPOSIXLocale = NULL; /* Heap allocated */
1.1143 +#endif
1.1144 +
1.1145 +static UBool U_CALLCONV putil_cleanup(void)
1.1146 +{
1.1147 + if (gDataDirectory && *gDataDirectory) {
1.1148 + uprv_free(gDataDirectory);
1.1149 + }
1.1150 + gDataDirectory = NULL;
1.1151 +#if U_POSIX_LOCALE || U_PLATFORM_USES_ONLY_WIN32_API
1.1152 + if (gCorrectedPOSIXLocale) {
1.1153 + uprv_free(gCorrectedPOSIXLocale);
1.1154 + gCorrectedPOSIXLocale = NULL;
1.1155 + }
1.1156 +#endif
1.1157 + return TRUE;
1.1158 +}
1.1159 +
1.1160 +/*
1.1161 + * Set the data directory.
1.1162 + * Make a copy of the passed string, and set the global data dir to point to it.
1.1163 + */
1.1164 +U_CAPI void U_EXPORT2
1.1165 +u_setDataDirectory(const char *directory) {
1.1166 + char *newDataDir;
1.1167 + int32_t length;
1.1168 +
1.1169 + if(directory==NULL || *directory==0) {
1.1170 + /* A small optimization to prevent the malloc and copy when the
1.1171 + shared library is used, and this is a way to make sure that NULL
1.1172 + is never returned.
1.1173 + */
1.1174 + newDataDir = (char *)"";
1.1175 + }
1.1176 + else {
1.1177 + length=(int32_t)uprv_strlen(directory);
1.1178 + newDataDir = (char *)uprv_malloc(length + 2);
1.1179 + /* Exit out if newDataDir could not be created. */
1.1180 + if (newDataDir == NULL) {
1.1181 + return;
1.1182 + }
1.1183 + uprv_strcpy(newDataDir, directory);
1.1184 +
1.1185 +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR)
1.1186 + {
1.1187 + char *p;
1.1188 + while(p = uprv_strchr(newDataDir, U_FILE_ALT_SEP_CHAR)) {
1.1189 + *p = U_FILE_SEP_CHAR;
1.1190 + }
1.1191 + }
1.1192 +#endif
1.1193 + }
1.1194 +
1.1195 + if (gDataDirectory && *gDataDirectory) {
1.1196 + uprv_free(gDataDirectory);
1.1197 + }
1.1198 + gDataDirectory = newDataDir;
1.1199 + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup);
1.1200 +}
1.1201 +
1.1202 +U_CAPI UBool U_EXPORT2
1.1203 +uprv_pathIsAbsolute(const char *path)
1.1204 +{
1.1205 + if(!path || !*path) {
1.1206 + return FALSE;
1.1207 + }
1.1208 +
1.1209 + if(*path == U_FILE_SEP_CHAR) {
1.1210 + return TRUE;
1.1211 + }
1.1212 +
1.1213 +#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR)
1.1214 + if(*path == U_FILE_ALT_SEP_CHAR) {
1.1215 + return TRUE;
1.1216 + }
1.1217 +#endif
1.1218 +
1.1219 +#if U_PLATFORM_USES_ONLY_WIN32_API
1.1220 + if( (((path[0] >= 'A') && (path[0] <= 'Z')) ||
1.1221 + ((path[0] >= 'a') && (path[0] <= 'z'))) &&
1.1222 + path[1] == ':' ) {
1.1223 + return TRUE;
1.1224 + }
1.1225 +#endif
1.1226 +
1.1227 + return FALSE;
1.1228 +}
1.1229 +
1.1230 +/* Temporary backup setting of ICU_DATA_DIR_PREFIX_ENV_VAR
1.1231 + until some client wrapper makefiles are updated */
1.1232 +#if U_PLATFORM_IS_DARWIN_BASED && TARGET_IPHONE_SIMULATOR
1.1233 +# if !defined(ICU_DATA_DIR_PREFIX_ENV_VAR)
1.1234 +# define ICU_DATA_DIR_PREFIX_ENV_VAR "IPHONE_SIMULATOR_ROOT"
1.1235 +# endif
1.1236 +#endif
1.1237 +
1.1238 +U_CAPI const char * U_EXPORT2
1.1239 +u_getDataDirectory(void) {
1.1240 + const char *path = NULL;
1.1241 +#if defined(ICU_DATA_DIR_PREFIX_ENV_VAR)
1.1242 + char datadir_path_buffer[PATH_MAX];
1.1243 +#endif
1.1244 +
1.1245 + /* if we have the directory, then return it immediately */
1.1246 + if(gDataDirectory) {
1.1247 + return gDataDirectory;
1.1248 + }
1.1249 +
1.1250 + /*
1.1251 + When ICU_NO_USER_DATA_OVERRIDE is defined, users aren't allowed to
1.1252 + override ICU's data with the ICU_DATA environment variable. This prevents
1.1253 + problems where multiple custom copies of ICU's specific version of data
1.1254 + are installed on a system. Either the application must define the data
1.1255 + directory with u_setDataDirectory, define ICU_DATA_DIR when compiling
1.1256 + ICU, set the data with udata_setCommonData or trust that all of the
1.1257 + required data is contained in ICU's data library that contains
1.1258 + the entry point defined by U_ICUDATA_ENTRY_POINT.
1.1259 +
1.1260 + There may also be some platforms where environment variables
1.1261 + are not allowed.
1.1262 + */
1.1263 +# if !defined(ICU_NO_USER_DATA_OVERRIDE) && !UCONFIG_NO_FILE_IO
1.1264 + /* First try to get the environment variable */
1.1265 + path=getenv("ICU_DATA");
1.1266 +# endif
1.1267 +
1.1268 + /* ICU_DATA_DIR may be set as a compile option.
1.1269 + * U_ICU_DATA_DEFAULT_DIR is provided and is set by ICU at compile time
1.1270 + * and is used only when data is built in archive mode eliminating the need
1.1271 + * for ICU_DATA_DIR to be set. U_ICU_DATA_DEFAULT_DIR is set to the installation
1.1272 + * directory of the data dat file. Users should use ICU_DATA_DIR if they want to
1.1273 + * set their own path.
1.1274 + */
1.1275 +#if defined(ICU_DATA_DIR) || defined(U_ICU_DATA_DEFAULT_DIR)
1.1276 + if(path==NULL || *path==0) {
1.1277 +# if defined(ICU_DATA_DIR_PREFIX_ENV_VAR)
1.1278 + const char *prefix = getenv(ICU_DATA_DIR_PREFIX_ENV_VAR);
1.1279 +# endif
1.1280 +# ifdef ICU_DATA_DIR
1.1281 + path=ICU_DATA_DIR;
1.1282 +# else
1.1283 + path=U_ICU_DATA_DEFAULT_DIR;
1.1284 +# endif
1.1285 +# if defined(ICU_DATA_DIR_PREFIX_ENV_VAR)
1.1286 + if (prefix != NULL) {
1.1287 + snprintf(datadir_path_buffer, PATH_MAX, "%s%s", prefix, path);
1.1288 + path=datadir_path_buffer;
1.1289 + }
1.1290 +# endif
1.1291 + }
1.1292 +#endif
1.1293 +
1.1294 + if(path==NULL) {
1.1295 + /* It looks really bad, set it to something. */
1.1296 + path = "";
1.1297 + }
1.1298 +
1.1299 + u_setDataDirectory(path);
1.1300 + return gDataDirectory;
1.1301 +}
1.1302 +
1.1303 +
1.1304 +
1.1305 +
1.1306 +
1.1307 +/* Macintosh-specific locale information ------------------------------------ */
1.1308 +#if U_PLATFORM == U_PF_CLASSIC_MACOS
1.1309 +
1.1310 +typedef struct {
1.1311 + int32_t script;
1.1312 + int32_t region;
1.1313 + int32_t lang;
1.1314 + int32_t date_region;
1.1315 + const char* posixID;
1.1316 +} mac_lc_rec;
1.1317 +
1.1318 +/* Todo: This will be updated with a newer version from www.unicode.org web
1.1319 + page when it's available.*/
1.1320 +#define MAC_LC_MAGIC_NUMBER -5
1.1321 +#define MAC_LC_INIT_NUMBER -9
1.1322 +
1.1323 +static const mac_lc_rec mac_lc_recs[] = {
1.1324 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 0, "en_US",
1.1325 + /* United States*/
1.1326 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 1, "fr_FR",
1.1327 + /* France*/
1.1328 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 2, "en_GB",
1.1329 + /* Great Britain*/
1.1330 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 3, "de_DE",
1.1331 + /* Germany*/
1.1332 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 4, "it_IT",
1.1333 + /* Italy*/
1.1334 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 5, "nl_NL",
1.1335 + /* Metherlands*/
1.1336 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 6, "fr_BE",
1.1337 + /* French for Belgium or Lxembourg*/
1.1338 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 7, "sv_SE",
1.1339 + /* Sweden*/
1.1340 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 9, "da_DK",
1.1341 + /* Denmark*/
1.1342 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 10, "pt_PT",
1.1343 + /* Portugal*/
1.1344 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 11, "fr_CA",
1.1345 + /* French Canada*/
1.1346 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 13, "is_IS",
1.1347 + /* Israel*/
1.1348 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 14, "ja_JP",
1.1349 + /* Japan*/
1.1350 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 15, "en_AU",
1.1351 + /* Australia*/
1.1352 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 16, "ar_AE",
1.1353 + /* the Arabic world (?)*/
1.1354 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 17, "fi_FI",
1.1355 + /* Finland*/
1.1356 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 18, "fr_CH",
1.1357 + /* French for Switzerland*/
1.1358 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 19, "de_CH",
1.1359 + /* German for Switzerland*/
1.1360 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 20, "el_GR",
1.1361 + /* Greece*/
1.1362 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 21, "is_IS",
1.1363 + /* Iceland ===*/
1.1364 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 22, "",*/
1.1365 + /* Malta ===*/
1.1366 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 23, "",*/
1.1367 + /* Cyprus ===*/
1.1368 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 24, "tr_TR",
1.1369 + /* Turkey ===*/
1.1370 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 25, "sh_YU",
1.1371 + /* Croatian system for Yugoslavia*/
1.1372 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 33, "",*/
1.1373 + /* Hindi system for India*/
1.1374 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 34, "",*/
1.1375 + /* Pakistan*/
1.1376 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 41, "lt_LT",
1.1377 + /* Lithuania*/
1.1378 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 42, "pl_PL",
1.1379 + /* Poland*/
1.1380 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 43, "hu_HU",
1.1381 + /* Hungary*/
1.1382 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 44, "et_EE",
1.1383 + /* Estonia*/
1.1384 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 45, "lv_LV",
1.1385 + /* Latvia*/
1.1386 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 46, "",*/
1.1387 + /* Lapland [Ask Rich for the data. HS]*/
1.1388 + /*MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 47, "",*/
1.1389 + /* Faeroe Islands*/
1.1390 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 48, "fa_IR",
1.1391 + /* Iran*/
1.1392 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 49, "ru_RU",
1.1393 + /* Russia*/
1.1394 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 50, "en_IE",
1.1395 + /* Ireland*/
1.1396 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 51, "ko_KR",
1.1397 + /* Korea*/
1.1398 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 52, "zh_CN",
1.1399 + /* People's Republic of China*/
1.1400 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 53, "zh_TW",
1.1401 + /* Taiwan*/
1.1402 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, 54, "th_TH",
1.1403 + /* Thailand*/
1.1404 +
1.1405 + /* fallback is en_US*/
1.1406 + MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER, MAC_LC_MAGIC_NUMBER,
1.1407 + MAC_LC_MAGIC_NUMBER, "en_US"
1.1408 +};
1.1409 +
1.1410 +#endif
1.1411 +
1.1412 +#if U_POSIX_LOCALE
1.1413 +/* A helper function used by uprv_getPOSIXIDForDefaultLocale and
1.1414 + * uprv_getPOSIXIDForDefaultCodepage. Returns the posix locale id for
1.1415 + * LC_CTYPE and LC_MESSAGES. It doesn't support other locale categories.
1.1416 + */
1.1417 +static const char *uprv_getPOSIXIDForCategory(int category)
1.1418 +{
1.1419 + const char* posixID = NULL;
1.1420 + if (category == LC_MESSAGES || category == LC_CTYPE) {
1.1421 + /*
1.1422 + * On Solaris two different calls to setlocale can result in
1.1423 + * different values. Only get this value once.
1.1424 + *
1.1425 + * We must check this first because an application can set this.
1.1426 + *
1.1427 + * LC_ALL can't be used because it's platform dependent. The LANG
1.1428 + * environment variable seems to affect LC_CTYPE variable by default.
1.1429 + * Here is what setlocale(LC_ALL, NULL) can return.
1.1430 + * HPUX can return 'C C C C C C C'
1.1431 + * Solaris can return /en_US/C/C/C/C/C on the second try.
1.1432 + * Linux can return LC_CTYPE=C;LC_NUMERIC=C;...
1.1433 + *
1.1434 + * The default codepage detection also needs to use LC_CTYPE.
1.1435 + *
1.1436 + * Do not call setlocale(LC_*, "")! Using an empty string instead
1.1437 + * of NULL, will modify the libc behavior.
1.1438 + */
1.1439 + posixID = setlocale(category, NULL);
1.1440 + if ((posixID == 0)
1.1441 + || (uprv_strcmp("C", posixID) == 0)
1.1442 + || (uprv_strcmp("POSIX", posixID) == 0))
1.1443 + {
1.1444 + /* Maybe we got some garbage. Try something more reasonable */
1.1445 + posixID = getenv("LC_ALL");
1.1446 + if (posixID == 0) {
1.1447 + posixID = getenv(category == LC_MESSAGES ? "LC_MESSAGES" : "LC_CTYPE");
1.1448 + if (posixID == 0) {
1.1449 + posixID = getenv("LANG");
1.1450 + }
1.1451 + }
1.1452 + }
1.1453 + }
1.1454 + if ((posixID==0)
1.1455 + || (uprv_strcmp("C", posixID) == 0)
1.1456 + || (uprv_strcmp("POSIX", posixID) == 0))
1.1457 + {
1.1458 + /* Nothing worked. Give it a nice POSIX default value. */
1.1459 + posixID = "en_US_POSIX";
1.1460 + }
1.1461 + return posixID;
1.1462 +}
1.1463 +
1.1464 +/* Return just the POSIX id for the default locale, whatever happens to be in
1.1465 + * it. It gets the value from LC_MESSAGES and indirectly from LC_ALL and LANG.
1.1466 + */
1.1467 +static const char *uprv_getPOSIXIDForDefaultLocale(void)
1.1468 +{
1.1469 + static const char* posixID = NULL;
1.1470 + if (posixID == 0) {
1.1471 + posixID = uprv_getPOSIXIDForCategory(LC_MESSAGES);
1.1472 + }
1.1473 + return posixID;
1.1474 +}
1.1475 +
1.1476 +#if !U_CHARSET_IS_UTF8
1.1477 +/* Return just the POSIX id for the default codepage, whatever happens to be in
1.1478 + * it. It gets the value from LC_CTYPE and indirectly from LC_ALL and LANG.
1.1479 + */
1.1480 +static const char *uprv_getPOSIXIDForDefaultCodepage(void)
1.1481 +{
1.1482 + static const char* posixID = NULL;
1.1483 + if (posixID == 0) {
1.1484 + posixID = uprv_getPOSIXIDForCategory(LC_CTYPE);
1.1485 + }
1.1486 + return posixID;
1.1487 +}
1.1488 +#endif
1.1489 +#endif
1.1490 +
1.1491 +/* NOTE: The caller should handle thread safety */
1.1492 +U_CAPI const char* U_EXPORT2
1.1493 +uprv_getDefaultLocaleID()
1.1494 +{
1.1495 +#if U_POSIX_LOCALE
1.1496 +/*
1.1497 + Note that: (a '!' means the ID is improper somehow)
1.1498 + LC_ALL ----> default_loc codepage
1.1499 +--------------------------------------------------------
1.1500 + ab.CD ab CD
1.1501 + ab@CD ab__CD -
1.1502 + ab@CD.EF ab__CD EF
1.1503 +
1.1504 + ab_CD.EF@GH ab_CD_GH EF
1.1505 +
1.1506 +Some 'improper' ways to do the same as above:
1.1507 + ! ab_CD@GH.EF ab_CD_GH EF
1.1508 + ! ab_CD.EF@GH.IJ ab_CD_GH EF
1.1509 + ! ab_CD@ZZ.EF@GH.IJ ab_CD_GH EF
1.1510 +
1.1511 + _CD@GH _CD_GH -
1.1512 + _CD.EF@GH _CD_GH EF
1.1513 +
1.1514 +The variant cannot have dots in it.
1.1515 +The 'rightmost' variant (@xxx) wins.
1.1516 +The leftmost codepage (.xxx) wins.
1.1517 +*/
1.1518 + char *correctedPOSIXLocale = 0;
1.1519 + const char* posixID = uprv_getPOSIXIDForDefaultLocale();
1.1520 + const char *p;
1.1521 + const char *q;
1.1522 + int32_t len;
1.1523 +
1.1524 + /* Format: (no spaces)
1.1525 + ll [ _CC ] [ . MM ] [ @ VV]
1.1526 +
1.1527 + l = lang, C = ctry, M = charmap, V = variant
1.1528 + */
1.1529 +
1.1530 + if (gCorrectedPOSIXLocale != NULL) {
1.1531 + return gCorrectedPOSIXLocale;
1.1532 + }
1.1533 +
1.1534 + if ((p = uprv_strchr(posixID, '.')) != NULL) {
1.1535 + /* assume new locale can't be larger than old one? */
1.1536 + correctedPOSIXLocale = static_cast<char *>(uprv_malloc(uprv_strlen(posixID)+1));
1.1537 + /* Exit on memory allocation error. */
1.1538 + if (correctedPOSIXLocale == NULL) {
1.1539 + return NULL;
1.1540 + }
1.1541 + uprv_strncpy(correctedPOSIXLocale, posixID, p-posixID);
1.1542 + correctedPOSIXLocale[p-posixID] = 0;
1.1543 +
1.1544 + /* do not copy after the @ */
1.1545 + if ((p = uprv_strchr(correctedPOSIXLocale, '@')) != NULL) {
1.1546 + correctedPOSIXLocale[p-correctedPOSIXLocale] = 0;
1.1547 + }
1.1548 + }
1.1549 +
1.1550 + /* Note that we scan the *uncorrected* ID. */
1.1551 + if ((p = uprv_strrchr(posixID, '@')) != NULL) {
1.1552 + if (correctedPOSIXLocale == NULL) {
1.1553 + correctedPOSIXLocale = static_cast<char *>(uprv_malloc(uprv_strlen(posixID)+1));
1.1554 + /* Exit on memory allocation error. */
1.1555 + if (correctedPOSIXLocale == NULL) {
1.1556 + return NULL;
1.1557 + }
1.1558 + uprv_strncpy(correctedPOSIXLocale, posixID, p-posixID);
1.1559 + correctedPOSIXLocale[p-posixID] = 0;
1.1560 + }
1.1561 + p++;
1.1562 +
1.1563 + /* Take care of any special cases here.. */
1.1564 + if (!uprv_strcmp(p, "nynorsk")) {
1.1565 + p = "NY";
1.1566 + /* Don't worry about no__NY. In practice, it won't appear. */
1.1567 + }
1.1568 +
1.1569 + if (uprv_strchr(correctedPOSIXLocale,'_') == NULL) {
1.1570 + uprv_strcat(correctedPOSIXLocale, "__"); /* aa@b -> aa__b */
1.1571 + }
1.1572 + else {
1.1573 + uprv_strcat(correctedPOSIXLocale, "_"); /* aa_CC@b -> aa_CC_b */
1.1574 + }
1.1575 +
1.1576 + if ((q = uprv_strchr(p, '.')) != NULL) {
1.1577 + /* How big will the resulting string be? */
1.1578 + len = (int32_t)(uprv_strlen(correctedPOSIXLocale) + (q-p));
1.1579 + uprv_strncat(correctedPOSIXLocale, p, q-p);
1.1580 + correctedPOSIXLocale[len] = 0;
1.1581 + }
1.1582 + else {
1.1583 + /* Anything following the @ sign */
1.1584 + uprv_strcat(correctedPOSIXLocale, p);
1.1585 + }
1.1586 +
1.1587 + /* Should there be a map from 'no@nynorsk' -> no_NO_NY here?
1.1588 + * How about 'russian' -> 'ru'?
1.1589 + * Many of the other locales using ISO codes will be handled by the
1.1590 + * canonicalization functions in uloc_getDefault.
1.1591 + */
1.1592 + }
1.1593 +
1.1594 + /* Was a correction made? */
1.1595 + if (correctedPOSIXLocale != NULL) {
1.1596 + posixID = correctedPOSIXLocale;
1.1597 + }
1.1598 + else {
1.1599 + /* copy it, just in case the original pointer goes away. See j2395 */
1.1600 + correctedPOSIXLocale = (char *)uprv_malloc(uprv_strlen(posixID) + 1);
1.1601 + /* Exit on memory allocation error. */
1.1602 + if (correctedPOSIXLocale == NULL) {
1.1603 + return NULL;
1.1604 + }
1.1605 + posixID = uprv_strcpy(correctedPOSIXLocale, posixID);
1.1606 + }
1.1607 +
1.1608 + if (gCorrectedPOSIXLocale == NULL) {
1.1609 + gCorrectedPOSIXLocale = correctedPOSIXLocale;
1.1610 + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup);
1.1611 + correctedPOSIXLocale = NULL;
1.1612 + }
1.1613 +
1.1614 + if (correctedPOSIXLocale != NULL) { /* Was already set - clean up. */
1.1615 + uprv_free(correctedPOSIXLocale);
1.1616 + }
1.1617 +
1.1618 + return posixID;
1.1619 +
1.1620 +#elif U_PLATFORM_USES_ONLY_WIN32_API
1.1621 +#define POSIX_LOCALE_CAPACITY 64
1.1622 + UErrorCode status = U_ZERO_ERROR;
1.1623 + char *correctedPOSIXLocale = 0;
1.1624 +
1.1625 + if (gCorrectedPOSIXLocale != NULL) {
1.1626 + return gCorrectedPOSIXLocale;
1.1627 + }
1.1628 +
1.1629 + LCID id = GetThreadLocale();
1.1630 + correctedPOSIXLocale = static_cast<char *>(uprv_malloc(POSIX_LOCALE_CAPACITY + 1));
1.1631 + if (correctedPOSIXLocale) {
1.1632 + int32_t posixLen = uprv_convertToPosix(id, correctedPOSIXLocale, POSIX_LOCALE_CAPACITY, &status);
1.1633 + if (U_SUCCESS(status)) {
1.1634 + *(correctedPOSIXLocale + posixLen) = 0;
1.1635 + gCorrectedPOSIXLocale = correctedPOSIXLocale;
1.1636 + ucln_common_registerCleanup(UCLN_COMMON_PUTIL, putil_cleanup);
1.1637 + } else {
1.1638 + uprv_free(correctedPOSIXLocale);
1.1639 + }
1.1640 + }
1.1641 +
1.1642 + if (gCorrectedPOSIXLocale == NULL) {
1.1643 + return "en_US";
1.1644 + }
1.1645 + return gCorrectedPOSIXLocale;
1.1646 +
1.1647 +#elif U_PLATFORM == U_PF_CLASSIC_MACOS
1.1648 + int32_t script = MAC_LC_INIT_NUMBER;
1.1649 + /* = IntlScript(); or GetScriptManagerVariable(smSysScript);*/
1.1650 + int32_t region = MAC_LC_INIT_NUMBER;
1.1651 + /* = GetScriptManagerVariable(smRegionCode);*/
1.1652 + int32_t lang = MAC_LC_INIT_NUMBER;
1.1653 + /* = GetScriptManagerVariable(smScriptLang);*/
1.1654 + int32_t date_region = MAC_LC_INIT_NUMBER;
1.1655 + const char* posixID = 0;
1.1656 + int32_t count = sizeof(mac_lc_recs) / sizeof(mac_lc_rec);
1.1657 + int32_t i;
1.1658 + Intl1Hndl ih;
1.1659 +
1.1660 + ih = (Intl1Hndl) GetIntlResource(1);
1.1661 + if (ih)
1.1662 + date_region = ((uint16_t)(*ih)->intl1Vers) >> 8;
1.1663 +
1.1664 + for (i = 0; i < count; i++) {
1.1665 + if ( ((mac_lc_recs[i].script == MAC_LC_MAGIC_NUMBER)
1.1666 + || (mac_lc_recs[i].script == script))
1.1667 + && ((mac_lc_recs[i].region == MAC_LC_MAGIC_NUMBER)
1.1668 + || (mac_lc_recs[i].region == region))
1.1669 + && ((mac_lc_recs[i].lang == MAC_LC_MAGIC_NUMBER)
1.1670 + || (mac_lc_recs[i].lang == lang))
1.1671 + && ((mac_lc_recs[i].date_region == MAC_LC_MAGIC_NUMBER)
1.1672 + || (mac_lc_recs[i].date_region == date_region))
1.1673 + )
1.1674 + {
1.1675 + posixID = mac_lc_recs[i].posixID;
1.1676 + break;
1.1677 + }
1.1678 + }
1.1679 +
1.1680 + return posixID;
1.1681 +
1.1682 +#elif U_PLATFORM == U_PF_OS400
1.1683 + /* locales are process scoped and are by definition thread safe */
1.1684 + static char correctedLocale[64];
1.1685 + const char *localeID = getenv("LC_ALL");
1.1686 + char *p;
1.1687 +
1.1688 + if (localeID == NULL)
1.1689 + localeID = getenv("LANG");
1.1690 + if (localeID == NULL)
1.1691 + localeID = setlocale(LC_ALL, NULL);
1.1692 + /* Make sure we have something... */
1.1693 + if (localeID == NULL)
1.1694 + return "en_US_POSIX";
1.1695 +
1.1696 + /* Extract the locale name from the path. */
1.1697 + if((p = uprv_strrchr(localeID, '/')) != NULL)
1.1698 + {
1.1699 + /* Increment p to start of locale name. */
1.1700 + p++;
1.1701 + localeID = p;
1.1702 + }
1.1703 +
1.1704 + /* Copy to work location. */
1.1705 + uprv_strcpy(correctedLocale, localeID);
1.1706 +
1.1707 + /* Strip off the '.locale' extension. */
1.1708 + if((p = uprv_strchr(correctedLocale, '.')) != NULL) {
1.1709 + *p = 0;
1.1710 + }
1.1711 +
1.1712 + /* Upper case the locale name. */
1.1713 + T_CString_toUpperCase(correctedLocale);
1.1714 +
1.1715 + /* See if we are using the POSIX locale. Any of the
1.1716 + * following are equivalent and use the same QLGPGCMA
1.1717 + * (POSIX) locale.
1.1718 + * QLGPGCMA2 means UCS2
1.1719 + * QLGPGCMA_4 means UTF-32
1.1720 + * QLGPGCMA_8 means UTF-8
1.1721 + */
1.1722 + if ((uprv_strcmp("C", correctedLocale) == 0) ||
1.1723 + (uprv_strcmp("POSIX", correctedLocale) == 0) ||
1.1724 + (uprv_strncmp("QLGPGCMA", correctedLocale, 8) == 0))
1.1725 + {
1.1726 + uprv_strcpy(correctedLocale, "en_US_POSIX");
1.1727 + }
1.1728 + else
1.1729 + {
1.1730 + int16_t LocaleLen;
1.1731 +
1.1732 + /* Lower case the lang portion. */
1.1733 + for(p = correctedLocale; *p != 0 && *p != '_'; p++)
1.1734 + {
1.1735 + *p = uprv_tolower(*p);
1.1736 + }
1.1737 +
1.1738 + /* Adjust for Euro. After '_E' add 'URO'. */
1.1739 + LocaleLen = uprv_strlen(correctedLocale);
1.1740 + if (correctedLocale[LocaleLen - 2] == '_' &&
1.1741 + correctedLocale[LocaleLen - 1] == 'E')
1.1742 + {
1.1743 + uprv_strcat(correctedLocale, "URO");
1.1744 + }
1.1745 +
1.1746 + /* If using Lotus-based locale then convert to
1.1747 + * equivalent non Lotus.
1.1748 + */
1.1749 + else if (correctedLocale[LocaleLen - 2] == '_' &&
1.1750 + correctedLocale[LocaleLen - 1] == 'L')
1.1751 + {
1.1752 + correctedLocale[LocaleLen - 2] = 0;
1.1753 + }
1.1754 +
1.1755 + /* There are separate simplified and traditional
1.1756 + * locales called zh_HK_S and zh_HK_T.
1.1757 + */
1.1758 + else if (uprv_strncmp(correctedLocale, "zh_HK", 5) == 0)
1.1759 + {
1.1760 + uprv_strcpy(correctedLocale, "zh_HK");
1.1761 + }
1.1762 +
1.1763 + /* A special zh_CN_GBK locale...
1.1764 + */
1.1765 + else if (uprv_strcmp(correctedLocale, "zh_CN_GBK") == 0)
1.1766 + {
1.1767 + uprv_strcpy(correctedLocale, "zh_CN");
1.1768 + }
1.1769 +
1.1770 + }
1.1771 +
1.1772 + return correctedLocale;
1.1773 +#endif
1.1774 +
1.1775 +}
1.1776 +
1.1777 +#if !U_CHARSET_IS_UTF8
1.1778 +#if U_POSIX_LOCALE
1.1779 +/*
1.1780 +Due to various platform differences, one platform may specify a charset,
1.1781 +when they really mean a different charset. Remap the names so that they are
1.1782 +compatible with ICU. Only conflicting/ambiguous aliases should be resolved
1.1783 +here. Before adding anything to this function, please consider adding unique
1.1784 +names to the ICU alias table in the data directory.
1.1785 +*/
1.1786 +static const char*
1.1787 +remapPlatformDependentCodepage(const char *locale, const char *name) {
1.1788 + if (locale != NULL && *locale == 0) {
1.1789 + /* Make sure that an empty locale is handled the same way. */
1.1790 + locale = NULL;
1.1791 + }
1.1792 + if (name == NULL) {
1.1793 + return NULL;
1.1794 + }
1.1795 +#if U_PLATFORM == U_PF_AIX
1.1796 + if (uprv_strcmp(name, "IBM-943") == 0) {
1.1797 + /* Use the ASCII compatible ibm-943 */
1.1798 + name = "Shift-JIS";
1.1799 + }
1.1800 + else if (uprv_strcmp(name, "IBM-1252") == 0) {
1.1801 + /* Use the windows-1252 that contains the Euro */
1.1802 + name = "IBM-5348";
1.1803 + }
1.1804 +#elif U_PLATFORM == U_PF_SOLARIS
1.1805 + if (locale != NULL && uprv_strcmp(name, "EUC") == 0) {
1.1806 + /* Solaris underspecifies the "EUC" name. */
1.1807 + if (uprv_strcmp(locale, "zh_CN") == 0) {
1.1808 + name = "EUC-CN";
1.1809 + }
1.1810 + else if (uprv_strcmp(locale, "zh_TW") == 0) {
1.1811 + name = "EUC-TW";
1.1812 + }
1.1813 + else if (uprv_strcmp(locale, "ko_KR") == 0) {
1.1814 + name = "EUC-KR";
1.1815 + }
1.1816 + }
1.1817 + else if (uprv_strcmp(name, "eucJP") == 0) {
1.1818 + /*
1.1819 + ibm-954 is the best match.
1.1820 + ibm-33722 is the default for eucJP (similar to Windows).
1.1821 + */
1.1822 + name = "eucjis";
1.1823 + }
1.1824 + else if (uprv_strcmp(name, "646") == 0) {
1.1825 + /*
1.1826 + * The default codepage given by Solaris is 646 but the C library routines treat it as if it was
1.1827 + * ISO-8859-1 instead of US-ASCII(646).
1.1828 + */
1.1829 + name = "ISO-8859-1";
1.1830 + }
1.1831 +#elif U_PLATFORM_IS_DARWIN_BASED
1.1832 + if (locale == NULL && *name == 0) {
1.1833 + /*
1.1834 + No locale was specified, and an empty name was passed in.
1.1835 + This usually indicates that nl_langinfo didn't return valid information.
1.1836 + Mac OS X uses UTF-8 by default (especially the locale data and console).
1.1837 + */
1.1838 + name = "UTF-8";
1.1839 + }
1.1840 + else if (uprv_strcmp(name, "CP949") == 0) {
1.1841 + /* Remap CP949 to a similar codepage to avoid issues with backslash and won symbol. */
1.1842 + name = "EUC-KR";
1.1843 + }
1.1844 + else if (locale != NULL && uprv_strcmp(locale, "en_US_POSIX") != 0 && uprv_strcmp(name, "US-ASCII") == 0) {
1.1845 + /*
1.1846 + * For non C/POSIX locale, default the code page to UTF-8 instead of US-ASCII.
1.1847 + */
1.1848 + name = "UTF-8";
1.1849 + }
1.1850 +#elif U_PLATFORM == U_PF_BSD
1.1851 + if (uprv_strcmp(name, "CP949") == 0) {
1.1852 + /* Remap CP949 to a similar codepage to avoid issues with backslash and won symbol. */
1.1853 + name = "EUC-KR";
1.1854 + }
1.1855 +#elif U_PLATFORM == U_PF_HPUX
1.1856 + if (locale != NULL && uprv_strcmp(locale, "zh_HK") == 0 && uprv_strcmp(name, "big5") == 0) {
1.1857 + /* HP decided to extend big5 as hkbig5 even though it's not compatible :-( */
1.1858 + /* zh_TW.big5 is not the same charset as zh_HK.big5! */
1.1859 + name = "hkbig5";
1.1860 + }
1.1861 + else if (uprv_strcmp(name, "eucJP") == 0) {
1.1862 + /*
1.1863 + ibm-1350 is the best match, but unavailable.
1.1864 + ibm-954 is mostly a superset of ibm-1350.
1.1865 + ibm-33722 is the default for eucJP (similar to Windows).
1.1866 + */
1.1867 + name = "eucjis";
1.1868 + }
1.1869 +#elif U_PLATFORM == U_PF_LINUX
1.1870 + if (locale != NULL && uprv_strcmp(name, "euc") == 0) {
1.1871 + /* Linux underspecifies the "EUC" name. */
1.1872 + if (uprv_strcmp(locale, "korean") == 0) {
1.1873 + name = "EUC-KR";
1.1874 + }
1.1875 + else if (uprv_strcmp(locale, "japanese") == 0) {
1.1876 + /* See comment below about eucJP */
1.1877 + name = "eucjis";
1.1878 + }
1.1879 + }
1.1880 + else if (uprv_strcmp(name, "eucjp") == 0) {
1.1881 + /*
1.1882 + ibm-1350 is the best match, but unavailable.
1.1883 + ibm-954 is mostly a superset of ibm-1350.
1.1884 + ibm-33722 is the default for eucJP (similar to Windows).
1.1885 + */
1.1886 + name = "eucjis";
1.1887 + }
1.1888 + else if (locale != NULL && uprv_strcmp(locale, "en_US_POSIX") != 0 &&
1.1889 + (uprv_strcmp(name, "ANSI_X3.4-1968") == 0 || uprv_strcmp(name, "US-ASCII") == 0)) {
1.1890 + /*
1.1891 + * For non C/POSIX locale, default the code page to UTF-8 instead of US-ASCII.
1.1892 + */
1.1893 + name = "UTF-8";
1.1894 + }
1.1895 + /*
1.1896 + * Linux returns ANSI_X3.4-1968 for C/POSIX, but the call site takes care of
1.1897 + * it by falling back to 'US-ASCII' when NULL is returned from this
1.1898 + * function. So, we don't have to worry about it here.
1.1899 + */
1.1900 +#endif
1.1901 + /* return NULL when "" is passed in */
1.1902 + if (*name == 0) {
1.1903 + name = NULL;
1.1904 + }
1.1905 + return name;
1.1906 +}
1.1907 +
1.1908 +static const char*
1.1909 +getCodepageFromPOSIXID(const char *localeName, char * buffer, int32_t buffCapacity)
1.1910 +{
1.1911 + char localeBuf[100];
1.1912 + const char *name = NULL;
1.1913 + char *variant = NULL;
1.1914 +
1.1915 + if (localeName != NULL && (name = (uprv_strchr(localeName, '.'))) != NULL) {
1.1916 + size_t localeCapacity = uprv_min(sizeof(localeBuf), (name-localeName)+1);
1.1917 + uprv_strncpy(localeBuf, localeName, localeCapacity);
1.1918 + localeBuf[localeCapacity-1] = 0; /* ensure NULL termination */
1.1919 + name = uprv_strncpy(buffer, name+1, buffCapacity);
1.1920 + buffer[buffCapacity-1] = 0; /* ensure NULL termination */
1.1921 + if ((variant = const_cast<char *>(uprv_strchr(name, '@'))) != NULL) {
1.1922 + *variant = 0;
1.1923 + }
1.1924 + name = remapPlatformDependentCodepage(localeBuf, name);
1.1925 + }
1.1926 + return name;
1.1927 +}
1.1928 +#endif
1.1929 +
1.1930 +static const char*
1.1931 +int_getDefaultCodepage()
1.1932 +{
1.1933 +#if U_PLATFORM == U_PF_OS400
1.1934 + uint32_t ccsid = 37; /* Default to ibm-37 */
1.1935 + static char codepage[64];
1.1936 + Qwc_JOBI0400_t jobinfo;
1.1937 + Qus_EC_t error = { sizeof(Qus_EC_t) }; /* SPI error code */
1.1938 +
1.1939 + EPT_CALL(QUSRJOBI)(&jobinfo, sizeof(jobinfo), "JOBI0400",
1.1940 + "* ", " ", &error);
1.1941 +
1.1942 + if (error.Bytes_Available == 0) {
1.1943 + if (jobinfo.Coded_Char_Set_ID != 0xFFFF) {
1.1944 + ccsid = (uint32_t)jobinfo.Coded_Char_Set_ID;
1.1945 + }
1.1946 + else if (jobinfo.Default_Coded_Char_Set_Id != 0xFFFF) {
1.1947 + ccsid = (uint32_t)jobinfo.Default_Coded_Char_Set_Id;
1.1948 + }
1.1949 + /* else use the default */
1.1950 + }
1.1951 + sprintf(codepage,"ibm-%d", ccsid);
1.1952 + return codepage;
1.1953 +
1.1954 +#elif U_PLATFORM == U_PF_OS390
1.1955 + static char codepage[64];
1.1956 +
1.1957 + strncpy(codepage, nl_langinfo(CODESET),63-strlen(UCNV_SWAP_LFNL_OPTION_STRING));
1.1958 + strcat(codepage,UCNV_SWAP_LFNL_OPTION_STRING);
1.1959 + codepage[63] = 0; /* NULL terminate */
1.1960 +
1.1961 + return codepage;
1.1962 +
1.1963 +#elif U_PLATFORM == U_PF_CLASSIC_MACOS
1.1964 + return "macintosh"; /* TODO: Macintosh Roman. There must be a better way. fixme! */
1.1965 +
1.1966 +#elif U_PLATFORM_USES_ONLY_WIN32_API
1.1967 + static char codepage[64];
1.1968 + sprintf(codepage, "windows-%d", GetACP());
1.1969 + return codepage;
1.1970 +
1.1971 +#elif U_POSIX_LOCALE
1.1972 + static char codesetName[100];
1.1973 + const char *localeName = NULL;
1.1974 + const char *name = NULL;
1.1975 +
1.1976 + localeName = uprv_getPOSIXIDForDefaultCodepage();
1.1977 + uprv_memset(codesetName, 0, sizeof(codesetName));
1.1978 +#if U_HAVE_NL_LANGINFO_CODESET
1.1979 + /* When available, check nl_langinfo first because it usually gives more
1.1980 + useful names. It depends on LC_CTYPE.
1.1981 + nl_langinfo may use the same buffer as setlocale. */
1.1982 + {
1.1983 + const char *codeset = nl_langinfo(U_NL_LANGINFO_CODESET);
1.1984 +#if U_PLATFORM_IS_DARWIN_BASED || U_PLATFORM_IS_LINUX_BASED
1.1985 + /*
1.1986 + * On Linux and MacOSX, ensure that default codepage for non C/POSIX locale is UTF-8
1.1987 + * instead of ASCII.
1.1988 + */
1.1989 + if (uprv_strcmp(localeName, "en_US_POSIX") != 0) {
1.1990 + codeset = remapPlatformDependentCodepage(localeName, codeset);
1.1991 + } else
1.1992 +#endif
1.1993 + {
1.1994 + codeset = remapPlatformDependentCodepage(NULL, codeset);
1.1995 + }
1.1996 +
1.1997 + if (codeset != NULL) {
1.1998 + uprv_strncpy(codesetName, codeset, sizeof(codesetName));
1.1999 + codesetName[sizeof(codesetName)-1] = 0;
1.2000 + return codesetName;
1.2001 + }
1.2002 + }
1.2003 +#endif
1.2004 +
1.2005 + /* Use setlocale in a nice way, and then check some environment variables.
1.2006 + Maybe the application used setlocale already.
1.2007 + */
1.2008 + uprv_memset(codesetName, 0, sizeof(codesetName));
1.2009 + name = getCodepageFromPOSIXID(localeName, codesetName, sizeof(codesetName));
1.2010 + if (name) {
1.2011 + /* if we can find the codeset name from setlocale, return that. */
1.2012 + return name;
1.2013 + }
1.2014 +
1.2015 + if (*codesetName == 0)
1.2016 + {
1.2017 + /* Everything failed. Return US ASCII (ISO 646). */
1.2018 + (void)uprv_strcpy(codesetName, "US-ASCII");
1.2019 + }
1.2020 + return codesetName;
1.2021 +#else
1.2022 + return "US-ASCII";
1.2023 +#endif
1.2024 +}
1.2025 +
1.2026 +
1.2027 +U_CAPI const char* U_EXPORT2
1.2028 +uprv_getDefaultCodepage()
1.2029 +{
1.2030 + static char const *name = NULL;
1.2031 + umtx_lock(NULL);
1.2032 + if (name == NULL) {
1.2033 + name = int_getDefaultCodepage();
1.2034 + }
1.2035 + umtx_unlock(NULL);
1.2036 + return name;
1.2037 +}
1.2038 +#endif /* !U_CHARSET_IS_UTF8 */
1.2039 +
1.2040 +
1.2041 +/* end of platform-specific implementation -------------- */
1.2042 +
1.2043 +/* version handling --------------------------------------------------------- */
1.2044 +
1.2045 +U_CAPI void U_EXPORT2
1.2046 +u_versionFromString(UVersionInfo versionArray, const char *versionString) {
1.2047 + char *end;
1.2048 + uint16_t part=0;
1.2049 +
1.2050 + if(versionArray==NULL) {
1.2051 + return;
1.2052 + }
1.2053 +
1.2054 + if(versionString!=NULL) {
1.2055 + for(;;) {
1.2056 + versionArray[part]=(uint8_t)uprv_strtoul(versionString, &end, 10);
1.2057 + if(end==versionString || ++part==U_MAX_VERSION_LENGTH || *end!=U_VERSION_DELIMITER) {
1.2058 + break;
1.2059 + }
1.2060 + versionString=end+1;
1.2061 + }
1.2062 + }
1.2063 +
1.2064 + while(part<U_MAX_VERSION_LENGTH) {
1.2065 + versionArray[part++]=0;
1.2066 + }
1.2067 +}
1.2068 +
1.2069 +U_CAPI void U_EXPORT2
1.2070 +u_versionFromUString(UVersionInfo versionArray, const UChar *versionString) {
1.2071 + if(versionArray!=NULL && versionString!=NULL) {
1.2072 + char versionChars[U_MAX_VERSION_STRING_LENGTH+1];
1.2073 + int32_t len = u_strlen(versionString);
1.2074 + if(len>U_MAX_VERSION_STRING_LENGTH) {
1.2075 + len = U_MAX_VERSION_STRING_LENGTH;
1.2076 + }
1.2077 + u_UCharsToChars(versionString, versionChars, len);
1.2078 + versionChars[len]=0;
1.2079 + u_versionFromString(versionArray, versionChars);
1.2080 + }
1.2081 +}
1.2082 +
1.2083 +U_CAPI void U_EXPORT2
1.2084 +u_versionToString(const UVersionInfo versionArray, char *versionString) {
1.2085 + uint16_t count, part;
1.2086 + uint8_t field;
1.2087 +
1.2088 + if(versionString==NULL) {
1.2089 + return;
1.2090 + }
1.2091 +
1.2092 + if(versionArray==NULL) {
1.2093 + versionString[0]=0;
1.2094 + return;
1.2095 + }
1.2096 +
1.2097 + /* count how many fields need to be written */
1.2098 + for(count=4; count>0 && versionArray[count-1]==0; --count) {
1.2099 + }
1.2100 +
1.2101 + if(count <= 1) {
1.2102 + count = 2;
1.2103 + }
1.2104 +
1.2105 + /* write the first part */
1.2106 + /* write the decimal field value */
1.2107 + field=versionArray[0];
1.2108 + if(field>=100) {
1.2109 + *versionString++=(char)('0'+field/100);
1.2110 + field%=100;
1.2111 + }
1.2112 + if(field>=10) {
1.2113 + *versionString++=(char)('0'+field/10);
1.2114 + field%=10;
1.2115 + }
1.2116 + *versionString++=(char)('0'+field);
1.2117 +
1.2118 + /* write the following parts */
1.2119 + for(part=1; part<count; ++part) {
1.2120 + /* write a dot first */
1.2121 + *versionString++=U_VERSION_DELIMITER;
1.2122 +
1.2123 + /* write the decimal field value */
1.2124 + field=versionArray[part];
1.2125 + if(field>=100) {
1.2126 + *versionString++=(char)('0'+field/100);
1.2127 + field%=100;
1.2128 + }
1.2129 + if(field>=10) {
1.2130 + *versionString++=(char)('0'+field/10);
1.2131 + field%=10;
1.2132 + }
1.2133 + *versionString++=(char)('0'+field);
1.2134 + }
1.2135 +
1.2136 + /* NUL-terminate */
1.2137 + *versionString=0;
1.2138 +}
1.2139 +
1.2140 +U_CAPI void U_EXPORT2
1.2141 +u_getVersion(UVersionInfo versionArray) {
1.2142 + u_versionFromString(versionArray, U_ICU_VERSION);
1.2143 +}
1.2144 +
1.2145 +/**
1.2146 + * icucfg.h dependent code
1.2147 + */
1.2148 +
1.2149 +#if U_ENABLE_DYLOAD
1.2150 +
1.2151 +#if HAVE_DLOPEN && !U_PLATFORM_USES_ONLY_WIN32_API
1.2152 +
1.2153 +#if HAVE_DLFCN_H
1.2154 +
1.2155 +#ifdef __MVS__
1.2156 +#ifndef __SUSV3
1.2157 +#define __SUSV3 1
1.2158 +#endif
1.2159 +#endif
1.2160 +#include <dlfcn.h>
1.2161 +#endif
1.2162 +
1.2163 +U_INTERNAL void * U_EXPORT2
1.2164 +uprv_dl_open(const char *libName, UErrorCode *status) {
1.2165 + void *ret = NULL;
1.2166 + if(U_FAILURE(*status)) return ret;
1.2167 + ret = dlopen(libName, RTLD_NOW|RTLD_GLOBAL);
1.2168 + if(ret==NULL) {
1.2169 +#ifdef U_TRACE_DYLOAD
1.2170 + printf("dlerror on dlopen(%s): %s\n", libName, dlerror());
1.2171 +#endif
1.2172 + *status = U_MISSING_RESOURCE_ERROR;
1.2173 + }
1.2174 + return ret;
1.2175 +}
1.2176 +
1.2177 +U_INTERNAL void U_EXPORT2
1.2178 +uprv_dl_close(void *lib, UErrorCode *status) {
1.2179 + if(U_FAILURE(*status)) return;
1.2180 + dlclose(lib);
1.2181 +}
1.2182 +
1.2183 +U_INTERNAL UVoidFunction* U_EXPORT2
1.2184 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) {
1.2185 + union {
1.2186 + UVoidFunction *fp;
1.2187 + void *vp;
1.2188 + } uret;
1.2189 + uret.fp = NULL;
1.2190 + if(U_FAILURE(*status)) return uret.fp;
1.2191 + uret.vp = dlsym(lib, sym);
1.2192 + if(uret.vp == NULL) {
1.2193 +#ifdef U_TRACE_DYLOAD
1.2194 + printf("dlerror on dlsym(%p,%s): %s\n", lib,sym, dlerror());
1.2195 +#endif
1.2196 + *status = U_MISSING_RESOURCE_ERROR;
1.2197 + }
1.2198 + return uret.fp;
1.2199 +}
1.2200 +
1.2201 +#else
1.2202 +
1.2203 +/* null (nonexistent) implementation. */
1.2204 +
1.2205 +U_INTERNAL void * U_EXPORT2
1.2206 +uprv_dl_open(const char *libName, UErrorCode *status) {
1.2207 + if(U_FAILURE(*status)) return NULL;
1.2208 + *status = U_UNSUPPORTED_ERROR;
1.2209 + return NULL;
1.2210 +}
1.2211 +
1.2212 +U_INTERNAL void U_EXPORT2
1.2213 +uprv_dl_close(void *lib, UErrorCode *status) {
1.2214 + if(U_FAILURE(*status)) return;
1.2215 + *status = U_UNSUPPORTED_ERROR;
1.2216 + return;
1.2217 +}
1.2218 +
1.2219 +
1.2220 +U_INTERNAL UVoidFunction* U_EXPORT2
1.2221 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) {
1.2222 + if(U_SUCCESS(*status)) {
1.2223 + *status = U_UNSUPPORTED_ERROR;
1.2224 + }
1.2225 + return (UVoidFunction*)NULL;
1.2226 +}
1.2227 +
1.2228 +
1.2229 +
1.2230 +#endif
1.2231 +
1.2232 +#elif U_PLATFORM_USES_ONLY_WIN32_API
1.2233 +
1.2234 +U_INTERNAL void * U_EXPORT2
1.2235 +uprv_dl_open(const char *libName, UErrorCode *status) {
1.2236 + HMODULE lib = NULL;
1.2237 +
1.2238 + if(U_FAILURE(*status)) return NULL;
1.2239 +
1.2240 + lib = LoadLibraryA(libName);
1.2241 +
1.2242 + if(lib==NULL) {
1.2243 + *status = U_MISSING_RESOURCE_ERROR;
1.2244 + }
1.2245 +
1.2246 + return (void*)lib;
1.2247 +}
1.2248 +
1.2249 +U_INTERNAL void U_EXPORT2
1.2250 +uprv_dl_close(void *lib, UErrorCode *status) {
1.2251 + HMODULE handle = (HMODULE)lib;
1.2252 + if(U_FAILURE(*status)) return;
1.2253 +
1.2254 + FreeLibrary(handle);
1.2255 +
1.2256 + return;
1.2257 +}
1.2258 +
1.2259 +
1.2260 +U_INTERNAL UVoidFunction* U_EXPORT2
1.2261 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) {
1.2262 + HMODULE handle = (HMODULE)lib;
1.2263 + UVoidFunction* addr = NULL;
1.2264 +
1.2265 + if(U_FAILURE(*status) || lib==NULL) return NULL;
1.2266 +
1.2267 + addr = (UVoidFunction*)GetProcAddress(handle, sym);
1.2268 +
1.2269 + if(addr==NULL) {
1.2270 + DWORD lastError = GetLastError();
1.2271 + if(lastError == ERROR_PROC_NOT_FOUND) {
1.2272 + *status = U_MISSING_RESOURCE_ERROR;
1.2273 + } else {
1.2274 + *status = U_UNSUPPORTED_ERROR; /* other unknown error. */
1.2275 + }
1.2276 + }
1.2277 +
1.2278 + return addr;
1.2279 +}
1.2280 +
1.2281 +
1.2282 +#else
1.2283 +
1.2284 +/* No dynamic loading set. */
1.2285 +
1.2286 +U_INTERNAL void * U_EXPORT2
1.2287 +uprv_dl_open(const char *libName, UErrorCode *status) {
1.2288 + if(U_FAILURE(*status)) return NULL;
1.2289 + *status = U_UNSUPPORTED_ERROR;
1.2290 + return NULL;
1.2291 +}
1.2292 +
1.2293 +U_INTERNAL void U_EXPORT2
1.2294 +uprv_dl_close(void *lib, UErrorCode *status) {
1.2295 + if(U_FAILURE(*status)) return;
1.2296 + *status = U_UNSUPPORTED_ERROR;
1.2297 + return;
1.2298 +}
1.2299 +
1.2300 +
1.2301 +U_INTERNAL UVoidFunction* U_EXPORT2
1.2302 +uprv_dlsym_func(void *lib, const char* sym, UErrorCode *status) {
1.2303 + if(U_SUCCESS(*status)) {
1.2304 + *status = U_UNSUPPORTED_ERROR;
1.2305 + }
1.2306 + return (UVoidFunction*)NULL;
1.2307 +}
1.2308 +
1.2309 +#endif /* U_ENABLE_DYLOAD */
1.2310 +
1.2311 +/*
1.2312 + * Hey, Emacs, please set the following:
1.2313 + *
1.2314 + * Local Variables:
1.2315 + * indent-tabs-mode: nil
1.2316 + * End:
1.2317 + *
1.2318 + */
