intl/icu/source/common/locutil.cpp

Sat, 03 Jan 2015 20:18:00 +0100

author
Michael Schloh von Bennewitz <michael@schloh.com>
date
Sat, 03 Jan 2015 20:18:00 +0100
branch
TOR_BUG_3246
changeset 7
129ffea94266
permissions
-rw-r--r--

Conditionally enable double key logic according to:
private browsing mode or privacy.thirdparty.isolate preference and
implement in GetCookieStringCommon and FindCookie where it counts...
With some reservations of how to convince FindCookie users to test
condition and pass a nullptr when disabling double key logic.

michael@0 1 /*
michael@0 2 *******************************************************************************
michael@0 3 * Copyright (C) 2002-2011, International Business Machines Corporation and
michael@0 4 * others. All Rights Reserved.
michael@0 5 *******************************************************************************
michael@0 6 */
michael@0 7 #include "unicode/utypes.h"
michael@0 8
michael@0 9 #if !UCONFIG_NO_SERVICE || !UCONFIG_NO_TRANSLITERATION
michael@0 10
michael@0 11 #include "unicode/resbund.h"
michael@0 12 #include "cmemory.h"
michael@0 13 #include "ustrfmt.h"
michael@0 14 #include "locutil.h"
michael@0 15 #include "charstr.h"
michael@0 16 #include "ucln_cmn.h"
michael@0 17 #include "uassert.h"
michael@0 18 #include "umutex.h"
michael@0 19
michael@0 20 // see LocaleUtility::getAvailableLocaleNames
michael@0 21 static icu::Hashtable * LocaleUtility_cache = NULL;
michael@0 22
michael@0 23 #define UNDERSCORE_CHAR ((UChar)0x005f)
michael@0 24 #define AT_SIGN_CHAR ((UChar)64)
michael@0 25 #define PERIOD_CHAR ((UChar)46)
michael@0 26
michael@0 27 /*
michael@0 28 ******************************************************************
michael@0 29 */
michael@0 30
michael@0 31 /**
michael@0 32 * Release all static memory held by Locale Utility.
michael@0 33 */
michael@0 34 U_CDECL_BEGIN
michael@0 35 static UBool U_CALLCONV service_cleanup(void) {
michael@0 36 if (LocaleUtility_cache) {
michael@0 37 delete LocaleUtility_cache;
michael@0 38 LocaleUtility_cache = NULL;
michael@0 39 }
michael@0 40 return TRUE;
michael@0 41 }
michael@0 42 U_CDECL_END
michael@0 43
michael@0 44 U_NAMESPACE_BEGIN
michael@0 45
michael@0 46 UnicodeString&
michael@0 47 LocaleUtility::canonicalLocaleString(const UnicodeString* id, UnicodeString& result)
michael@0 48 {
michael@0 49 if (id == NULL) {
michael@0 50 result.setToBogus();
michael@0 51 } else {
michael@0 52 // Fix case only (no other changes) up to the first '@' or '.' or
michael@0 53 // end of string, whichever comes first. In 3.0 I changed this to
michael@0 54 // stop at first '@' or '.'. It used to run out to the end of
michael@0 55 // string. My fix makes the tests pass but is probably
michael@0 56 // structurally incorrect. See below. [alan 3.0]
michael@0 57
michael@0 58 // TODO: Doug, you might want to revise this...
michael@0 59 result = *id;
michael@0 60 int32_t i = 0;
michael@0 61 int32_t end = result.indexOf(AT_SIGN_CHAR);
michael@0 62 int32_t n = result.indexOf(PERIOD_CHAR);
michael@0 63 if (n >= 0 && n < end) {
michael@0 64 end = n;
michael@0 65 }
michael@0 66 if (end < 0) {
michael@0 67 end = result.length();
michael@0 68 }
michael@0 69 n = result.indexOf(UNDERSCORE_CHAR);
michael@0 70 if (n < 0) {
michael@0 71 n = end;
michael@0 72 }
michael@0 73 for (; i < n; ++i) {
michael@0 74 UChar c = result.charAt(i);
michael@0 75 if (c >= 0x0041 && c <= 0x005a) {
michael@0 76 c += 0x20;
michael@0 77 result.setCharAt(i, c);
michael@0 78 }
michael@0 79 }
michael@0 80 for (n = end; i < n; ++i) {
michael@0 81 UChar c = result.charAt(i);
michael@0 82 if (c >= 0x0061 && c <= 0x007a) {
michael@0 83 c -= 0x20;
michael@0 84 result.setCharAt(i, c);
michael@0 85 }
michael@0 86 }
michael@0 87 }
michael@0 88 return result;
michael@0 89
michael@0 90 #if 0
michael@0 91 // This code does a proper full level 2 canonicalization of id.
michael@0 92 // It's nasty to go from UChar to char to char to UChar -- but
michael@0 93 // that's what you have to do to use the uloc_canonicalize
michael@0 94 // function on UnicodeStrings.
michael@0 95
michael@0 96 // I ended up doing the alternate fix (see above) not for
michael@0 97 // performance reasons, although performance will certainly be
michael@0 98 // better, but because doing a full level 2 canonicalization
michael@0 99 // causes some tests to fail. [alan 3.0]
michael@0 100
michael@0 101 // TODO: Doug, you might want to revisit this...
michael@0 102 result.setToBogus();
michael@0 103 if (id != 0) {
michael@0 104 int32_t buflen = id->length() + 8; // space for NUL
michael@0 105 char* buf = (char*) uprv_malloc(buflen);
michael@0 106 char* canon = (buf == 0) ? 0 : (char*) uprv_malloc(buflen);
michael@0 107 if (buf != 0 && canon != 0) {
michael@0 108 U_ASSERT(id->extract(0, INT32_MAX, buf, buflen) < buflen);
michael@0 109 UErrorCode ec = U_ZERO_ERROR;
michael@0 110 uloc_canonicalize(buf, canon, buflen, &ec);
michael@0 111 if (U_SUCCESS(ec)) {
michael@0 112 result = UnicodeString(canon);
michael@0 113 }
michael@0 114 }
michael@0 115 uprv_free(buf);
michael@0 116 uprv_free(canon);
michael@0 117 }
michael@0 118 return result;
michael@0 119 #endif
michael@0 120 }
michael@0 121
michael@0 122 Locale&
michael@0 123 LocaleUtility::initLocaleFromName(const UnicodeString& id, Locale& result)
michael@0 124 {
michael@0 125 enum { BUFLEN = 128 }; // larger than ever needed
michael@0 126
michael@0 127 if (id.isBogus() || id.length() >= BUFLEN) {
michael@0 128 result.setToBogus();
michael@0 129 } else {
michael@0 130 /*
michael@0 131 * We need to convert from a UnicodeString to char * in order to
michael@0 132 * create a Locale.
michael@0 133 *
michael@0 134 * Problem: Locale ID strings may contain '@' which is a variant
michael@0 135 * character and cannot be handled by invariant-character conversion.
michael@0 136 *
michael@0 137 * Hack: Since ICU code can handle locale IDs with multiple encodings
michael@0 138 * of '@' (at least for EBCDIC; it's not known to be a problem for
michael@0 139 * ASCII-based systems),
michael@0 140 * we use regular invariant-character conversion for everything else
michael@0 141 * and manually convert U+0040 into a compiler-char-constant '@'.
michael@0 142 * While this compilation-time constant may not match the runtime
michael@0 143 * encoding of '@', it should be one of the encodings which ICU
michael@0 144 * recognizes.
michael@0 145 *
michael@0 146 * There should be only at most one '@' in a locale ID.
michael@0 147 */
michael@0 148 char buffer[BUFLEN];
michael@0 149 int32_t prev, i;
michael@0 150 prev = 0;
michael@0 151 for(;;) {
michael@0 152 i = id.indexOf((UChar)0x40, prev);
michael@0 153 if(i < 0) {
michael@0 154 // no @ between prev and the rest of the string
michael@0 155 id.extract(prev, INT32_MAX, buffer + prev, BUFLEN - prev, US_INV);
michael@0 156 break; // done
michael@0 157 } else {
michael@0 158 // normal invariant-character conversion for text between @s
michael@0 159 id.extract(prev, i - prev, buffer + prev, BUFLEN - prev, US_INV);
michael@0 160 // manually "convert" U+0040 at id[i] into '@' at buffer[i]
michael@0 161 buffer[i] = '@';
michael@0 162 prev = i + 1;
michael@0 163 }
michael@0 164 }
michael@0 165 result = Locale::createFromName(buffer);
michael@0 166 }
michael@0 167 return result;
michael@0 168 }
michael@0 169
michael@0 170 UnicodeString&
michael@0 171 LocaleUtility::initNameFromLocale(const Locale& locale, UnicodeString& result)
michael@0 172 {
michael@0 173 if (locale.isBogus()) {
michael@0 174 result.setToBogus();
michael@0 175 } else {
michael@0 176 result.append(UnicodeString(locale.getName(), -1, US_INV));
michael@0 177 }
michael@0 178 return result;
michael@0 179 }
michael@0 180
michael@0 181 const Hashtable*
michael@0 182 LocaleUtility::getAvailableLocaleNames(const UnicodeString& bundleID)
michael@0 183 {
michael@0 184 // LocaleUtility_cache is a hash-of-hashes. The top-level keys
michael@0 185 // are path strings ('bundleID') passed to
michael@0 186 // ures_openAvailableLocales. The top-level values are
michael@0 187 // second-level hashes. The second-level keys are result strings
michael@0 188 // from ures_openAvailableLocales. The second-level values are
michael@0 189 // garbage ((void*)1 or other random pointer).
michael@0 190
michael@0 191 UErrorCode status = U_ZERO_ERROR;
michael@0 192 Hashtable* cache;
michael@0 193 umtx_lock(NULL);
michael@0 194 cache = LocaleUtility_cache;
michael@0 195 umtx_unlock(NULL);
michael@0 196
michael@0 197 if (cache == NULL) {
michael@0 198 cache = new Hashtable(status);
michael@0 199 if (cache == NULL || U_FAILURE(status)) {
michael@0 200 return NULL; // catastrophic failure; e.g. out of memory
michael@0 201 }
michael@0 202 cache->setValueDeleter(uhash_deleteHashtable);
michael@0 203 Hashtable* h; // set this to final LocaleUtility_cache value
michael@0 204 umtx_lock(NULL);
michael@0 205 h = LocaleUtility_cache;
michael@0 206 if (h == NULL) {
michael@0 207 LocaleUtility_cache = h = cache;
michael@0 208 cache = NULL;
michael@0 209 ucln_common_registerCleanup(UCLN_COMMON_SERVICE, service_cleanup);
michael@0 210 }
michael@0 211 umtx_unlock(NULL);
michael@0 212 if(cache != NULL) {
michael@0 213 delete cache;
michael@0 214 }
michael@0 215 cache = h;
michael@0 216 }
michael@0 217
michael@0 218 U_ASSERT(cache != NULL);
michael@0 219
michael@0 220 Hashtable* htp;
michael@0 221 umtx_lock(NULL);
michael@0 222 htp = (Hashtable*) cache->get(bundleID);
michael@0 223 umtx_unlock(NULL);
michael@0 224
michael@0 225 if (htp == NULL) {
michael@0 226 htp = new Hashtable(status);
michael@0 227 if (htp && U_SUCCESS(status)) {
michael@0 228 CharString cbundleID;
michael@0 229 cbundleID.appendInvariantChars(bundleID, status);
michael@0 230 const char* path = cbundleID.isEmpty() ? NULL : cbundleID.data();
michael@0 231 UEnumeration *uenum = ures_openAvailableLocales(path, &status);
michael@0 232 for (;;) {
michael@0 233 const UChar* id = uenum_unext(uenum, NULL, &status);
michael@0 234 if (id == NULL) {
michael@0 235 break;
michael@0 236 }
michael@0 237 htp->put(UnicodeString(id), (void*)htp, status);
michael@0 238 }
michael@0 239 uenum_close(uenum);
michael@0 240 if (U_FAILURE(status)) {
michael@0 241 delete htp;
michael@0 242 return NULL;
michael@0 243 }
michael@0 244 umtx_lock(NULL);
michael@0 245 cache->put(bundleID, (void*)htp, status);
michael@0 246 umtx_unlock(NULL);
michael@0 247 }
michael@0 248 }
michael@0 249 return htp;
michael@0 250 }
michael@0 251
michael@0 252 UBool
michael@0 253 LocaleUtility::isFallbackOf(const UnicodeString& root, const UnicodeString& child)
michael@0 254 {
michael@0 255 return child.indexOf(root) == 0 &&
michael@0 256 (child.length() == root.length() ||
michael@0 257 child.charAt(root.length()) == UNDERSCORE_CHAR);
michael@0 258 }
michael@0 259
michael@0 260 U_NAMESPACE_END
michael@0 261
michael@0 262 /* !UCONFIG_NO_SERVICE */
michael@0 263 #endif
michael@0 264
michael@0 265

mercurial