michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (C) 1997-2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *
michael@0: * File resbund.cpp
michael@0: *
michael@0: * Modification History:
michael@0: *
michael@0: *   Date        Name        Description
michael@0: *   02/05/97    aliu        Fixed bug in chopLocale.  Added scanForLocaleInFile
michael@0: *                           based on code taken from scanForLocale.  Added
michael@0: *                           constructor which attempts to read resource bundle
michael@0: *                           from a specific file, without searching other files.
michael@0: *   02/11/97    aliu        Added UErrorCode return values to constructors. Fixed
michael@0: *                           infinite loops in scanForFile and scanForLocale.
michael@0: *                           Modified getRawResourceData to not delete storage in
michael@0: *                           localeData and resourceData which it doesn't own.
michael@0: *                           Added Mac compatibility #ifdefs for tellp() and
michael@0: *                           ios::nocreate.
michael@0: *   03/04/97    aliu        Modified to use ExpandingDataSink objects instead of
michael@0: *                           the highly inefficient ostrstream objects.
michael@0: *   03/13/97    aliu        Rewrote to load in entire resource bundle and store
michael@0: *                           it as a Hashtable of ResourceBundleData objects.
michael@0: *                           Added state table to govern parsing of files.
michael@0: *                           Modified to load locale index out of new file distinct
michael@0: *                           from default.txt.
michael@0: *   03/25/97    aliu        Modified to support 2-d arrays, needed for timezone data.
michael@0: *                           Added support for custom file suffixes.  Again, needed
michael@0: *                           to support timezone data.  Improved error handling to
michael@0: *                           detect duplicate tags and subtags.
michael@0: *   04/07/97    aliu        Fixed bug in getHashtableForLocale().  Fixed handling
michael@0: *                           of failing UErrorCode values on entry to API methods.
michael@0: *                           Fixed bugs in getArrayItem() for negative indices.
michael@0: *   04/29/97    aliu        Update to use new Hashtable deletion protocol.
michael@0: *   05/06/97    aliu        Flattened kTransitionTable for HP compiler.
michael@0: *                           Fixed usage of CharString.
michael@0: * 06/11/99      stephen     Removed parsing of .txt files.
michael@0: *                           Reworked to use new binary format.
michael@0: *                           Cleaned up.
michael@0: * 06/14/99      stephen     Removed methods taking a filename suffix.
michael@0: * 06/22/99      stephen     Added missing T_FileStream_close in parse()
michael@0: * 11/09/99      weiv        Added getLocale(), rewritten constructForLocale()
michael@0: * March 2000    weiv        complete overhaul.
michael@0: ******************************************************************************
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/resbund.h"
michael@0: 
michael@0: #include "mutex.h"
michael@0: #include "uassert.h"
michael@0: #include "umutex.h"
michael@0: 
michael@0: #include "uresimp.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: /*-----------------------------------------------------------------------------
michael@0:  * Implementation Notes
michael@0:  *
michael@0:  * Resource bundles are read in once, and thereafter cached.
michael@0:  * ResourceBundle statically keeps track of which files have been
michael@0:  * read, so we are guaranteed that each file is read at most once.
michael@0:  * Resource bundles can be loaded from different data directories and
michael@0:  * will be treated as distinct, even if they are for the same locale.
michael@0:  *
michael@0:  * Resource bundles are lightweight objects, which have pointers to
michael@0:  * one or more shared Hashtable objects containing all the data.
michael@0:  * Copying would be cheap, but there is no copy constructor, since
michael@0:  * there wasn't one in the original API.
michael@0:  *
michael@0:  * The ResourceBundle parsing mechanism is implemented as a transition
michael@0:  * network, for easy maintenance and modification.  The network is
michael@0:  * implemented as a matrix (instead of in code) to make this even
michael@0:  * easier.  The matrix contains Transition objects.  Each Transition
michael@0:  * object describes a destination node and an action to take before
michael@0:  * moving to the destination node.  The source node is encoded by the
michael@0:  * index of the object in the array that contains it.  The pieces
michael@0:  * needed to understand the transition network are the enums for node
michael@0:  * IDs and actions, the parse() method, which walks through the
michael@0:  * network and implements the actions, and the network itself.  The
michael@0:  * network guarantees certain conditions, for example, that a new
michael@0:  * resource will not be closed until one has been opened first; or
michael@0:  * that data will not be stored into a TaggedList until a TaggedList
michael@0:  * has been created.  Nonetheless, the code in parse() does some
michael@0:  * consistency checks as it runs the network, and fails with an
michael@0:  * U_INTERNAL_PROGRAM_ERROR if one of these checks fails.  If the input
michael@0:  * data has a bad format, an U_INVALID_FORMAT_ERROR is returned.  If you
michael@0:  * see an U_INTERNAL_PROGRAM_ERROR the transition matrix has a bug in
michael@0:  * it.
michael@0:  *
michael@0:  * Old functionality of multiple locales in a single file is still
michael@0:  * supported.  For this reason, LOCALE names override FILE names.  If
michael@0:  * data for en_US is located in the en.txt file, once it is loaded,
michael@0:  * the code will not care where it came from (other than remembering
michael@0:  * which directory it came from).  However, if there is an en_US
michael@0:  * resource in en_US.txt, that will take precedence.  There is no
michael@0:  * limit to the number or type of resources that can be stored in a
michael@0:  * file, however, files are only searched in a specific way.  If
michael@0:  * en_US_CA is requested, then first en_US_CA.txt is searched, then
michael@0:  * en_US.txt, then en.txt, then default.txt.  So it only makes sense
michael@0:  * to put certain locales in certain files.  In this example, it would
michael@0:  * be logical to put en_US_CA, en_US, and en into the en.txt file,
michael@0:  * since they would be found there if asked for.  The extreme example
michael@0:  * is to place all locale resources into default.txt, which should
michael@0:  * also work.
michael@0:  *
michael@0:  * Inheritance is implemented.  For example, xx_YY_zz inherits as
michael@0:  * follows: xx_YY_zz, xx_YY, xx, default.  Inheritance is implemented
michael@0:  * as an array of hashtables.  There will be from 1 to 4 hashtables in
michael@0:  * the array.
michael@0:  *
michael@0:  * Fallback files are implemented.  The fallback pattern is Language
michael@0:  * Country Variant (LCV) -> LC -> L.  Fallback is first done for the
michael@0:  * requested locale.  Then it is done for the default locale, as
michael@0:  * returned by Locale::getDefault().  Then the special file
michael@0:  * default.txt is searched for the default locale.  The overall FILE
michael@0:  * fallback path is LCV -> LC -> L -> dLCV -> dLC -> dL -> default.
michael@0:  *
michael@0:  * Note that although file name searching includes the default locale,
michael@0:  * once a ResourceBundle object is constructed, the inheritance path
michael@0:  * no longer includes the default locale.  The path is LCV -> LC -> L
michael@0:  * -> default.
michael@0:  *
michael@0:  * File parsing is lazy.  Nothing is parsed unless it is called for by
michael@0:  * someone.  So when a ResourceBundle for xx_YY_zz is constructed,
michael@0:  * only that locale is parsed (along with anything else in the same
michael@0:  * file).  Later, if the FooBar tag is asked for, and if it isn't
michael@0:  * found in xx_YY_zz, then xx_YY.txt will be parsed and checked, and
michael@0:  * so forth, until the chain is exhausted or the tag is found.
michael@0:  *
michael@0:  * Thread-safety is implemented around caches, both the cache that
michael@0:  * stores all the resouce data, and the cache that stores flags
michael@0:  * indicating whether or not a file has been visited.  These caches
michael@0:  * delete their storage at static cleanup time, when the process
michael@0:  * quits.
michael@0:  *
michael@0:  * ResourceBundle supports TableCollation as a special case.  This
michael@0:  * involves having special ResourceBundle objects which DO own their
michael@0:  * data, since we don't want large collation rule strings in the
michael@0:  * ResourceBundle cache (these are already cached in the
michael@0:  * TableCollation cache).  TableCollation files (.ctx files) have the
michael@0:  * same format as normal resource data files, with a different
michael@0:  * interpretation, from the standpoint of ResourceBundle.  .ctx files
michael@0:  * are loaded into otherwise ordinary ResourceBundle objects.  They
michael@0:  * don't inherit (that's implemented by TableCollation) and they own
michael@0:  * their data (as mentioned above).  However, they still support
michael@0:  * possible multiple locales in a single .ctx file.  (This is in
michael@0:  * practice a bad idea, since you only want the one locale you're
michael@0:  * looking for, and only one tag will be present
michael@0:  * ("CollationElements"), so you don't need an inheritance chain of
michael@0:  * multiple locales.)  Up to 4 locale resources will be loaded from a
michael@0:  * .ctx file; everything after the first 4 is ignored (parsed and
michael@0:  * deleted).  (Normal .txt files have no limit.)  Instead of being
michael@0:  * loaded into the cache, and then looked up as needed, the locale
michael@0:  * resources are read straight into the ResourceBundle object.
michael@0:  *
michael@0:  * The Index, which used to reside in default.txt, has been moved to a
michael@0:  * new file, index.txt.  This file contains a slightly modified format
michael@0:  * with the addition of the "InstalledLocales" tag; it looks like:
michael@0:  *
michael@0:  * Index {
michael@0:  *   InstalledLocales {
michael@0:  *     ar
michael@0:  *     ..
michael@0:  *     zh_TW
michael@0:  *   }
michael@0:  * }
michael@0:  */
michael@0: //-----------------------------------------------------------------------------
michael@0: 
michael@0: UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle)
michael@0: 
michael@0: ResourceBundle::ResourceBundle(UErrorCode &err)
michael@0:                                 :UObject(), fLocale(NULL)
michael@0: {
michael@0:     fResource = ures_open(0, Locale::getDefault().getName(), &err);
michael@0: }
michael@0: 
michael@0: ResourceBundle::ResourceBundle(const ResourceBundle &other)
michael@0:                               :UObject(other), fLocale(NULL)
michael@0: {
michael@0:     UErrorCode status = U_ZERO_ERROR;
michael@0: 
michael@0:     if (other.fResource) {
michael@0:         fResource = ures_copyResb(0, other.fResource, &status);
michael@0:     } else {
michael@0:         /* Copying a bad resource bundle */
michael@0:         fResource = NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: ResourceBundle::ResourceBundle(UResourceBundle *res, UErrorCode& err)
michael@0:                                :UObject(), fLocale(NULL)
michael@0: {
michael@0:     if (res) {
michael@0:         fResource = ures_copyResb(0, res, &err);
michael@0:     } else {
michael@0:         /* Copying a bad resource bundle */
michael@0:         fResource = NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: ResourceBundle::ResourceBundle(const char* path, const Locale& locale, UErrorCode& err) 
michael@0:                                :UObject(), fLocale(NULL)
michael@0: {
michael@0:     fResource = ures_open(path, locale.getName(), &err);
michael@0: }
michael@0: 
michael@0: 
michael@0: ResourceBundle& ResourceBundle::operator=(const ResourceBundle& other)
michael@0: {
michael@0:     if(this == &other) {
michael@0:         return *this;
michael@0:     }
michael@0:     if(fResource != 0) {
michael@0:         ures_close(fResource);
michael@0:         fResource = NULL;
michael@0:     }
michael@0:     if (fLocale != NULL) {
michael@0:         delete fLocale;
michael@0:         fLocale = NULL;
michael@0:     }
michael@0:     UErrorCode status = U_ZERO_ERROR;
michael@0:     if (other.fResource) {
michael@0:         fResource = ures_copyResb(0, other.fResource, &status);
michael@0:     } else {
michael@0:         /* Copying a bad resource bundle */
michael@0:         fResource = NULL;
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: ResourceBundle::~ResourceBundle()
michael@0: {
michael@0:     if(fResource != 0) {
michael@0:         ures_close(fResource);
michael@0:     }
michael@0:     if(fLocale != NULL) {
michael@0:       delete(fLocale);
michael@0:     }
michael@0: }
michael@0: 
michael@0: ResourceBundle *
michael@0: ResourceBundle::clone() const {
michael@0:     return new ResourceBundle(*this);
michael@0: }
michael@0: 
michael@0: UnicodeString ResourceBundle::getString(UErrorCode& status) const {
michael@0:     int32_t len = 0;
michael@0:     const UChar *r = ures_getString(fResource, &len, &status);
michael@0:     return UnicodeString(TRUE, r, len);
michael@0: }
michael@0: 
michael@0: const uint8_t *ResourceBundle::getBinary(int32_t& len, UErrorCode& status) const {
michael@0:     return ures_getBinary(fResource, &len, &status);
michael@0: }
michael@0: 
michael@0: const int32_t *ResourceBundle::getIntVector(int32_t& len, UErrorCode& status) const {
michael@0:     return ures_getIntVector(fResource, &len, &status);
michael@0: }
michael@0: 
michael@0: uint32_t ResourceBundle::getUInt(UErrorCode& status) const {
michael@0:     return ures_getUInt(fResource, &status);
michael@0: }
michael@0: 
michael@0: int32_t ResourceBundle::getInt(UErrorCode& status) const {
michael@0:     return ures_getInt(fResource, &status);
michael@0: }
michael@0: 
michael@0: const char *ResourceBundle::getName(void) const {
michael@0:     return ures_getName(fResource);
michael@0: }
michael@0: 
michael@0: const char *ResourceBundle::getKey(void) const {
michael@0:     return ures_getKey(fResource);
michael@0: }
michael@0: 
michael@0: UResType ResourceBundle::getType(void) const {
michael@0:     return ures_getType(fResource);
michael@0: }
michael@0: 
michael@0: int32_t ResourceBundle::getSize(void) const {
michael@0:     return ures_getSize(fResource);
michael@0: }
michael@0: 
michael@0: UBool ResourceBundle::hasNext(void) const {
michael@0:     return ures_hasNext(fResource);
michael@0: }
michael@0: 
michael@0: void ResourceBundle::resetIterator(void) {
michael@0:     ures_resetIterator(fResource);
michael@0: }
michael@0: 
michael@0: ResourceBundle ResourceBundle::getNext(UErrorCode& status) {
michael@0:     UResourceBundle r;
michael@0: 
michael@0:     ures_initStackObject(&r);
michael@0:     ures_getNextResource(fResource, &r, &status);
michael@0:     ResourceBundle res(&r, status);
michael@0:     if (U_SUCCESS(status)) {
michael@0:         ures_close(&r);
michael@0:     }
michael@0:     return res;
michael@0: }
michael@0: 
michael@0: UnicodeString ResourceBundle::getNextString(UErrorCode& status) {
michael@0:     int32_t len = 0;
michael@0:     const UChar* r = ures_getNextString(fResource, &len, 0, &status);
michael@0:     return UnicodeString(TRUE, r, len);
michael@0: }
michael@0: 
michael@0: UnicodeString ResourceBundle::getNextString(const char ** key, UErrorCode& status) {
michael@0:     int32_t len = 0;
michael@0:     const UChar* r = ures_getNextString(fResource, &len, key, &status);
michael@0:     return UnicodeString(TRUE, r, len);
michael@0: }
michael@0: 
michael@0: ResourceBundle ResourceBundle::get(int32_t indexR, UErrorCode& status) const {
michael@0:     UResourceBundle r;
michael@0: 
michael@0:     ures_initStackObject(&r);
michael@0:     ures_getByIndex(fResource, indexR, &r, &status);
michael@0:     ResourceBundle res(&r, status);
michael@0:     if (U_SUCCESS(status)) {
michael@0:         ures_close(&r);
michael@0:     }
michael@0:     return res;
michael@0: }
michael@0: 
michael@0: UnicodeString ResourceBundle::getStringEx(int32_t indexS, UErrorCode& status) const {
michael@0:     int32_t len = 0;
michael@0:     const UChar* r = ures_getStringByIndex(fResource, indexS, &len, &status);
michael@0:     return UnicodeString(TRUE, r, len);
michael@0: }
michael@0: 
michael@0: ResourceBundle ResourceBundle::get(const char* key, UErrorCode& status) const {
michael@0:     UResourceBundle r;
michael@0: 
michael@0:     ures_initStackObject(&r);
michael@0:     ures_getByKey(fResource, key, &r, &status);
michael@0:     ResourceBundle res(&r, status);
michael@0:     if (U_SUCCESS(status)) {
michael@0:         ures_close(&r);
michael@0:     }
michael@0:     return res;
michael@0: }
michael@0: 
michael@0: ResourceBundle ResourceBundle::getWithFallback(const char* key, UErrorCode& status){
michael@0:     UResourceBundle r;
michael@0:     ures_initStackObject(&r);
michael@0:     ures_getByKeyWithFallback(fResource, key, &r, &status);
michael@0:     ResourceBundle res(&r, status);
michael@0:     if(U_SUCCESS(status)){
michael@0:         ures_close(&r);
michael@0:     }
michael@0:     return res;
michael@0: }
michael@0: UnicodeString ResourceBundle::getStringEx(const char* key, UErrorCode& status) const {
michael@0:     int32_t len = 0;
michael@0:     const UChar* r = ures_getStringByKey(fResource, key, &len, &status);
michael@0:     return UnicodeString(TRUE, r, len);
michael@0: }
michael@0: 
michael@0: const char*
michael@0: ResourceBundle::getVersionNumber()  const
michael@0: {
michael@0:     return ures_getVersionNumberInternal(fResource);
michael@0: }
michael@0: 
michael@0: void ResourceBundle::getVersion(UVersionInfo versionInfo) const {
michael@0:     ures_getVersion(fResource, versionInfo);
michael@0: }
michael@0: 
michael@0: static UMutex gLocaleLock = U_MUTEX_INITIALIZER;
michael@0: const Locale &ResourceBundle::getLocale(void) const {
michael@0:     Mutex lock(&gLocaleLock);
michael@0:     if (fLocale != NULL) {
michael@0:         return *fLocale;
michael@0:     }
michael@0:     UErrorCode status = U_ZERO_ERROR;
michael@0:     const char *localeName = ures_getLocaleInternal(fResource, &status);
michael@0:     ResourceBundle *ncThis = const_cast<ResourceBundle *>(this);
michael@0:     ncThis->fLocale = new Locale(localeName);
michael@0:     return ncThis->fLocale != NULL ? *ncThis->fLocale : Locale::getDefault();
michael@0: }
michael@0: 
michael@0: const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const
michael@0: {
michael@0:   return ures_getLocaleByType(fResource, type, &status);
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: //eof