X-Git-Url: https://git.saurik.com/apple/icu.git/blobdiff_plain/374ca955a76ecab1204ca8bfa63ff9238d998416..a01113dcd0f39d5da295ef82785beff9ed86fe38:/icuSources/i18n/transreg.cpp?ds=sidebyside diff --git a/icuSources/i18n/transreg.cpp b/icuSources/i18n/transreg.cpp index 781ea87a..5cc29625 100644 --- a/icuSources/i18n/transreg.cpp +++ b/icuSources/i18n/transreg.cpp @@ -1,6 +1,8 @@ +// © 2016 and later: Unicode, Inc. and others. +// License & terms of use: http://www.unicode.org/copyright.html /* ********************************************************************** -* Copyright (c) 2001-2004, International Business Machines +* Copyright (c) 2001-2014, International Business Machines * Corporation and others. All Rights Reserved. ********************************************************************** * Date Name Description @@ -12,11 +14,8 @@ #if !UCONFIG_NO_TRANSLITERATION -#include "unicode/utypes.h" -#include "unicode/uobject.h" -#include "unicode/parseerr.h" -#include "unicode/resbund.h" #include "unicode/translit.h" +#include "unicode/resbund.h" #include "unicode/uniset.h" #include "unicode/uscript.h" #include "rbt.h" @@ -28,6 +27,7 @@ #include "tridpars.h" #include "charstr.h" #include "uassert.h" +#include "locutil.h" // Enable the following symbol to add debugging code that tracks the // allocation, deletion, and use of Entry objects. BoundsChecker has @@ -42,17 +42,43 @@ // UChar constants static const UChar LOCALE_SEP = 95; // '_' -static const UChar ID_SEP = 0x002D; /*-*/ -static const UChar VARIANT_SEP = 0x002F; // '/' +//static const UChar ID_SEP = 0x002D; /*-*/ +//static const UChar VARIANT_SEP = 0x002F; // '/' // String constants -static const UChar NO_VARIANT[] = { 0 }; // empty string -static const UChar ANY[] = { 65, 110, 121, 0 }; // Any +static const UChar ANY[] = { 0x41, 0x6E, 0x79, 0 }; // Any +static const UChar LAT[] = { 0x4C, 0x61, 0x74, 0 }; // Lat + +// empty string +#define NO_VARIANT UnicodeString() + +// initial estimate for specDAG size +// ICU 60 Transliterator::countAvailableSources() +//#define SPECDAG_INIT_SIZE 149 +// Apple adjustment +#define SPECDAG_INIT_SIZE 134 + +// initial estimate for number of variant names +#define VARIANT_LIST_INIT_SIZE 11 +#define VARIANT_LIST_MAX_SIZE 31 + +// initial estimate for availableIDs count (default estimate is 8 => multiple reallocs) +// ICU 60 Transliterator::countAvailableIDs() +//#define AVAILABLE_IDS_INIT_SIZE 641 +// Apple adjustment +#define AVAILABLE_IDS_INIT_SIZE 493 + +// initial estimate for number of targets for source "Any", "Lat" +// ICU 60 Transliterator::countAvailableTargets("Any")/("Latn") +//#define ANY_TARGETS_INIT_SIZE 125 +// Apple adjustmennt +#define ANY_TARGETS_INIT_SIZE 102 +#define LAT_TARGETS_INIT_SIZE 23 /** * Resource bundle key for the RuleBasedTransliterator rule. */ -static const char RB_RULE[] = "Rule"; +//static const char RB_RULE[] = "Rule"; U_NAMESPACE_BEGIN @@ -60,25 +86,25 @@ U_NAMESPACE_BEGIN // Alias //------------------------------------------------------------------ -TransliteratorAlias::TransliteratorAlias(const UnicodeString& theAliasID) : +TransliteratorAlias::TransliteratorAlias(const UnicodeString& theAliasID, + const UnicodeSet* cpdFilter) : ID(), - aliasID(theAliasID), - trans(0), - compoundFilter(0), - idSplitPoint(-1), + aliasesOrRules(theAliasID), + transes(0), + compoundFilter(cpdFilter), + direction(UTRANS_FORWARD), type(TransliteratorAlias::SIMPLE) { } TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, - const UnicodeString& idBlock, - Transliterator* adopted, - int32_t theIDSplitPoint, + const UnicodeString& idBlocks, + UVector* adoptedTransliterators, const UnicodeSet* cpdFilter) : ID(theID), - aliasID(idBlock), - trans(adopted), + aliasesOrRules(idBlocks), + transes(adoptedTransliterators), compoundFilter(cpdFilter), - idSplitPoint(theIDSplitPoint), + direction(UTRANS_FORWARD), type(TransliteratorAlias::COMPOUND) { } @@ -86,15 +112,15 @@ TransliteratorAlias::TransliteratorAlias(const UnicodeString& theID, const UnicodeString& rules, UTransDirection dir) : ID(theID), - aliasID(rules), // bad name -- rename aliasID! - trans(0), + aliasesOrRules(rules), + transes(0), compoundFilter(0), - idSplitPoint((int32_t) dir), // bad name -- rename idSplitPoint! + direction(dir), type(TransliteratorAlias::RULES) { } TransliteratorAlias::~TransliteratorAlias() { - delete trans; + delete transes; } @@ -106,28 +132,67 @@ Transliterator* TransliteratorAlias::create(UParseError& pe, Transliterator *t = NULL; switch (type) { case SIMPLE: - t = Transliterator::createInstance(aliasID, UTRANS_FORWARD, pe, ec); - break; - case COMPOUND: - t = new CompoundTransliterator(ID, aliasID, idSplitPoint, - trans, ec); - /* test for NULL */ - if (t == 0) { - ec = U_MEMORY_ALLOCATION_ERROR; + t = Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec); + if(U_FAILURE(ec)){ return 0; } - trans = 0; // so we don't delete it later - if (compoundFilter) { - // TODO: Is this right? Are we leaking memory here? - // I'm suspicious because of the "trans = 0" line above; - // doesn't seem to fit the cloning here. Don't have time - // to track this down right now. [alan 3.0] - t->adoptFilter((UnicodeSet*) compoundFilter->clone()); + if (compoundFilter != 0) + t->adoptFilter((UnicodeSet*)compoundFilter->clone()); + break; + case COMPOUND: + { + // the total number of transliterators in the compound is the total number of anonymous transliterators + // plus the total number of ID blocks-- we start by assuming the list begins and ends with an ID + // block and that each pair anonymous transliterators has an ID block between them. Then we go back + // to see whether there really are ID blocks at the beginning and end (by looking for U+FFFF, which + // marks the position where an anonymous transliterator goes) and adjust accordingly + int32_t anonymousRBTs = transes->size(); + int32_t transCount = anonymousRBTs * 2 + 1; + if (!aliasesOrRules.isEmpty() && aliasesOrRules[0] == (UChar)(0xffff)) + --transCount; + if (aliasesOrRules.length() >= 2 && aliasesOrRules[aliasesOrRules.length() - 1] == (UChar)(0xffff)) + --transCount; + UnicodeString noIDBlock((UChar)(0xffff)); + noIDBlock += ((UChar)(0xffff)); + int32_t pos = aliasesOrRules.indexOf(noIDBlock); + while (pos >= 0) { + --transCount; + pos = aliasesOrRules.indexOf(noIDBlock, pos + 1); + } + + UVector transliterators(ec); + UnicodeString idBlock; + int32_t blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); + while (blockSeparatorPos >= 0) { + aliasesOrRules.extract(0, blockSeparatorPos, idBlock); + aliasesOrRules.remove(0, blockSeparatorPos + 1); + if (!idBlock.isEmpty()) + transliterators.addElement(Transliterator::createInstance(idBlock, UTRANS_FORWARD, pe, ec), ec); + if (!transes->isEmpty()) + transliterators.addElement(transes->orphanElementAt(0), ec); + blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff)); + } + if (!aliasesOrRules.isEmpty()) + transliterators.addElement(Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec), ec); + while (!transes->isEmpty()) + transliterators.addElement(transes->orphanElementAt(0), ec); + + if (U_SUCCESS(ec)) { + t = new CompoundTransliterator(ID, transliterators, + (compoundFilter ? (UnicodeSet*)(compoundFilter->clone()) : 0), + anonymousRBTs, pe, ec); + if (t == 0) { + ec = U_MEMORY_ALLOCATION_ERROR; + return 0; + } + } else { + for (int32_t i = 0; i < transliterators.size(); i++) + delete (Transliterator*)(transliterators.elementAt(i)); + } } break; case RULES: - U_ASSERT(FALSE); // don't call create() if isRuleBased() returns TRUE! - break; + UPRV_UNREACHABLE; // don't call create() if isRuleBased() returns TRUE! } return t; } @@ -143,17 +208,15 @@ void TransliteratorAlias::parse(TransliteratorParser& parser, return; } - // aliasID is really rules -- rename it! - // idSplitPoint is really UTransDirection -- rename it! - parser.parse(aliasID, (UTransDirection) idSplitPoint, pe, ec); + parser.parse(aliasesOrRules, direction, pe, ec); } //---------------------------------------------------------------------- -// class Spec +// class TransliteratorSpec //---------------------------------------------------------------------- /** - * A Spec is a string specifying either a source or a target. In more + * A TransliteratorSpec is a string specifying either a source or a target. In more * general terms, it may also specify a variant, but we only use the * Spec class for sources and targets. * @@ -167,10 +230,10 @@ void TransliteratorAlias::parse(TransliteratorParser& parser, * canonical form, or the script is transformed from an abbreviation * to a full name. */ -class Spec : public UMemory { +class TransliteratorSpec : public UMemory { public: - Spec(const UnicodeString& spec); - ~Spec(); + TransliteratorSpec(const UnicodeString& spec); + ~TransliteratorSpec(); const UnicodeString& get() const; UBool hasFallback() const; @@ -194,41 +257,46 @@ class Spec : public UMemory { UBool isNextLocale; // TRUE if nextSpec is a locale ResourceBundle* res; - Spec(const Spec &other); // forbid copying of this class - Spec &operator=(const Spec &other); // forbid copying of this class + TransliteratorSpec(const TransliteratorSpec &other); // forbid copying of this class + TransliteratorSpec &operator=(const TransliteratorSpec &other); // forbid copying of this class }; -Spec::Spec(const UnicodeString& theSpec) : top(theSpec) { +TransliteratorSpec::TransliteratorSpec(const UnicodeString& theSpec) +: top(theSpec), + res(0) +{ UErrorCode status = U_ZERO_ERROR; - CharString topch(top); - Locale toploc(topch); - res = new ResourceBundle(U_ICUDATA_TRANSLIT, toploc, status); - /* test for NULL */ - if (res == 0) { - return; - } - if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { - delete res; - res = 0; + Locale topLoc(""); + LocaleUtility::initLocaleFromName(theSpec, topLoc); + if (!topLoc.isBogus()) { + res = new ResourceBundle(U_ICUDATA_TRANSLIT, topLoc, status); + /* test for NULL */ + if (res == 0) { + return; + } + if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { + delete res; + res = 0; + } } // Canonicalize script name -or- do locale->script mapping status = U_ZERO_ERROR; - const int32_t capacity = 10; + static const int32_t capacity = 10; UScriptCode script[capacity]={USCRIPT_INVALID_CODE}; - int32_t num = uscript_getCode(topch,script,capacity, &status); + int32_t num = uscript_getCode(CharString().appendInvariantChars(theSpec, status).data(), + script, capacity, &status); if (num > 0 && script[0] != USCRIPT_INVALID_CODE) { - scriptName = UnicodeString(uscript_getName(script[0]), ""); + scriptName = UnicodeString(uscript_getName(script[0]), -1, US_INV); } // Canonicalize top - char buf[256]; if (res != 0) { // Canonicalize locale name - status = U_ZERO_ERROR; - uloc_getName(topch, buf, sizeof(buf), &status); - if (U_SUCCESS(status) && status != U_STRING_NOT_TERMINATED_WARNING) { - top = UnicodeString(buf, ""); + UnicodeString locStr; + LocaleUtility::initNameFromLocale(topLoc, locStr); + if (!locStr.isBogus()) { + top = locStr; } } else if (scriptName.length() != 0) { // We are a script; use canonical name @@ -239,15 +307,15 @@ Spec::Spec(const UnicodeString& theSpec) : top(theSpec) { reset(); } -Spec::~Spec() { +TransliteratorSpec::~TransliteratorSpec() { delete res; } -UBool Spec::hasFallback() const { +UBool TransliteratorSpec::hasFallback() const { return nextSpec.length() != 0; } -void Spec::reset() { +void TransliteratorSpec::reset() { if (spec != top) { spec = top; isSpecLocale = (res != 0); @@ -255,7 +323,7 @@ void Spec::reset() { } } -void Spec::setupNext() { +void TransliteratorSpec::setupNext() { isNextLocale = FALSE; if (isSpecLocale) { nextSpec = spec; @@ -278,22 +346,22 @@ void Spec::setupNext() { // for(const UnicodeString& s(spec.get()); // spec.hasFallback(); s(spec.next())) { ... -const UnicodeString& Spec::next() { +const UnicodeString& TransliteratorSpec::next() { spec = nextSpec; isSpecLocale = isNextLocale; setupNext(); return spec; } -const UnicodeString& Spec::get() const { +const UnicodeString& TransliteratorSpec::get() const { return spec; } -UBool Spec::isLocale() const { +UBool TransliteratorSpec::isLocale() const { return isSpecLocale; } -ResourceBundle& Spec::getBundle() const { +ResourceBundle& TransliteratorSpec::getBundle() const { return *res; } @@ -313,9 +381,9 @@ static void DEBUG_setup() { // Caller must call DEBUG_setup first. Return index of given Entry, // if it is in use (not deleted yet), or -1 if not found. -static int DEBUG_findEntry(Entry* e) { +static int DEBUG_findEntry(TransliteratorEntry* e) { for (int i=0; isize(); ++i) { - if (e == (Entry*) DEBUG_entries->elementAt(i)) { + if (e == (TransliteratorEntry*) DEBUG_entries->elementAt(i)) { return i; } } @@ -323,7 +391,7 @@ static int DEBUG_findEntry(Entry* e) { } // Track object creation -static void DEBUG_newEntry(Entry* e) { +static void DEBUG_newEntry(TransliteratorEntry* e) { DEBUG_setup(); if (DEBUG_findEntry(e) >= 0) { // This should really never happen unless the heap is broken @@ -335,7 +403,7 @@ static void DEBUG_newEntry(Entry* e) { } // Track object deletion -static void DEBUG_delEntry(Entry* e) { +static void DEBUG_delEntry(TransliteratorEntry* e) { DEBUG_setup(); int i = DEBUG_findEntry(e); if (i < 0) { @@ -346,7 +414,7 @@ static void DEBUG_delEntry(Entry* e) { } // Track object usage -static void DEBUG_useEntry(Entry* e) { +static void DEBUG_useEntry(TransliteratorEntry* e) { if (e == NULL) return; DEBUG_setup(); int i = DEBUG_findEntry(e); @@ -376,7 +444,7 @@ static void DEBUG_useEntry(Entry* e) { * for it. We could easily add this if there is a need for it in the * future. */ -class Entry : public UMemory { +class TransliteratorEntry : public UMemory { public: enum Type { RULES_FORWARD, @@ -396,46 +464,51 @@ public: UnicodeSet* compoundFilter; // For COMPOUND_RBT union { Transliterator* prototype; // For PROTOTYPE - TransliterationRuleData* data; // For RBT_DATA, COMPOUND_RBT + TransliterationRuleData* data; // For RBT_DATA + UVector* dataVector; // For COMPOUND_RBT struct { Transliterator::Factory function; Transliterator::Token context; } factory; // For FACTORY } u; - Entry(); - ~Entry(); + TransliteratorEntry(); + ~TransliteratorEntry(); void adoptPrototype(Transliterator* adopted); void setFactory(Transliterator::Factory factory, Transliterator::Token context); private: - Entry(const Entry &other); // forbid copying of this class - Entry &operator=(const Entry &other); // forbid copying of this class + TransliteratorEntry(const TransliteratorEntry &other); // forbid copying of this class + TransliteratorEntry &operator=(const TransliteratorEntry &other); // forbid copying of this class }; -Entry::Entry() { +TransliteratorEntry::TransliteratorEntry() { u.prototype = 0; compoundFilter = NULL; entryType = NONE; DEBUG_newEntry(this); } -Entry::~Entry() { +TransliteratorEntry::~TransliteratorEntry() { DEBUG_delEntry(this); if (entryType == PROTOTYPE) { delete u.prototype; - } else if (entryType == RBT_DATA || entryType == COMPOUND_RBT) { + } else if (entryType == RBT_DATA) { // The data object is shared between instances of RBT. The // entry object owns it. It should only be deleted when the // transliterator component is being cleaned up. Doing so // invalidates any RBTs that the user has instantiated. delete u.data; + } else if (entryType == COMPOUND_RBT) { + while (u.dataVector != NULL && !u.dataVector->isEmpty()) + delete (TransliterationRuleData*)u.dataVector->orphanElementAt(0); + delete u.dataVector; } delete compoundFilter; } -void Entry::adoptPrototype(Transliterator* adopted) { +void TransliteratorEntry::adoptPrototype(Transliterator* adopted) { if (entryType == PROTOTYPE) { delete u.prototype; } @@ -443,7 +516,7 @@ void Entry::adoptPrototype(Transliterator* adopted) { u.prototype = adopted; } -void Entry::setFactory(Transliterator::Factory factory, +void TransliteratorEntry::setFactory(Transliterator::Factory factory, Transliterator::Token context) { if (entryType == PROTOTYPE) { delete u.prototype; @@ -455,9 +528,9 @@ void Entry::setFactory(Transliterator::Factory factory, // UObjectDeleter for Hashtable::setValueDeleter U_CDECL_BEGIN -static void U_EXPORT2 U_CALLCONV +static void U_CALLCONV deleteEntry(void* obj) { - delete (Entry*) obj; + delete (TransliteratorEntry*) obj; } U_CDECL_END @@ -467,11 +540,18 @@ U_CDECL_END TransliteratorRegistry::TransliteratorRegistry(UErrorCode& status) : registry(TRUE, status), - specDAG(TRUE, status), - availableIDs(status) + specDAG(TRUE, SPECDAG_INIT_SIZE, status), + variantList(VARIANT_LIST_INIT_SIZE, status), + availableIDs(AVAILABLE_IDS_INIT_SIZE, status) { registry.setValueDeleter(deleteEntry); - availableIDs.setDeleter(uhash_deleteUnicodeString); + variantList.setDeleter(uprv_deleteUObject); + variantList.setComparer(uhash_compareCaselessUnicodeString); + UnicodeString *emptyString = new UnicodeString(); + if (emptyString != NULL) { + variantList.addElement(emptyString, status); + } + availableIDs.setDeleter(uprv_deleteUObject); availableIDs.setComparer(uhash_compareCaselessUnicodeString); specDAG.setValueDeleter(uhash_deleteHashtable); } @@ -484,7 +564,7 @@ Transliterator* TransliteratorRegistry::get(const UnicodeString& ID, TransliteratorAlias*& aliasReturn, UErrorCode& status) { U_ASSERT(aliasReturn == NULL); - Entry *entry = find(ID); + TransliteratorEntry *entry = find(ID); return (entry == 0) ? 0 : instantiateEntry(ID, entry, aliasReturn, status); } @@ -494,7 +574,7 @@ Transliterator* TransliteratorRegistry::reget(const UnicodeString& ID, TransliteratorAlias*& aliasReturn, UErrorCode& status) { U_ASSERT(aliasReturn == NULL); - Entry *entry = find(ID); + TransliteratorEntry *entry = find(ID); if (entry == 0) { // We get to this point if there are two threads, one of which @@ -515,43 +595,45 @@ Transliterator* TransliteratorRegistry::reget(const UnicodeString& ID, // We have to detect this so we don't stomp over existing entry // data members and potentially leak memory (u.data and compoundFilter). - if (entry->entryType == Entry::RULES_FORWARD || - entry->entryType == Entry::RULES_REVERSE || - entry->entryType == Entry::LOCALE_RULES) { + if (entry->entryType == TransliteratorEntry::RULES_FORWARD || + entry->entryType == TransliteratorEntry::RULES_REVERSE || + entry->entryType == TransliteratorEntry::LOCALE_RULES) { - entry->u.data = parser.orphanData(); - entry->stringArg = parser.idBlock; - entry->intArg = parser.idSplitPoint; - entry->compoundFilter = parser.orphanCompoundFilter(); - - // Reset entry->entryType to encapsulate the parsed data. The - // next time we instantiate this ID (including this very next - // time, at the end of this function) we won't have to parse - // again. - // NOTE: The logic here matches that in - // Transliterator::createFromRules(). - if (entry->stringArg.length() == 0) { - if (entry->u.data == 0) { - // No idBlock, no data -- this is just an - // alias for Null - entry->entryType = Entry::ALIAS; - entry->stringArg = NullTransliterator::ID; - } else { - // No idBlock, data != 0 -- this is an - // ordinary RBT_DATA - entry->entryType = Entry::RBT_DATA; - } - } else { - if (entry->u.data == 0) { - // idBlock, no data -- this is an alias. The ID has - // been munged from reverse into forward mode, if - // necessary, so instantiate the ID in the forward - // direction. - entry->entryType = Entry::ALIAS; - } else { - // idBlock and data -- this is a compound - // RBT - entry->entryType = Entry::COMPOUND_RBT; + if (parser.idBlockVector.isEmpty() && parser.dataVector.isEmpty()) { + entry->u.data = 0; + entry->entryType = TransliteratorEntry::ALIAS; + entry->stringArg = UNICODE_STRING_SIMPLE("Any-NULL"); + } + else if (parser.idBlockVector.isEmpty() && parser.dataVector.size() == 1) { + entry->u.data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); + entry->entryType = TransliteratorEntry::RBT_DATA; + } + else if (parser.idBlockVector.size() == 1 && parser.dataVector.isEmpty()) { + entry->stringArg = *(UnicodeString*)(parser.idBlockVector.elementAt(0)); + entry->compoundFilter = parser.orphanCompoundFilter(); + entry->entryType = TransliteratorEntry::ALIAS; + } + else { + entry->entryType = TransliteratorEntry::COMPOUND_RBT; + entry->compoundFilter = parser.orphanCompoundFilter(); + entry->u.dataVector = new UVector(status); + entry->stringArg.remove(); + + int32_t limit = parser.idBlockVector.size(); + if (parser.dataVector.size() > limit) + limit = parser.dataVector.size(); + + for (int32_t i = 0; i < limit; i++) { + if (i < parser.idBlockVector.size()) { + UnicodeString* idBlock = (UnicodeString*)parser.idBlockVector.elementAt(i); + if (!idBlock->isEmpty()) + entry->stringArg += *idBlock; + } + if (!parser.dataVector.isEmpty()) { + TransliterationRuleData* data = (TransliterationRuleData*)parser.dataVector.orphanElementAt(0); + entry->u.dataVector->addElement(data, status); + entry->stringArg += (UChar)0xffff; // use U+FFFF to mark position of RBTs in ID block + } } } } @@ -562,8 +644,14 @@ Transliterator* TransliteratorRegistry::reget(const UnicodeString& ID, } void TransliteratorRegistry::put(Transliterator* adoptedProto, - UBool visible) { - Entry *entry = new Entry(); + UBool visible, + UErrorCode& ec) +{ + TransliteratorEntry *entry = new TransliteratorEntry(); + if (entry == NULL) { + ec = U_MEMORY_ALLOCATION_ERROR; + return; + } entry->adoptPrototype(adoptedProto); registerEntry(adoptedProto->getID(), entry, visible); } @@ -571,8 +659,13 @@ void TransliteratorRegistry::put(Transliterator* adoptedProto, void TransliteratorRegistry::put(const UnicodeString& ID, Transliterator::Factory factory, Transliterator::Token context, - UBool visible) { - Entry *entry = new Entry(); + UBool visible, + UErrorCode& ec) { + TransliteratorEntry *entry = new TransliteratorEntry(); + if (entry == NULL) { + ec = U_MEMORY_ALLOCATION_ERROR; + return; + } entry->setFactory(factory, context); registerEntry(ID, entry, visible); } @@ -580,21 +673,42 @@ void TransliteratorRegistry::put(const UnicodeString& ID, void TransliteratorRegistry::put(const UnicodeString& ID, const UnicodeString& resourceName, UTransDirection dir, - UBool visible) { - Entry *entry = new Entry(); - entry->entryType = (dir == UTRANS_FORWARD) ? Entry::RULES_FORWARD - : Entry::RULES_REVERSE; - entry->stringArg = resourceName; + UBool readonlyResourceAlias, + UBool visible, + UErrorCode& ec) { + TransliteratorEntry *entry = new TransliteratorEntry(); + if (entry == NULL) { + ec = U_MEMORY_ALLOCATION_ERROR; + return; + } + entry->entryType = (dir == UTRANS_FORWARD) ? TransliteratorEntry::RULES_FORWARD + : TransliteratorEntry::RULES_REVERSE; + if (readonlyResourceAlias) { + entry->stringArg.setTo(TRUE, resourceName.getBuffer(), -1); + } + else { + entry->stringArg = resourceName; + } registerEntry(ID, entry, visible); } void TransliteratorRegistry::put(const UnicodeString& ID, const UnicodeString& alias, - UBool visible) { - Entry *entry = new Entry(); - entry->entryType = Entry::ALIAS; - entry->stringArg = alias; - registerEntry(ID, entry, visible); + UBool readonlyAliasAlias, + UBool visible, + UErrorCode& /*ec*/) { + TransliteratorEntry *entry = new TransliteratorEntry(); + // Null pointer check + if (entry != NULL) { + entry->entryType = TransliteratorEntry::ALIAS; + if (readonlyAliasAlias) { + entry->stringArg.setTo(TRUE, alias.getBuffer(), -1); + } + else { + entry->stringArg = alias; + } + registerEntry(ID, entry, visible); + } } void TransliteratorRegistry::remove(const UnicodeString& ID) { @@ -646,7 +760,7 @@ int32_t TransliteratorRegistry::countAvailableSources(void) const { UnicodeString& TransliteratorRegistry::getAvailableSource(int32_t index, UnicodeString& result) const { - int32_t pos = -1; + int32_t pos = UHASH_FIRST; const UHashElement *e = 0; while (index-- >= 0) { e = specDAG.nextElement(pos); @@ -675,7 +789,7 @@ UnicodeString& TransliteratorRegistry::getAvailableTarget(int32_t index, result.truncate(0); // invalid source return result; } - int32_t pos = -1; + int32_t pos = UHASH_FIRST; const UHashElement *e = 0; while (index-- >= 0) { e = targets->nextElement(pos); @@ -697,9 +811,15 @@ int32_t TransliteratorRegistry::countAvailableVariants(const UnicodeString& sour if (targets == 0) { return 0; } - UVector *variants = (UVector*) targets->get(target); - // variants may be 0 if the source/target are invalid - return (variants == 0) ? 0 : variants->size(); + uint32_t varMask = targets->geti(target); + int32_t varCount = 0; + while (varMask > 0) { + if (varMask & 1) { + varCount++; + } + varMask >>= 1; + } + return varCount; } UnicodeString& TransliteratorRegistry::getAvailableVariant(int32_t index, @@ -711,17 +831,25 @@ UnicodeString& TransliteratorRegistry::getAvailableVariant(int32_t index, result.truncate(0); // invalid source return result; } - UVector *variants = (UVector*) targets->get(target); - if (variants == 0) { - result.truncate(0); // invalid target - return result; - } - UnicodeString *v = (UnicodeString*) variants->elementAt(index); - if (v == 0) { - result.truncate(0); // invalid index - } else { - result = *v; + uint32_t varMask = targets->geti(target); + int32_t varCount = 0; + int32_t varListIndex = 0; + while (varMask > 0) { + if (varMask & 1) { + if (varCount == index) { + UnicodeString *v = (UnicodeString*) variantList.elementAt(varListIndex); + if (v != NULL) { + result = *v; + return result; + } + break; + } + varCount++; + } + varMask >>= 1; + varListIndex++; } + result.truncate(0); // invalid target or index return result; } @@ -783,12 +911,12 @@ UOBJECT_DEFINE_RTTI_IMPLEMENTATION(TransliteratorRegistry::Enumeration) void TransliteratorRegistry::registerEntry(const UnicodeString& source, const UnicodeString& target, const UnicodeString& variant, - Entry* adopted, + TransliteratorEntry* adopted, UBool visible) { UnicodeString ID; UnicodeString s(source); if (s.length() == 0) { - s = ANY; + s.setTo(TRUE, ANY, 3); } TransliteratorIDParser::STVtoID(source, target, variant, ID); registerEntry(ID, s, target, variant, adopted, visible); @@ -798,7 +926,7 @@ void TransliteratorRegistry::registerEntry(const UnicodeString& source, * Convenience method. Calls 6-arg registerEntry(). */ void TransliteratorRegistry::registerEntry(const UnicodeString& ID, - Entry* adopted, + TransliteratorEntry* adopted, UBool visible) { UnicodeString source, target, variant; UBool sawSource; @@ -817,7 +945,7 @@ void TransliteratorRegistry::registerEntry(const UnicodeString& ID, const UnicodeString& source, const UnicodeString& target, const UnicodeString& variant, - Entry* adopted, + TransliteratorEntry* adopted, UBool visible) { UErrorCode status = U_ZERO_ERROR; registry.put(ID, adopted, status); @@ -825,9 +953,12 @@ void TransliteratorRegistry::registerEntry(const UnicodeString& ID, registerSTV(source, target, variant); if (!availableIDs.contains((void*) &ID)) { UnicodeString *newID = (UnicodeString *)ID.clone(); - // NUL-terminate the ID string - newID->getTerminatedBuffer(); - availableIDs.addElement(newID, status); + // Check to make sure newID was created. + if (newID != NULL) { + // NUL-terminate the ID string + newID->getTerminatedBuffer(); + availableIDs.addElement(newID, status); + } } } else { removeSTV(source, target, variant); @@ -837,9 +968,7 @@ void TransliteratorRegistry::registerEntry(const UnicodeString& ID, /** * Register a source-target/variant in the specDAG. Variant may be - * empty, but source and target must not be. If variant is empty then - * the special variant NO_VARIANT is stored in slot zero of the - * UVector of variants. + * empty, but source and target must not be. */ void TransliteratorRegistry::registerSTV(const UnicodeString& source, const UnicodeString& target, @@ -849,32 +978,38 @@ void TransliteratorRegistry::registerSTV(const UnicodeString& source, UErrorCode status = U_ZERO_ERROR; Hashtable *targets = (Hashtable*) specDAG.get(source); if (targets == 0) { - targets = new Hashtable(TRUE, status); - if (U_FAILURE(status) || targets == 0) { + int32_t size = 3; + if (source.compare(ANY,3) == 0) { + size = ANY_TARGETS_INIT_SIZE; + } else if (source.compare(LAT,3) == 0) { + size = LAT_TARGETS_INIT_SIZE; + } + targets = new Hashtable(TRUE, size, status); + if (U_FAILURE(status) || targets == NULL) { return; } - targets->setValueDeleter(uhash_deleteUVector); specDAG.put(source, targets, status); } - UVector *variants = (UVector*) targets->get(target); - if (variants == 0) { - variants = new UVector(uhash_deleteUnicodeString, - uhash_compareCaselessUnicodeString, status); - if (variants == 0) { + int32_t variantListIndex = variantList.indexOf((void*) &variant, 0); + if (variantListIndex < 0) { + if (variantList.size() >= VARIANT_LIST_MAX_SIZE) { + // can't handle any more variants return; } - targets->put(target, variants, status); - } - // assert(NO_VARIANT == ""); - // We add the variant string. If it is the special "no variant" - // string, that is, the empty string, we add it at position zero. - if (!variants->contains((void*) &variant)) { - if (variant.length() > 0) { - variants->addElement(new UnicodeString(variant), status); - } else { - variants->insertElementAt(new UnicodeString(NO_VARIANT), 0, status); + UnicodeString *variantEntry = new UnicodeString(variant); + if (variantEntry != NULL) { + variantList.addElement(variantEntry, status); + if (U_SUCCESS(status)) { + variantListIndex = variantList.size() - 1; + } + } + if (variantListIndex < 0) { + return; } } + uint32_t addMask = 1 << variantListIndex; + uint32_t varMask = targets->geti(target); + targets->puti(target, varMask | addMask, status); } /** @@ -885,17 +1020,24 @@ void TransliteratorRegistry::removeSTV(const UnicodeString& source, const UnicodeString& variant) { // assert(source.length() > 0); // assert(target.length() > 0); -// UErrorCode status = U_ZERO_ERROR; + UErrorCode status = U_ZERO_ERROR; Hashtable *targets = (Hashtable*) specDAG.get(source); - if (targets == 0) { + if (targets == NULL) { + return; // should never happen for valid s-t/v + } + uint32_t varMask = targets->geti(target); + if (varMask == 0) { return; // should never happen for valid s-t/v } - UVector *variants = (UVector*) targets->get(target); - if (variants == 0) { + int32_t variantListIndex = variantList.indexOf((void*) &variant, 0); + if (variantListIndex < 0) { return; // should never happen for valid s-t/v } - variants->removeElement((void*) &variant); - if (variants->size() == 0) { + int32_t remMask = 1 << variantListIndex; + varMask &= (~remMask); + if (varMask != 0) { + targets->puti(target, varMask, status); + } else { targets->remove(target); // should delete variants if (targets->count() == 0) { specDAG.remove(source); // should delete targets @@ -909,12 +1051,12 @@ void TransliteratorRegistry::removeSTV(const UnicodeString& source, * * Caller does NOT own returned object. */ -Entry* TransliteratorRegistry::findInDynamicStore(const Spec& src, - const Spec& trg, +TransliteratorEntry* TransliteratorRegistry::findInDynamicStore(const TransliteratorSpec& src, + const TransliteratorSpec& trg, const UnicodeString& variant) const { UnicodeString ID; TransliteratorIDParser::STVtoID(src, trg, variant, ID); - Entry *e = (Entry*) registry.get(ID); + TransliteratorEntry *e = (TransliteratorEntry*) registry.get(ID); DEBUG_useEntry(e); return e; } @@ -930,10 +1072,10 @@ Entry* TransliteratorRegistry::findInDynamicStore(const Spec& src, * * Caller does NOT own returned object. */ -Entry* TransliteratorRegistry::findInStaticStore(const Spec& src, - const Spec& trg, +TransliteratorEntry* TransliteratorRegistry::findInStaticStore(const TransliteratorSpec& src, + const TransliteratorSpec& trg, const UnicodeString& variant) { - Entry* entry = 0; + TransliteratorEntry* entry = 0; if (src.isLocale()) { entry = findInBundle(src, trg, variant, UTRANS_FORWARD); } else if (trg.isLocale()) { @@ -966,8 +1108,8 @@ static const UChar TRANSLITERATE[] = {84,114,97,110,115,108,105,116,101,114,97,1 * On success, create a new Entry object, populate it, and return it. * The caller owns the returned object. */ -Entry* TransliteratorRegistry::findInBundle(const Spec& specToOpen, - const Spec& specToFind, +TransliteratorEntry* TransliteratorRegistry::findInBundle(const TransliteratorSpec& specToOpen, + const TransliteratorSpec& specToFind, const UnicodeString& variant, UTransDirection direction) { @@ -983,35 +1125,33 @@ Entry* TransliteratorRegistry::findInBundle(const Spec& specToOpen, // but must be consistent and documented. if (pass == 0) { utag.append(direction == UTRANS_FORWARD ? - TRANSLITERATE_TO : TRANSLITERATE_FROM); + TRANSLITERATE_TO : TRANSLITERATE_FROM, -1); } else { - utag.append(TRANSLITERATE); + utag.append(TRANSLITERATE, -1); } UnicodeString s(specToFind.get()); - utag.append(s.toUpper()); - CharString tag(utag); - + utag.append(s.toUpper("")); UErrorCode status = U_ZERO_ERROR; - ResourceBundle subres(specToOpen.getBundle().get(tag, status)); + ResourceBundle subres(specToOpen.getBundle().get( + CharString().appendInvariantChars(utag, status).data(), status)); if (U_FAILURE(status) || status == U_USING_DEFAULT_WARNING) { continue; } - - if (specToOpen.get() != subres.getLocale().getName()) { + + s.truncate(0); + if (specToOpen.get() != LocaleUtility::initNameFromLocale(subres.getLocale(), s)) { continue; } - + if (variant.length() != 0) { - CharString var(variant); status = U_ZERO_ERROR; - resStr = subres.getStringEx(var, status); + resStr = subres.getStringEx( + CharString().appendInvariantChars(variant, status).data(), status); if (U_SUCCESS(status)) { // Exit loop successfully break; } - } - - else { + } else { // Variant is empty, which means match the first variant listed. status = U_ZERO_ERROR; resStr = subres.getStringEx(1, status); @@ -1029,7 +1169,7 @@ Entry* TransliteratorRegistry::findInBundle(const Spec& specToOpen, // We have succeeded in loading a string from the locale // resources. Create a new registry entry to hold it and return it. - Entry *entry = new Entry(); + TransliteratorEntry *entry = new TransliteratorEntry(); if (entry != 0) { // The direction is always forward for the // TransliterateTo_xxx and TransliterateFrom_xxx @@ -1038,7 +1178,7 @@ Entry* TransliteratorRegistry::findInBundle(const Spec& specToOpen, // the direction is the value passed in to this // function. int32_t dir = (pass == 0) ? UTRANS_FORWARD : direction; - entry->entryType = Entry::LOCALE_RULES; + entry->entryType = TransliteratorEntry::LOCALE_RULES; entry->stringArg = resStr; entry->intArg = dir; } @@ -1049,7 +1189,7 @@ Entry* TransliteratorRegistry::findInBundle(const Spec& specToOpen, /** * Convenience method. Calls 3-arg find(). */ -Entry* TransliteratorRegistry::find(const UnicodeString& ID) { +TransliteratorEntry* TransliteratorRegistry::find(const UnicodeString& ID) { UnicodeString source, target, variant; UBool sawSource; TransliteratorIDParser::IDtoSTV(ID, source, target, variant, sawSource); @@ -1077,13 +1217,25 @@ Entry* TransliteratorRegistry::find(const UnicodeString& ID) { * * Caller does NOT own returned object. Return 0 on failure. */ -Entry* TransliteratorRegistry::find(UnicodeString& source, +TransliteratorEntry* TransliteratorRegistry::find(UnicodeString& source, UnicodeString& target, UnicodeString& variant) { - Spec src(source); - Spec trg(target); - Entry* entry; + TransliteratorSpec src(source); + TransliteratorSpec trg(target); + TransliteratorEntry* entry; + + // Seek exact match in hashtable. Temporary fix for ICU 4.6. + // TODO: The general logic for finding a matching transliterator needs to be reviewed. + // ICU ticket #8089 + UnicodeString ID; + TransliteratorIDParser::STVtoID(source, target, variant, ID); + entry = (TransliteratorEntry*) registry.get(ID); + if (entry != 0) { + // std::string ss; + // std::cout << ID.toUTF8String(ss) << std::endl; + return entry; + } if (variant.length() != 0) { @@ -1141,65 +1293,79 @@ Entry* TransliteratorRegistry::find(UnicodeString& source, * modified. */ Transliterator* TransliteratorRegistry::instantiateEntry(const UnicodeString& ID, - Entry *entry, + TransliteratorEntry *entry, TransliteratorAlias* &aliasReturn, UErrorCode& status) { Transliterator *t = 0; U_ASSERT(aliasReturn == 0); switch (entry->entryType) { - case Entry::RBT_DATA: + case TransliteratorEntry::RBT_DATA: t = new RuleBasedTransliterator(ID, entry->u.data); if (t == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return t; - case Entry::PROTOTYPE: + case TransliteratorEntry::PROTOTYPE: t = entry->u.prototype->clone(); if (t == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return t; - case Entry::ALIAS: - aliasReturn = new TransliteratorAlias(entry->stringArg); + case TransliteratorEntry::ALIAS: + aliasReturn = new TransliteratorAlias(entry->stringArg, entry->compoundFilter); if (aliasReturn == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return 0; - case Entry::FACTORY: + case TransliteratorEntry::FACTORY: t = entry->u.factory.function(ID, entry->u.factory.context); if (t == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return t; - case Entry::COMPOUND_RBT: + case TransliteratorEntry::COMPOUND_RBT: { - UnicodeString id((UChar)0x005F); /* "_" */ - Transliterator *t = new RuleBasedTransliterator(id, entry->u.data); - if (t == 0) { + UVector* rbts = new UVector(entry->u.dataVector->size(), status); + // Check for null pointer + if (rbts == NULL) { status = U_MEMORY_ALLOCATION_ERROR; + return NULL; + } + int32_t passNumber = 1; + for (int32_t i = 0; U_SUCCESS(status) && i < entry->u.dataVector->size(); i++) { + // TODO: Should passNumber be turned into a decimal-string representation (1 -> "1")? + Transliterator* tl = new RuleBasedTransliterator(UnicodeString(CompoundTransliterator::PASS_STRING) + UnicodeString(passNumber++), + (TransliterationRuleData*)(entry->u.dataVector->elementAt(i)), FALSE); + if (tl == 0) + status = U_MEMORY_ALLOCATION_ERROR; + else + rbts->addElement(tl, status); + } + if (U_FAILURE(status)) { + delete rbts; return 0; } - aliasReturn = new TransliteratorAlias(ID, entry->stringArg, t, entry->intArg, entry->compoundFilter); + aliasReturn = new TransliteratorAlias(ID, entry->stringArg, rbts, entry->compoundFilter); } if (aliasReturn == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return 0; - case Entry::LOCALE_RULES: + case TransliteratorEntry::LOCALE_RULES: aliasReturn = new TransliteratorAlias(ID, entry->stringArg, (UTransDirection) entry->intArg); if (aliasReturn == 0) { status = U_MEMORY_ALLOCATION_ERROR; } return 0; - case Entry::RULES_FORWARD: - case Entry::RULES_REVERSE: + case TransliteratorEntry::RULES_FORWARD: + case TransliteratorEntry::RULES_REVERSE: // Process the rule data into a TransliteratorRuleData object, // and possibly also into an ::id header and/or footer. Then // we modify the registry with the parsed data and retry. { - TransliteratorParser parser; + TransliteratorParser parser(status); // We use the file name, taken from another resource bundle // 2-d array at static init time, as a locale language. We're @@ -1226,7 +1392,7 @@ Transliterator* TransliteratorRegistry::instantiateEntry(const UnicodeString& ID // transliterators; if it lists something that's not // installed, we'll get an error from ResourceBundle. aliasReturn = new TransliteratorAlias(ID, rules, - ((entry->entryType == Entry::RULES_REVERSE) ? + ((entry->entryType == TransliteratorEntry::RULES_REVERSE) ? UTRANS_REVERSE : UTRANS_FORWARD)); if (aliasReturn == 0) { status = U_MEMORY_ALLOCATION_ERROR; @@ -1235,8 +1401,7 @@ Transliterator* TransliteratorRegistry::instantiateEntry(const UnicodeString& ID } return 0; default: - U_ASSERT(FALSE); // can't get here - return 0; + UPRV_UNREACHABLE; // can't get here } } U_NAMESPACE_END