am 3ce4c17d: Update has shortcut targets and has bigrams flags.
* commit '3ce4c17d00c7e2c41b8dc7081cc1aeb29c75d8f4': Update has shortcut targets and has bigrams flags.main
commit
58a7ecfc3f
|
@ -39,6 +39,13 @@ class DynamicPatriciaTrieWritingUtils {
|
||||||
static bool writePtNodeArraySizeAndAdvancePosition(BufferWithExtendableBuffer *const buffer,
|
static bool writePtNodeArraySizeAndAdvancePosition(BufferWithExtendableBuffer *const buffer,
|
||||||
const size_t arraySize, int *const arraySizeFieldPos);
|
const size_t arraySize, int *const arraySizeFieldPos);
|
||||||
|
|
||||||
|
static bool writeFlags(BufferWithExtendableBuffer *const buffer,
|
||||||
|
const DynamicPatriciaTrieReadingUtils::NodeFlags nodeFlags,
|
||||||
|
const int nodeFlagsFieldPos) {
|
||||||
|
int writingPos = nodeFlagsFieldPos;
|
||||||
|
return writeFlagsAndAdvancePosition(buffer, nodeFlags, &writingPos);
|
||||||
|
}
|
||||||
|
|
||||||
static bool writeFlagsAndAdvancePosition(BufferWithExtendableBuffer *const buffer,
|
static bool writeFlagsAndAdvancePosition(BufferWithExtendableBuffer *const buffer,
|
||||||
const DynamicPatriciaTrieReadingUtils::NodeFlags nodeFlags,
|
const DynamicPatriciaTrieReadingUtils::NodeFlags nodeFlags,
|
||||||
int *const nodeFlagsFieldPos);
|
int *const nodeFlagsFieldPos);
|
||||||
|
|
|
@ -160,8 +160,21 @@ bool Ver4PatriciaTrieNodeWriter::addNewBigramEntry(
|
||||||
const PtNodeParams *const sourcePtNodeParams,
|
const PtNodeParams *const sourcePtNodeParams,
|
||||||
const PtNodeParams *const targetPtNodeParam, const int probability, const int timestamp,
|
const PtNodeParams *const targetPtNodeParam, const int probability, const int timestamp,
|
||||||
bool *const outAddedNewBigram) {
|
bool *const outAddedNewBigram) {
|
||||||
return mBigramPolicy->addNewEntry(sourcePtNodeParams->getTerminalId(),
|
if (!mBigramPolicy->addNewEntry(sourcePtNodeParams->getTerminalId(),
|
||||||
targetPtNodeParam->getTerminalId(), probability, timestamp, outAddedNewBigram);
|
targetPtNodeParam->getTerminalId(), probability, timestamp, outAddedNewBigram)) {
|
||||||
|
AKLOGE("Cannot add new bigram entry. terminalId: %d, targetTerminalId: %d",
|
||||||
|
sourcePtNodeParams->getTerminalId(), targetPtNodeParam->getTerminalId());
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
if (!sourcePtNodeParams->hasBigrams()) {
|
||||||
|
// Update has bigrams flag.
|
||||||
|
return updatePtNodeFlags(sourcePtNodeParams->getHeadPos(),
|
||||||
|
sourcePtNodeParams->isBlacklisted(), sourcePtNodeParams->isNotAWord(),
|
||||||
|
sourcePtNodeParams->isTerminal(), sourcePtNodeParams->hasShortcutTargets(),
|
||||||
|
true /* hasBigrams */,
|
||||||
|
sourcePtNodeParams->getCodePointCount() > 1 /* hasMultipleChars */);
|
||||||
|
}
|
||||||
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
bool Ver4PatriciaTrieNodeWriter::removeBigramEntry(
|
bool Ver4PatriciaTrieNodeWriter::removeBigramEntry(
|
||||||
|
@ -220,8 +233,31 @@ bool Ver4PatriciaTrieNodeWriter::updateAllPositionFields(
|
||||||
bool Ver4PatriciaTrieNodeWriter::addShortcutTarget(const PtNodeParams *const ptNodeParams,
|
bool Ver4PatriciaTrieNodeWriter::addShortcutTarget(const PtNodeParams *const ptNodeParams,
|
||||||
const int *const targetCodePoints, const int targetCodePointCount,
|
const int *const targetCodePoints, const int targetCodePointCount,
|
||||||
const int shortcutProbability) {
|
const int shortcutProbability) {
|
||||||
return mShortcutPolicy->addNewShortcut(ptNodeParams->getTerminalId(),
|
if (!mShortcutPolicy->addNewShortcut(ptNodeParams->getTerminalId(),
|
||||||
targetCodePoints, targetCodePointCount, shortcutProbability);
|
targetCodePoints, targetCodePointCount, shortcutProbability)) {
|
||||||
|
AKLOGE("Cannot add new shortuct entry. terminalId: %d", ptNodeParams->getTerminalId());
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
if (!ptNodeParams->hasShortcutTargets()) {
|
||||||
|
// Update has shortcut targets flag.
|
||||||
|
return updatePtNodeFlags(ptNodeParams->getHeadPos(),
|
||||||
|
ptNodeParams->isBlacklisted(), ptNodeParams->isNotAWord(),
|
||||||
|
ptNodeParams->isTerminal(), true /* hasShortcutTargets */,
|
||||||
|
ptNodeParams->hasBigrams(),
|
||||||
|
ptNodeParams->getCodePointCount() > 1 /* hasMultipleChars */);
|
||||||
|
}
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
|
||||||
|
bool Ver4PatriciaTrieNodeWriter::updatePtNodeHasBigramsAndShortcutTargetsFlags(
|
||||||
|
const PtNodeParams *const ptNodeParams) {
|
||||||
|
const bool hasBigrams = mBuffers->getBigramDictContent()->getBigramListHeadPos(
|
||||||
|
ptNodeParams->getTerminalId()) != NOT_A_DICT_POS;
|
||||||
|
const bool hasShortcutTargets = mBuffers->getShortcutDictContent()->getShortcutListHeadPos(
|
||||||
|
ptNodeParams->getTerminalId()) != NOT_A_DICT_POS;
|
||||||
|
return updatePtNodeFlags(ptNodeParams->getHeadPos(), ptNodeParams->isBlacklisted(),
|
||||||
|
ptNodeParams->isNotAWord(), ptNodeParams->isTerminal(), hasShortcutTargets,
|
||||||
|
hasBigrams, ptNodeParams->getCodePointCount() > 1 /* hasMultipleChars */);
|
||||||
}
|
}
|
||||||
|
|
||||||
bool Ver4PatriciaTrieNodeWriter::writePtNodeAndGetTerminalIdAndAdvancePosition(
|
bool Ver4PatriciaTrieNodeWriter::writePtNodeAndGetTerminalIdAndAdvancePosition(
|
||||||
|
@ -273,19 +309,9 @@ bool Ver4PatriciaTrieNodeWriter::writePtNodeAndGetTerminalIdAndAdvancePosition(
|
||||||
ptNodeParams->getChildrenPos(), ptNodeWritingPos)) {
|
ptNodeParams->getChildrenPos(), ptNodeWritingPos)) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
// Create node flags and write them.
|
return updatePtNodeFlags(nodePos, ptNodeParams->isBlacklisted(), ptNodeParams->isNotAWord(),
|
||||||
PatriciaTrieReadingUtils::NodeFlags nodeFlags =
|
isTerminal, ptNodeParams->hasShortcutTargets(), ptNodeParams->hasBigrams(),
|
||||||
PatriciaTrieReadingUtils::createAndGetFlags(ptNodeParams->isBlacklisted(),
|
ptNodeParams->getCodePointCount() > 1 /* hasMultipleChars */);
|
||||||
ptNodeParams->isNotAWord(), isTerminal,
|
|
||||||
ptNodeParams->hasShortcutTargets(), ptNodeParams->hasBigrams(),
|
|
||||||
ptNodeParams->getCodePointCount() > 1 /* hasMultipleChars */,
|
|
||||||
CHILDREN_POSITION_FIELD_SIZE);
|
|
||||||
int flagsFieldPos = nodePos;
|
|
||||||
if (!DynamicPatriciaTrieWritingUtils::writeFlagsAndAdvancePosition(mTrieBuffer, nodeFlags,
|
|
||||||
&flagsFieldPos)) {
|
|
||||||
return false;
|
|
||||||
}
|
|
||||||
return true;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
const ProbabilityEntry Ver4PatriciaTrieNodeWriter::createUpdatedEntryFrom(
|
const ProbabilityEntry Ver4PatriciaTrieNodeWriter::createUpdatedEntryFrom(
|
||||||
|
@ -304,4 +330,19 @@ const ProbabilityEntry Ver4PatriciaTrieNodeWriter::createUpdatedEntryFrom(
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
bool Ver4PatriciaTrieNodeWriter::updatePtNodeFlags(const int ptNodePos,
|
||||||
|
const bool isBlacklisted, const bool isNotAWord, const bool isTerminal,
|
||||||
|
const bool hasShortcutTargets, const bool hasBigrams, const bool hasMultipleChars) {
|
||||||
|
// Create node flags and write them.
|
||||||
|
PatriciaTrieReadingUtils::NodeFlags nodeFlags =
|
||||||
|
PatriciaTrieReadingUtils::createAndGetFlags(isBlacklisted, isNotAWord, isTerminal,
|
||||||
|
hasShortcutTargets, hasBigrams, hasMultipleChars,
|
||||||
|
CHILDREN_POSITION_FIELD_SIZE);
|
||||||
|
if (!DynamicPatriciaTrieWritingUtils::writeFlags(mTrieBuffer, nodeFlags, ptNodePos)) {
|
||||||
|
AKLOGE("Cannot write PtNode flags. flags: %x, pos: %d", nodeFlags, ptNodePos);
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
|
@ -87,6 +87,8 @@ class Ver4PatriciaTrieNodeWriter : public PtNodeWriter {
|
||||||
const int *const targetCodePoints, const int targetCodePointCount,
|
const int *const targetCodePoints, const int targetCodePointCount,
|
||||||
const int shortcutProbability);
|
const int shortcutProbability);
|
||||||
|
|
||||||
|
bool updatePtNodeHasBigramsAndShortcutTargetsFlags(const PtNodeParams *const ptNodeParams);
|
||||||
|
|
||||||
private:
|
private:
|
||||||
DISALLOW_COPY_AND_ASSIGN(Ver4PatriciaTrieNodeWriter);
|
DISALLOW_COPY_AND_ASSIGN(Ver4PatriciaTrieNodeWriter);
|
||||||
|
|
||||||
|
@ -100,6 +102,10 @@ class Ver4PatriciaTrieNodeWriter : public PtNodeWriter {
|
||||||
const ProbabilityEntry *const originalProbabilityEntry, const int newProbability,
|
const ProbabilityEntry *const originalProbabilityEntry, const int newProbability,
|
||||||
const int timestamp) const;
|
const int timestamp) const;
|
||||||
|
|
||||||
|
bool updatePtNodeFlags(const int ptNodePos, const bool isBlacklisted, const bool isNotAWord,
|
||||||
|
const bool isTerminal, const bool hasShortcutTargets, const bool hasBigrams,
|
||||||
|
const bool hasMultipleChars);
|
||||||
|
|
||||||
static const int CHILDREN_POSITION_FIELD_SIZE;
|
static const int CHILDREN_POSITION_FIELD_SIZE;
|
||||||
|
|
||||||
BufferWithExtendableBuffer *const mTrieBuffer;
|
BufferWithExtendableBuffer *const mTrieBuffer;
|
||||||
|
|
|
@ -172,18 +172,18 @@ bool Ver4PatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
|
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
|
||||||
TraversePolicyToUpdateAllTerminalIds traversePolicyToUpdateAllTerminalIds(&newPtNodeWriter,
|
TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds
|
||||||
&terminalIdMap);
|
traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds(&newPtNodeWriter, &terminalIdMap);
|
||||||
if (!newDictReadingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
|
if (!newDictReadingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
|
||||||
&traversePolicyToUpdateAllTerminalIds)) {
|
&traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds)) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
*outUnigramCount = traversePolicyToUpdateAllPositionFields.getUnigramCount();
|
*outUnigramCount = traversePolicyToUpdateAllPositionFields.getUnigramCount();
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
bool Ver4PatriciaTrieWritingHelper::TraversePolicyToUpdateAllTerminalIds::onVisitingPtNode(
|
bool Ver4PatriciaTrieWritingHelper::TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds
|
||||||
const PtNodeParams *const ptNodeParams) {
|
::onVisitingPtNode(const PtNodeParams *const ptNodeParams) {
|
||||||
if (!ptNodeParams->isTerminal()) {
|
if (!ptNodeParams->isTerminal()) {
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
@ -194,7 +194,10 @@ bool Ver4PatriciaTrieWritingHelper::TraversePolicyToUpdateAllTerminalIds::onVisi
|
||||||
ptNodeParams->getTerminalId(), mTerminalIdMap->size());
|
ptNodeParams->getTerminalId(), mTerminalIdMap->size());
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
return mPtNodeWriter->updateTerminalId(ptNodeParams, it->second);
|
if (!mPtNodeWriter->updateTerminalId(ptNodeParams, it->second)) {
|
||||||
|
AKLOGE("Cannot update terminal id. %d -> %d", it->first, it->second);
|
||||||
|
}
|
||||||
|
return mPtNodeWriter->updatePtNodeHasBigramsAndShortcutTargetsFlags(ptNodeParams);
|
||||||
}
|
}
|
||||||
|
|
||||||
} // namespace latinime
|
} // namespace latinime
|
||||||
|
|
|
@ -41,12 +41,13 @@ class Ver4PatriciaTrieWritingHelper {
|
||||||
private:
|
private:
|
||||||
DISALLOW_IMPLICIT_CONSTRUCTORS(Ver4PatriciaTrieWritingHelper);
|
DISALLOW_IMPLICIT_CONSTRUCTORS(Ver4PatriciaTrieWritingHelper);
|
||||||
|
|
||||||
class TraversePolicyToUpdateAllTerminalIds
|
class TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds
|
||||||
: public DynamicPatriciaTrieReadingHelper::TraversingEventListener {
|
: public DynamicPatriciaTrieReadingHelper::TraversingEventListener {
|
||||||
public:
|
public:
|
||||||
TraversePolicyToUpdateAllTerminalIds(Ver4PatriciaTrieNodeWriter *const ptNodeWriter,
|
TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds(
|
||||||
|
Ver4PatriciaTrieNodeWriter *const ptNodeWriter,
|
||||||
const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap)
|
const TerminalPositionLookupTable::TerminalIdMap *const terminalIdMap)
|
||||||
: mPtNodeWriter(ptNodeWriter), mTerminalIdMap(terminalIdMap) {};
|
: mPtNodeWriter(ptNodeWriter), mTerminalIdMap(terminalIdMap) {}
|
||||||
|
|
||||||
bool onAscend() { return true; }
|
bool onAscend() { return true; }
|
||||||
|
|
||||||
|
@ -57,7 +58,7 @@ class Ver4PatriciaTrieWritingHelper {
|
||||||
bool onVisitingPtNode(const PtNodeParams *const ptNodeParams);
|
bool onVisitingPtNode(const PtNodeParams *const ptNodeParams);
|
||||||
|
|
||||||
private:
|
private:
|
||||||
DISALLOW_IMPLICIT_CONSTRUCTORS(TraversePolicyToUpdateAllTerminalIds);
|
DISALLOW_IMPLICIT_CONSTRUCTORS(TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds);
|
||||||
|
|
||||||
Ver4PatriciaTrieNodeWriter *const mPtNodeWriter;
|
Ver4PatriciaTrieNodeWriter *const mPtNodeWriter;
|
||||||
const TerminalPositionLookupTable::TerminalIdMap *const mTerminalIdMap;
|
const TerminalPositionLookupTable::TerminalIdMap *const mTerminalIdMap;
|
||||||
|
|
Loading…
Reference in New Issue