bool DynamicPatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
const HeaderPolicy *const headerPolicy, BufferWithExtendableBuffer *const bufferToWrite,
int *const outUnigramCount, int *const outBigramCount) {
DynamicPatriciaTrieReadingHelper readingHelper(mBuffer, mBigramPolicy, mShortcutPolicy);
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPatriciaTrieGcEventListeners
::TraversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted
traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted(
headerPolicy, this, mBuffer, mNeedsToDecay);
if (!readingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted)) {
return false;
}
if (mNeedsToDecay && traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted
.getValidUnigramCount() > ForgettingCurveUtils::MAX_UNIGRAM_COUNT_AFTER_GC) {
// TODO: Remove more unigrams.
}
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPatriciaTrieGcEventListeners::TraversePolicyToUpdateBigramProbability
traversePolicyToUpdateBigramProbability(mBigramPolicy);
if (!readingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateBigramProbability)) {
return false;
}
if (mNeedsToDecay && traversePolicyToUpdateBigramProbability.getValidBigramEntryCount()
> ForgettingCurveUtils::MAX_BIGRAM_COUNT_AFTER_GC) {
// TODO: Remove more bigrams.
}
// Mapping from positions in mBuffer to positions in bufferToWrite.
DictPositionRelocationMap dictPositionRelocationMap;
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPatriciaTrieGcEventListeners::TraversePolicyToPlaceAndWriteValidPtNodesToBuffer
traversePolicyToPlaceAndWriteValidPtNodesToBuffer(this, bufferToWrite,
&dictPositionRelocationMap);
if (!readingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToPlaceAndWriteValidPtNodesToBuffer)) {
return false;
}
// Create policy instance for the GCed dictionary.
DynamicShortcutListPolicy newDictShortcutPolicy(bufferToWrite);
DynamicBigramListPolicy newDictBigramPolicy(headerPolicy, bufferToWrite, &newDictShortcutPolicy,
mNeedsToDecay);
// Create reading helper for the GCed dictionary.
DynamicPatriciaTrieReadingHelper newDictReadingHelper(bufferToWrite, &newDictBigramPolicy,
&newDictShortcutPolicy);
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPatriciaTrieGcEventListeners::TraversePolicyToUpdateAllPositionFields
traversePolicyToUpdateAllPositionFields(this, &newDictBigramPolicy, bufferToWrite,
&dictPositionRelocationMap);
if (!newDictReadingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToUpdateAllPositionFields)) {
return false;
}
*outUnigramCount = traversePolicyToUpdateAllPositionFields.getUnigramCount();
*outBigramCount = traversePolicyToUpdateAllPositionFields.getBigramCount();
return true;
}
bool Ver4PatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
const HeaderPolicy *const headerPolicy, Ver4DictBuffers *const buffersToWrite,
int *const outUnigramCount, int *const outBigramCount) {
Ver4PatriciaTrieNodeReader ptNodeReader(mBuffers->getTrieBuffer(),
mBuffers->getProbabilityDictContent(), headerPolicy);
Ver4PtNodeArrayReader ptNodeArrayReader(mBuffers->getTrieBuffer());
Ver4BigramListPolicy bigramPolicy(mBuffers->getMutableBigramDictContent(),
mBuffers->getTerminalPositionLookupTable(), headerPolicy);
Ver4ShortcutListPolicy shortcutPolicy(mBuffers->getMutableShortcutDictContent(),
mBuffers->getTerminalPositionLookupTable());
Ver4PatriciaTrieNodeWriter ptNodeWriter(mBuffers->getWritableTrieBuffer(),
mBuffers, headerPolicy, &ptNodeReader, &ptNodeArrayReader, &bigramPolicy,
&shortcutPolicy);
DynamicPtReadingHelper readingHelper(&ptNodeReader, &ptNodeArrayReader);
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPtGcEventListeners
::TraversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted
traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted(
&ptNodeWriter);
if (!readingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted)) {
return false;
}
const int unigramCount = traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted
.getValidUnigramCount();
const int maxUnigramCount = headerPolicy->getMaxUnigramCount();
if (headerPolicy->isDecayingDict() && unigramCount > maxUnigramCount) {
if (!truncateUnigrams(&ptNodeReader, &ptNodeWriter, maxUnigramCount)) {
AKLOGE("Cannot remove unigrams. current: %d, max: %d", unigramCount,
maxUnigramCount);
return false;
}
}
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPtGcEventListeners::TraversePolicyToUpdateBigramProbability
traversePolicyToUpdateBigramProbability(&ptNodeWriter);
if (!readingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateBigramProbability)) {
return false;
}
const int bigramCount = traversePolicyToUpdateBigramProbability.getValidBigramEntryCount();
const int maxBigramCount = headerPolicy->getMaxBigramCount();
if (headerPolicy->isDecayingDict() && bigramCount > maxBigramCount) {
if (!truncateBigrams(maxBigramCount)) {
AKLOGE("Cannot remove bigrams. current: %d, max: %d", bigramCount, maxBigramCount);
return false;
}
}
// Mapping from positions in mBuffer to positions in bufferToWrite.
PtNodeWriter::DictPositionRelocationMap dictPositionRelocationMap;
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
Ver4PatriciaTrieNodeWriter ptNodeWriterForNewBuffers(buffersToWrite->getWritableTrieBuffer(),
buffersToWrite, headerPolicy, &ptNodeReader, &ptNodeArrayReader, &bigramPolicy,
&shortcutPolicy);
DynamicPtGcEventListeners::TraversePolicyToPlaceAndWriteValidPtNodesToBuffer
traversePolicyToPlaceAndWriteValidPtNodesToBuffer(&ptNodeWriterForNewBuffers,
buffersToWrite->getWritableTrieBuffer(), &dictPositionRelocationMap);
if (!readingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToPlaceAndWriteValidPtNodesToBuffer)) {
return false;
}
// Create policy instances for the GCed dictionary.
Ver4PatriciaTrieNodeReader newPtNodeReader(buffersToWrite->getTrieBuffer(),
buffersToWrite->getProbabilityDictContent(), headerPolicy);
Ver4PtNodeArrayReader newPtNodeArrayreader(buffersToWrite->getTrieBuffer());
Ver4BigramListPolicy newBigramPolicy(buffersToWrite->getMutableBigramDictContent(),
buffersToWrite->getTerminalPositionLookupTable(), headerPolicy);
Ver4ShortcutListPolicy newShortcutPolicy(buffersToWrite->getMutableShortcutDictContent(),
buffersToWrite->getTerminalPositionLookupTable());
Ver4PatriciaTrieNodeWriter newPtNodeWriter(buffersToWrite->getWritableTrieBuffer(),
buffersToWrite, headerPolicy, &newPtNodeReader, &newPtNodeArrayreader, &newBigramPolicy,
&newShortcutPolicy);
// Re-assign terminal IDs for valid terminal PtNodes.
TerminalPositionLookupTable::TerminalIdMap terminalIdMap;
if(!buffersToWrite->getMutableTerminalPositionLookupTable()->runGCTerminalIds(
&terminalIdMap)) {
return false;
}
// Run GC for probability dict content.
if (!buffersToWrite->getMutableProbabilityDictContent()->runGC(&terminalIdMap,
mBuffers->getProbabilityDictContent())) {
return false;
}
// Run GC for bigram dict content.
if(!buffersToWrite->getMutableBigramDictContent()->runGC(&terminalIdMap,
mBuffers->getBigramDictContent(), outBigramCount)) {
return false;
}
// Run GC for shortcut dict content.
if(!buffersToWrite->getMutableShortcutDictContent()->runGC(&terminalIdMap,
mBuffers->getShortcutDictContent())) {
return false;
}
DynamicPtReadingHelper newDictReadingHelper(&newPtNodeReader, &newPtNodeArrayreader);
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPtGcEventListeners::TraversePolicyToUpdateAllPositionFields
traversePolicyToUpdateAllPositionFields(&newPtNodeWriter, &dictPositionRelocationMap);
if (!newDictReadingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToUpdateAllPositionFields)) {
return false;
}
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds
traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds(&newPtNodeWriter, &terminalIdMap);
if (!newDictReadingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds)) {
return false;
}
*outUnigramCount = traversePolicyToUpdateAllPositionFields.getUnigramCount();
return true;
}
bool Ver4PatriciaTrieWritingHelper::runGC(const int rootPtNodeArrayPos,
const HeaderPolicy *const headerPolicy, Ver4DictBuffers *const buffersToWrite,
MutableEntryCounters *const outEntryCounters) {
Ver4PatriciaTrieNodeReader ptNodeReader(mBuffers->getTrieBuffer());
Ver4PtNodeArrayReader ptNodeArrayReader(mBuffers->getTrieBuffer());
Ver4ShortcutListPolicy shortcutPolicy(mBuffers->getMutableShortcutDictContent(),
mBuffers->getTerminalPositionLookupTable());
Ver4PatriciaTrieNodeWriter ptNodeWriter(mBuffers->getWritableTrieBuffer(),
mBuffers, &ptNodeReader, &ptNodeArrayReader, &shortcutPolicy);
if (!mBuffers->getMutableLanguageModelDictContent()->updateAllProbabilityEntriesForGC(
headerPolicy, outEntryCounters)) {
AKLOGE("Failed to update probabilities in language model dict content.");
return false;
}
if (headerPolicy->isDecayingDict()) {
const EntryCounts &maxEntryCounts = headerPolicy->getMaxNgramCounts();
if (!mBuffers->getMutableLanguageModelDictContent()->truncateEntries(
outEntryCounters->getEntryCounts(), maxEntryCounts, headerPolicy,
outEntryCounters)) {
AKLOGE("Failed to truncate entries in language model dict content.");
return false;
}
}
DynamicPtReadingHelper readingHelper(&ptNodeReader, &ptNodeArrayReader);
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPtGcEventListeners
::TraversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted
traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted(
&ptNodeWriter);
if (!readingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateUnigramProbabilityAndMarkUselessPtNodesAsDeleted)) {
return false;
}
// Mapping from positions in mBuffer to positions in bufferToWrite.
PtNodeWriter::DictPositionRelocationMap dictPositionRelocationMap;
readingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
Ver4PatriciaTrieNodeWriter ptNodeWriterForNewBuffers(buffersToWrite->getWritableTrieBuffer(),
buffersToWrite, &ptNodeReader, &ptNodeArrayReader, &shortcutPolicy);
DynamicPtGcEventListeners::TraversePolicyToPlaceAndWriteValidPtNodesToBuffer
traversePolicyToPlaceAndWriteValidPtNodesToBuffer(&ptNodeWriterForNewBuffers,
buffersToWrite->getWritableTrieBuffer(), &dictPositionRelocationMap);
if (!readingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToPlaceAndWriteValidPtNodesToBuffer)) {
return false;
}
// Create policy instances for the GCed dictionary.
Ver4PatriciaTrieNodeReader newPtNodeReader(buffersToWrite->getTrieBuffer());
Ver4PtNodeArrayReader newPtNodeArrayreader(buffersToWrite->getTrieBuffer());
Ver4ShortcutListPolicy newShortcutPolicy(buffersToWrite->getMutableShortcutDictContent(),
buffersToWrite->getTerminalPositionLookupTable());
Ver4PatriciaTrieNodeWriter newPtNodeWriter(buffersToWrite->getWritableTrieBuffer(),
buffersToWrite, &newPtNodeReader, &newPtNodeArrayreader,
&newShortcutPolicy);
// Re-assign terminal IDs for valid terminal PtNodes.
TerminalPositionLookupTable::TerminalIdMap terminalIdMap;
if(!buffersToWrite->getMutableTerminalPositionLookupTable()->runGCTerminalIds(
&terminalIdMap)) {
return false;
}
// Run GC for language model dict content.
if (!buffersToWrite->getMutableLanguageModelDictContent()->runGC(&terminalIdMap,
mBuffers->getLanguageModelDictContent())) {
return false;
}
// Run GC for shortcut dict content.
if(!buffersToWrite->getMutableShortcutDictContent()->runGC(&terminalIdMap,
mBuffers->getShortcutDictContent())) {
return false;
}
DynamicPtReadingHelper newDictReadingHelper(&newPtNodeReader, &newPtNodeArrayreader);
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
DynamicPtGcEventListeners::TraversePolicyToUpdateAllPositionFields
traversePolicyToUpdateAllPositionFields(&newPtNodeWriter, &dictPositionRelocationMap);
if (!newDictReadingHelper.traverseAllPtNodesInPtNodeArrayLevelPreorderDepthFirstManner(
&traversePolicyToUpdateAllPositionFields)) {
return false;
}
newDictReadingHelper.initWithPtNodeArrayPos(rootPtNodeArrayPos);
TraversePolicyToUpdateAllPtNodeFlagsAndTerminalIds
traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds(&newPtNodeWriter, &terminalIdMap);
if (!newDictReadingHelper.traverseAllPtNodesInPostorderDepthFirstManner(
&traversePolicyToUpdateAllPtNodeFlagsAndTerminalIds)) {
return false;
}
return true;
}
