diff options
Diffstat (limited to 'native/jni/src')
25 files changed, 164 insertions, 88 deletions
diff --git a/native/jni/src/suggest/core/dictionary/dictionary.cpp b/native/jni/src/suggest/core/dictionary/dictionary.cpp index c3f422916..8d3f8a9f8 100644 --- a/native/jni/src/suggest/core/dictionary/dictionary.cpp +++ b/native/jni/src/suggest/core/dictionary/dictionary.cpp @@ -155,6 +155,14 @@ bool Dictionary::removeNgramEntry(const PrevWordsInfo *const prevWordsInfo, return mDictionaryStructureWithBufferPolicy->removeNgramEntry(prevWordsInfo, codePoints); } +bool Dictionary::updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView codePoints, const bool isValidWord, + const HistoricalInfo historicalInfo) { + TimeKeeper::setCurrentTime(); + return mDictionaryStructureWithBufferPolicy->updateCounter(prevWordsInfo, codePoints, + isValidWord, historicalInfo); +} + bool Dictionary::flush(const char *const filePath) { TimeKeeper::setCurrentTime(); return mDictionaryStructureWithBufferPolicy->flush(filePath); diff --git a/native/jni/src/suggest/core/dictionary/dictionary.h b/native/jni/src/suggest/core/dictionary/dictionary.h index 09f8eaceb..a58dbfbd7 100644 --- a/native/jni/src/suggest/core/dictionary/dictionary.h +++ b/native/jni/src/suggest/core/dictionary/dictionary.h @@ -22,6 +22,7 @@ #include "defines.h" #include "jni.h" #include "suggest/core/dictionary/ngram_listener.h" +#include "suggest/core/dictionary/property/historical_info.h" #include "suggest/core/dictionary/property/word_property.h" #include "suggest/core/policy/dictionary_header_structure_policy.h" #include "suggest/core/policy/dictionary_structure_with_buffer_policy.h" @@ -90,6 +91,10 @@ class Dictionary { bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView codePoints); + bool updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView codePoints, const bool isValidWord, + const HistoricalInfo historicalInfo); + bool flush(const char *const filePath); bool flushWithGC(const char *const filePath); diff --git a/native/jni/src/suggest/policyimpl/dictionary/utils/historical_info.h b/native/jni/src/suggest/core/dictionary/property/historical_info.h index 428ca8626..5ed9ebfca 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/utils/historical_info.h +++ b/native/jni/src/suggest/core/dictionary/property/historical_info.h @@ -34,7 +34,7 @@ class HistoricalInfo { return mTimestamp != NOT_A_TIMESTAMP; } - int getTimeStamp() const { + int getTimestamp() const { return mTimestamp; } @@ -47,12 +47,12 @@ class HistoricalInfo { } private: - // Copy constructor is public to use this class as a type of return value. - DISALLOW_ASSIGNMENT_OPERATOR(HistoricalInfo); + // Default copy constructor and assign operator are used for using in std::vector. - const int mTimestamp; - const int mLevel; - const int mCount; + // TODO: Make members const. + int mTimestamp; + int mLevel; + int mCount; }; } // namespace latinime #endif /* LATINIME_HISTORICAL_INFO_H */ diff --git a/native/jni/src/suggest/core/dictionary/property/ngram_property.h b/native/jni/src/suggest/core/dictionary/property/ngram_property.h index 5d822fa3b..dce460099 100644 --- a/native/jni/src/suggest/core/dictionary/property/ngram_property.h +++ b/native/jni/src/suggest/core/dictionary/property/ngram_property.h @@ -20,15 +20,16 @@ #include <vector> #include "defines.h" +#include "suggest/core/dictionary/property/historical_info.h" namespace latinime { class NgramProperty { public: NgramProperty(const std::vector<int> &&targetCodePoints, const int probability, - const int timestamp, const int level, const int count) + const HistoricalInfo historicalInfo) : mTargetCodePoints(std::move(targetCodePoints)), mProbability(probability), - mTimestamp(timestamp), mLevel(level), mCount(count) {} + mHistoricalInfo(historicalInfo) {} const std::vector<int> *getTargetCodePoints() const { return &mTargetCodePoints; @@ -38,16 +39,8 @@ class NgramProperty { return mProbability; } - int getTimestamp() const { - return mTimestamp; - } - - int getLevel() const { - return mLevel; - } - - int getCount() const { - return mCount; + const HistoricalInfo getHistoricalInfo() const { + return mHistoricalInfo; } private: @@ -57,9 +50,7 @@ class NgramProperty { // TODO: Make members const. std::vector<int> mTargetCodePoints; int mProbability; - int mTimestamp; - int mLevel; - int mCount; + HistoricalInfo mHistoricalInfo; }; } // namespace latinime #endif // LATINIME_NGRAM_PROPERTY_H diff --git a/native/jni/src/suggest/core/dictionary/property/unigram_property.h b/native/jni/src/suggest/core/dictionary/property/unigram_property.h index b7e7d6686..d1f0ab4ca 100644 --- a/native/jni/src/suggest/core/dictionary/property/unigram_property.h +++ b/native/jni/src/suggest/core/dictionary/property/unigram_property.h @@ -20,6 +20,7 @@ #include <vector> #include "defines.h" +#include "suggest/core/dictionary/property/historical_info.h" namespace latinime { @@ -50,15 +51,21 @@ class UnigramProperty { UnigramProperty() : mRepresentsBeginningOfSentence(false), mIsNotAWord(false), mIsBlacklisted(false), - mProbability(NOT_A_PROBABILITY), mTimestamp(NOT_A_TIMESTAMP), mLevel(0), mCount(0), - mShortcuts() {} + mProbability(NOT_A_PROBABILITY), mHistoricalInfo(), mShortcuts() {} UnigramProperty(const bool representsBeginningOfSentence, const bool isNotAWord, - const bool isBlacklisted, const int probability, const int timestamp, const int level, - const int count, const std::vector<ShortcutProperty> *const shortcuts) + const bool isBlacklisted, const int probability, const HistoricalInfo historicalInfo, + const std::vector<ShortcutProperty> &&shortcuts) : mRepresentsBeginningOfSentence(representsBeginningOfSentence), mIsNotAWord(isNotAWord), mIsBlacklisted(isBlacklisted), mProbability(probability), - mTimestamp(timestamp), mLevel(level), mCount(count), mShortcuts(*shortcuts) {} + mHistoricalInfo(historicalInfo), mShortcuts(std::move(shortcuts)) {} + + // Without shortcuts. + UnigramProperty(const bool representsBeginningOfSentence, const bool isNotAWord, + const bool isBlacklisted, const int probability, const HistoricalInfo historicalInfo) + : mRepresentsBeginningOfSentence(representsBeginningOfSentence), + mIsNotAWord(isNotAWord), mIsBlacklisted(isBlacklisted), mProbability(probability), + mHistoricalInfo(historicalInfo), mShortcuts() {} bool representsBeginningOfSentence() const { return mRepresentsBeginningOfSentence; @@ -85,16 +92,8 @@ class UnigramProperty { return mProbability; } - int getTimestamp() const { - return mTimestamp; - } - - int getLevel() const { - return mLevel; - } - - int getCount() const { - return mCount; + const HistoricalInfo getHistoricalInfo() const { + return mHistoricalInfo; } const std::vector<ShortcutProperty> &getShortcuts() const { @@ -110,10 +109,7 @@ class UnigramProperty { bool mIsNotAWord; bool mIsBlacklisted; int mProbability; - // Historical information - int mTimestamp; - int mLevel; - int mCount; + HistoricalInfo mHistoricalInfo; std::vector<ShortcutProperty> mShortcuts; }; } // namespace latinime diff --git a/native/jni/src/suggest/core/dictionary/property/word_property.cpp b/native/jni/src/suggest/core/dictionary/property/word_property.cpp index e3406fac5..caac8fe79 100644 --- a/native/jni/src/suggest/core/dictionary/property/word_property.cpp +++ b/native/jni/src/suggest/core/dictionary/property/word_property.cpp @@ -17,6 +17,7 @@ #include "suggest/core/dictionary/property/word_property.h" #include "utils/jni_data_utils.h" +#include "suggest/core/dictionary/property/historical_info.h" namespace latinime { @@ -31,8 +32,9 @@ void WordProperty::outputProperties(JNIEnv *const env, jintArray outCodePoints, !mNgrams.empty(), mUnigramProperty.hasShortcuts(), mUnigramProperty.representsBeginningOfSentence()}; env->SetBooleanArrayRegion(outFlags, 0 /* start */, NELEMS(flags), flags); - int probabilityInfo[] = {mUnigramProperty.getProbability(), mUnigramProperty.getTimestamp(), - mUnigramProperty.getLevel(), mUnigramProperty.getCount()}; + const HistoricalInfo &historicalInfo = mUnigramProperty.getHistoricalInfo(); + int probabilityInfo[] = {mUnigramProperty.getProbability(), historicalInfo.getTimestamp(), + historicalInfo.getLevel(), historicalInfo.getCount()}; env->SetIntArrayRegion(outProbabilityInfo, 0 /* start */, NELEMS(probabilityInfo), probabilityInfo); @@ -51,10 +53,10 @@ void WordProperty::outputProperties(JNIEnv *const env, jintArray outCodePoints, false /* needsNullTermination */); env->CallBooleanMethod(outBigramTargets, addMethodId, bigramWord1CodePointArray); env->DeleteLocalRef(bigramWord1CodePointArray); - + const HistoricalInfo &ngramHistoricalInfo = ngramProperty.getHistoricalInfo(); int bigramProbabilityInfo[] = {ngramProperty.getProbability(), - ngramProperty.getTimestamp(), ngramProperty.getLevel(), - ngramProperty.getCount()}; + ngramHistoricalInfo.getTimestamp(), ngramHistoricalInfo.getLevel(), + ngramHistoricalInfo.getCount()}; jintArray bigramProbabilityInfoArray = env->NewIntArray(NELEMS(bigramProbabilityInfo)); env->SetIntArrayRegion(bigramProbabilityInfoArray, 0 /* start */, NELEMS(bigramProbabilityInfo), bigramProbabilityInfo); diff --git a/native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h b/native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h index f4b97989f..6624b7921 100644 --- a/native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h +++ b/native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h @@ -21,6 +21,7 @@ #include "defines.h" #include "suggest/core/dictionary/binary_dictionary_shortcut_iterator.h" +#include "suggest/core/dictionary/property/historical_info.h" #include "suggest/core/dictionary/property/word_property.h" #include "suggest/core/dictionary/word_attributes.h" #include "utils/int_array_view.h" @@ -87,6 +88,11 @@ class DictionaryStructureWithBufferPolicy { virtual bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView wordCodePoints) = 0; + // Returns whether the update was success or not. + virtual bool updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo) = 0; + // Returns whether the flush was success or not. virtual bool flush(const char *const filePath) = 0; diff --git a/native/jni/src/suggest/core/session/prev_words_info.h b/native/jni/src/suggest/core/session/prev_words_info.h index 02e82a8e0..553d5ad07 100644 --- a/native/jni/src/suggest/core/session/prev_words_info.h +++ b/native/jni/src/suggest/core/session/prev_words_info.h @@ -33,7 +33,7 @@ class PrevWordsInfo { clear(); } - PrevWordsInfo(PrevWordsInfo &&prevWordsInfo) + PrevWordsInfo(const PrevWordsInfo &prevWordsInfo) : mPrevWordCount(prevWordsInfo.mPrevWordCount) { for (size_t i = 0; i < mPrevWordCount; ++i) { mPrevWordCodePointCount[i] = prevWordsInfo.mPrevWordCodePointCount[i]; @@ -73,6 +73,16 @@ class PrevWordsInfo { mIsBeginningOfSentence[0] = isBeginningOfSentence; } + size_t getPrevWordCount() const { + return mPrevWordCount; + } + + // TODO: Remove. + const PrevWordsInfo getTrimmedPrevWordsInfo(const size_t maxPrevWordCount) const { + return PrevWordsInfo(mPrevWordCodePoints, mPrevWordCodePointCount, mIsBeginningOfSentence, + std::min(mPrevWordCount, maxPrevWordCount)); + } + bool isValid() const { if (mPrevWordCodePointCount[0] > 0) { return true; @@ -112,7 +122,7 @@ class PrevWordsInfo { } private: - DISALLOW_COPY_AND_ASSIGN(PrevWordsInfo); + DISALLOW_ASSIGNMENT_OPERATOR(PrevWordsInfo); static int getWordId(const DictionaryStructureWithBufferPolicy *const dictStructurePolicy, const int *const wordCodePoints, const int wordCodePointCount, diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/bigram/ver4_bigram_list_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/bigram/ver4_bigram_list_policy.cpp index c6397706d..bc0f47f79 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/bigram/ver4_bigram_list_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/bigram/ver4_bigram_list_policy.cpp @@ -267,8 +267,7 @@ const BigramEntry Ver4BigramListPolicy::createUpdatedBigramEntryFrom( const NgramProperty *const ngramProperty) const { // TODO: Consolidate historical info and probability. if (mHeaderPolicy->hasHistoricalInfoOfWords()) { - const HistoricalInfo historicalInfoForUpdate(ngramProperty->getTimestamp(), - ngramProperty->getLevel(), ngramProperty->getCount()); + const HistoricalInfo &historicalInfoForUpdate = ngramProperty->getHistoricalInfo(); const HistoricalInfo updatedHistoricalInfo = ForgettingCurveUtils::createUpdatedHistoricalInfo( originalBigramEntry->getHistoricalInfo(), ngramProperty->getProbability(), diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_dict_content.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_dict_content.cpp index e2dd93c5e..9e1adff70 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_dict_content.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_dict_content.cpp @@ -83,10 +83,10 @@ bool BigramDictContent::writeBigramEntryAndAdvancePosition( } if (mHasHistoricalInfo) { const HistoricalInfo *const historicalInfo = bigramEntryToWrite->getHistoricalInfo(); - if (!bigramListBuffer->writeUintAndAdvancePosition(historicalInfo->getTimeStamp(), + if (!bigramListBuffer->writeUintAndAdvancePosition(historicalInfo->getTimestamp(), Ver4DictConstants::TIME_STAMP_FIELD_SIZE, entryWritingPos)) { AKLOGE("Cannot write bigram timestamps. pos: %d, timestamp: %d", *entryWritingPos, - historicalInfo->getTimeStamp()); + historicalInfo->getTimestamp()); return false; } if (!bigramListBuffer->writeUintAndAdvancePosition(historicalInfo->getLevel(), diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_entry.h b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_entry.h index 40968b4d8..480095a2f 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_entry.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/bigram_entry.h @@ -25,8 +25,8 @@ #define LATINIME_BACKWARD_V402_BIGRAM_ENTRY_H #include "defines.h" +#include "suggest/core/dictionary/property/historical_info.h" #include "suggest/policyimpl/dictionary/structure/backward/v402/ver4_dict_constants.h" -#include "suggest/policyimpl/dictionary/utils/historical_info.h" namespace latinime { namespace backward { diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_dict_content.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_dict_content.cpp index c671647d4..4a740d47b 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_dict_content.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_dict_content.cpp @@ -147,7 +147,7 @@ bool ProbabilityDictContent::writeEntry(const ProbabilityEntry *const probabilit } if (mHasHistoricalInfo) { const HistoricalInfo *const historicalInfo = probabilityEntry->getHistoricalInfo(); - if (!bufferToWrite->writeUintAndAdvancePosition(historicalInfo->getTimeStamp(), + if (!bufferToWrite->writeUintAndAdvancePosition(historicalInfo->getTimestamp(), Ver4DictConstants::TIME_STAMP_FIELD_SIZE, &writingPos)) { AKLOGE("Cannot write timestamp in probability dict content. pos: %d", writingPos); return false; diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_entry.h b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_entry.h index 8ccfa33dc..4111a49c0 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_entry.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/content/probability_entry.h @@ -25,8 +25,8 @@ #define LATINIME_BACKWARD_V402_PROBABILITY_ENTRY_H #include "defines.h" +#include "suggest/core/dictionary/property/historical_info.h" #include "suggest/policyimpl/dictionary/structure/backward/v402/ver4_dict_constants.h" -#include "suggest/policyimpl/dictionary/utils/historical_info.h" namespace latinime { namespace backward { diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_node_writer.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_node_writer.cpp index 6827c3ee3..8d169743c 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_node_writer.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_node_writer.cpp @@ -396,8 +396,7 @@ const ProbabilityEntry Ver4PatriciaTrieNodeWriter::createUpdatedEntryFrom( const UnigramProperty *const unigramProperty) const { // TODO: Consolidate historical info and probability. if (mHeaderPolicy->hasHistoricalInfoOfWords()) { - const HistoricalInfo historicalInfoForUpdate(unigramProperty->getTimestamp(), - unigramProperty->getLevel(), unigramProperty->getCount()); + const HistoricalInfo &historicalInfoForUpdate = unigramProperty->getHistoricalInfo(); const HistoricalInfo updatedHistoricalInfo = ForgettingCurveUtils::createUpdatedHistoricalInfo( originalProbabilityEntry->getHistoricalInfo(), diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp index 9631cf1ae..11f7b305f 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp @@ -52,6 +52,7 @@ const char *const Ver4PatriciaTriePolicy::MAX_BIGRAM_COUNT_QUERY = "MAX_BIGRAM_C const int Ver4PatriciaTriePolicy::MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS = 1024; const int Ver4PatriciaTriePolicy::MIN_DICT_SIZE_TO_REFUSE_DYNAMIC_OPERATIONS = Ver4DictConstants::MAX_DICTIONARY_SIZE - MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS; +const int Ver4PatriciaTriePolicy::DUMMY_PROBABILITY_FOR_VALID_WORDS = 1; void Ver4PatriciaTriePolicy::createAndGetAllChildDicNodes(const DicNode *const dicNode, DicNodeVector *const childDicNodes) const { @@ -339,11 +340,9 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI } if (prevWordIds[0] == NOT_A_WORD_ID) { if (prevWordsInfo->isNthPrevWordBeginningOfSentence(1 /* n */)) { - const std::vector<UnigramProperty::ShortcutProperty> shortcuts; const UnigramProperty beginningOfSentenceUnigramProperty( true /* representsBeginningOfSentence */, true /* isNotAWord */, - false /* isBlacklisted */, MAX_PROBABILITY /* probability */, - NOT_A_TIMESTAMP /* timestamp */, 0 /* level */, 0 /* count */, &shortcuts); + false /* isBlacklisted */, MAX_PROBABILITY /* probability */, HistoricalInfo()); if (!addUnigramEntry(prevWordsInfo->getNthPrevWordCodePoints(1 /* n */), &beginningOfSentenceUnigramProperty)) { AKLOGE("Cannot add unigram entry for the beginning-of-sentence."); @@ -414,6 +413,29 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor } } + +bool Ver4PatriciaTriePolicy::updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo) { + if (!mBuffers->isUpdatable()) { + AKLOGI("Warning: updateCounter() is called for non-updatable dictionary."); + return false; + } + const int probability = isValidWord ? DUMMY_PROBABILITY_FOR_VALID_WORDS : NOT_A_PROBABILITY; + const UnigramProperty unigramProperty(false /* representsBeginningOfSentence */, + false /* isNotAWord */, false /*isBlacklisted*/, probability, historicalInfo); + if (!addUnigramEntry(wordCodePoints, &unigramProperty)) { + AKLOGE("Cannot update unigarm entry in updateCounter()."); + return false; + } + const NgramProperty ngramProperty(wordCodePoints.toVector(), probability, historicalInfo); + if (!addNgramEntry(prevWordsInfo, &ngramProperty)) { + AKLOGE("Cannot update unigarm entry in updateCounter()."); + return false; + } + return true; +} + bool Ver4PatriciaTriePolicy::flush(const char *const filePath) { if (!mBuffers->isUpdatable()) { AKLOGI("Warning: flush() is called for non-updatable dictionary. filePath: %s", filePath); @@ -528,8 +550,7 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty( bigramEntry.getProbability(); ngrams.emplace_back( CodePointArrayView(bigramWord1CodePoints, codePointCount).toVector(), - probability, historicalInfo->getTimeStamp(), historicalInfo->getLevel(), - historicalInfo->getCount()); + probability, *historicalInfo); } } // Fetch shortcut information. @@ -552,8 +573,7 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty( } const UnigramProperty unigramProperty(ptNodeParams.representsBeginningOfSentence(), ptNodeParams.isNotAWord(), ptNodeParams.isBlacklisted(), ptNodeParams.getProbability(), - historicalInfo->getTimeStamp(), historicalInfo->getLevel(), - historicalInfo->getCount(), &shortcuts); + *historicalInfo, std::move(shortcuts)); return WordProperty(wordCodePoints.toVector(), &unigramProperty, &ngrams); } diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h index 324a53e62..995d7764f 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h @@ -118,6 +118,10 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy { bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView wordCodePoints); + bool updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo); + bool flush(const char *const filePath); bool flushWithGC(const char *const filePath); @@ -147,6 +151,7 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy { // prevent the dictionary from overflowing. static const int MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS; static const int MIN_DICT_SIZE_TO_REFUSE_DYNAMIC_OPERATIONS; + static const int DUMMY_PROBABILITY_FOR_VALID_WORDS; const Ver4DictBuffers::Ver4DictBuffersPtr mBuffers; const HeaderPolicy *const mHeaderPolicy; diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_writing_helper.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_writing_helper.cpp index 3fb4caa08..2887dc6b1 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_writing_helper.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_writing_helper.cpp @@ -216,7 +216,7 @@ bool Ver4PatriciaTrieWritingHelper::truncateUnigrams( probabilityEntry.getHistoricalInfo(), mBuffers->getHeaderPolicy()) : probabilityEntry.getProbability(); priorityQueue.push(DictProbability(terminalPos, probability, - probabilityEntry.getHistoricalInfo()->getTimeStamp())); + probabilityEntry.getHistoricalInfo()->getTimestamp())); } // Delete unigrams. @@ -263,7 +263,7 @@ bool Ver4PatriciaTrieWritingHelper::truncateBigrams(const int maxBigramCount) { bigramEntry.getHistoricalInfo(), mBuffers->getHeaderPolicy()) : bigramEntry.getProbability(); priorityQueue.push(DictProbability(entryPos, probability, - bigramEntry.getHistoricalInfo()->getTimeStamp())); + bigramEntry.getHistoricalInfo()->getTimestamp())); } } diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.cpp index 49095d0e9..d3d684bfa 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.cpp @@ -452,7 +452,7 @@ const WordProperty PatriciaTriePolicy::getWordProperty( const int probability = getProbability(word1Probability, bigramsIt.getProbability()); ngrams.emplace_back( CodePointArrayView(bigramWord1CodePoints, word1CodePointCount).toVector(), - probability, NOT_A_TIMESTAMP /* timestamp */, 0 /* level */, 0 /* count */); + probability, HistoricalInfo()); } } // Fetch shortcut information. @@ -477,7 +477,7 @@ const WordProperty PatriciaTriePolicy::getWordProperty( } const UnigramProperty unigramProperty(ptNodeParams.representsBeginningOfSentence(), ptNodeParams.isNotAWord(), ptNodeParams.isBlacklisted(), ptNodeParams.getProbability(), - NOT_A_TIMESTAMP /* timestamp */, 0 /* level */, 0 /* count */, &shortcuts); + HistoricalInfo(), std::move(shortcuts)); return WordProperty(wordCodePoints.toVector(), &unigramProperty, &ngrams); } diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h b/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h index a912d03be..32a95bb6c 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h @@ -107,6 +107,14 @@ class PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy { return false; } + bool updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo) { + // This method should not be called for non-updatable dictionary. + AKLOGI("Warning: updateCounter() is called for non-updatable dictionary."); + return false; + } + bool flush(const char *const filePath) { // This method should not be called for non-updatable dictionary. AKLOGI("Warning: flush() is called for non-updatable dictionary."); diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.cpp index 139230228..956dabb4f 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.cpp @@ -296,7 +296,7 @@ bool LanguageModelDictContent::getEntryInfo(const HeaderPolicy *const headerPoli ForgettingCurveUtils::decodeProbability(probabilityEntry.getHistoricalInfo(), headerPolicy) : probabilityEntry.getProbability(); outEntryInfo->emplace_back(probability, - probabilityEntry.getHistoricalInfo()->getTimeStamp(), + probabilityEntry.getHistoricalInfo()->getTimestamp(), entry.key(), targetLevel, prevWordIds->data()); } return true; diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h index fd52e574c..fa1415633 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h @@ -21,10 +21,10 @@ #include <cstdint> #include "defines.h" +#include "suggest/core/dictionary/property/historical_info.h" #include "suggest/core/dictionary/property/ngram_property.h" #include "suggest/core/dictionary/property/unigram_property.h" #include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.h" -#include "suggest/policyimpl/dictionary/utils/historical_info.h" namespace latinime { @@ -53,15 +53,13 @@ class ProbabilityEntry { unigramProperty->isNotAWord(), unigramProperty->isBlacklisted(), unigramProperty->isPossiblyOffensive())), mProbability(unigramProperty->getProbability()), - mHistoricalInfo(unigramProperty->getTimestamp(), unigramProperty->getLevel(), - unigramProperty->getCount()) {} + mHistoricalInfo(unigramProperty->getHistoricalInfo()) {} // Create from ngram property. // TODO: Set flags. ProbabilityEntry(const NgramProperty *const ngramProperty) : mFlags(0), mProbability(ngramProperty->getProbability()), - mHistoricalInfo(ngramProperty->getTimestamp(), ngramProperty->getLevel(), - ngramProperty->getCount()) {} + mHistoricalInfo(ngramProperty->getHistoricalInfo()) {} bool isValid() const { return (mFlags & Ver4DictConstants::FLAG_NOT_A_VALID_ENTRY) == 0; @@ -103,7 +101,7 @@ class ProbabilityEntry { uint64_t encodedEntry = static_cast<uint64_t>(mFlags); if (hasHistoricalInfo) { encodedEntry = (encodedEntry << (Ver4DictConstants::TIME_STAMP_FIELD_SIZE * CHAR_BIT)) - ^ static_cast<uint64_t>(mHistoricalInfo.getTimeStamp()); + ^ static_cast<uint64_t>(mHistoricalInfo.getTimestamp()); encodedEntry = (encodedEntry << (Ver4DictConstants::WORD_LEVEL_FIELD_SIZE * CHAR_BIT)) ^ static_cast<uint64_t>(mHistoricalInfo.getLevel()); encodedEntry = (encodedEntry << (Ver4DictConstants::WORD_COUNT_FIELD_SIZE * CHAR_BIT)) diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp index d46acf594..41b109f95 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp @@ -43,6 +43,7 @@ const char *const Ver4PatriciaTriePolicy::MAX_BIGRAM_COUNT_QUERY = "MAX_BIGRAM_C const int Ver4PatriciaTriePolicy::MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS = 1024; const int Ver4PatriciaTriePolicy::MIN_DICT_SIZE_TO_REFUSE_DYNAMIC_OPERATIONS = Ver4DictConstants::MAX_DICTIONARY_SIZE - MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS; +const int Ver4PatriciaTriePolicy::DUMMY_PROBABILITY_FOR_VALID_WORDS = 1; void Ver4PatriciaTriePolicy::createAndGetAllChildDicNodes(const DicNode *const dicNode, DicNodeVector *const childDicNodes) const { @@ -298,11 +299,9 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI if (!prevWordsInfo->isNthPrevWordBeginningOfSentence(i + 1 /* n */)) { return false; } - const std::vector<UnigramProperty::ShortcutProperty> shortcuts; const UnigramProperty beginningOfSentenceUnigramProperty( true /* representsBeginningOfSentence */, true /* isNotAWord */, - false /* isBlacklisted */, MAX_PROBABILITY /* probability */, - NOT_A_TIMESTAMP /* timestamp */, 0 /* level */, 0 /* count */, &shortcuts); + false /* isBlacklisted */, MAX_PROBABILITY /* probability */, HistoricalInfo()); if (!addUnigramEntry(prevWordsInfo->getNthPrevWordCodePoints(1 /* n */), &beginningOfSentenceUnigramProperty)) { AKLOGE("Cannot add unigram entry for the beginning-of-sentence."); @@ -364,6 +363,32 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor } } +bool Ver4PatriciaTriePolicy::updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo) { + if (!mBuffers->isUpdatable()) { + AKLOGI("Warning: updateCounter() is called for non-updatable dictionary."); + return false; + } + // TODO: Have count up method in language model dict content. + const int probability = isValidWord ? DUMMY_PROBABILITY_FOR_VALID_WORDS : NOT_A_PROBABILITY; + const UnigramProperty unigramProperty(false /* representsBeginningOfSentence */, + false /* isNotAWord */, false /*isBlacklisted*/, probability, historicalInfo); + if (!addUnigramEntry(wordCodePoints, &unigramProperty)) { + AKLOGE("Cannot update unigarm entry in updateCounter()."); + return false; + } + const NgramProperty ngramProperty(wordCodePoints.toVector(), probability, historicalInfo); + for (size_t i = 1; i <= prevWordsInfo->getPrevWordCount(); ++i) { + const PrevWordsInfo trimmedPrevWordsInfo(prevWordsInfo->getTrimmedPrevWordsInfo(i)); + if (!addNgramEntry(&trimmedPrevWordsInfo, &ngramProperty)) { + AKLOGE("Cannot update ngram entry in updateCounter()."); + return false; + } + } + return true; +} + bool Ver4PatriciaTriePolicy::flush(const char *const filePath) { if (!mBuffers->isUpdatable()) { AKLOGI("Warning: flush() is called for non-updatable dictionary. filePath: %s", filePath); @@ -464,8 +489,7 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty( ForgettingCurveUtils::decodeProbability(historicalInfo, mHeaderPolicy) : probabilityEntry.getProbability(); ngrams.emplace_back(CodePointArrayView(bigramWord1CodePoints, codePointCount).toVector(), - probability, historicalInfo->getTimeStamp(), historicalInfo->getLevel(), - historicalInfo->getCount()); + probability, *historicalInfo); } // Fetch shortcut information. std::vector<UnigramProperty::ShortcutProperty> shortcuts; @@ -487,8 +511,7 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty( } const UnigramProperty unigramProperty(probabilityEntry.representsBeginningOfSentence(), probabilityEntry.isNotAWord(), probabilityEntry.isBlacklisted(), - probabilityEntry.getProbability(), historicalInfo->getTimeStamp(), - historicalInfo->getLevel(), historicalInfo->getCount(), &shortcuts); + probabilityEntry.getProbability(), *historicalInfo, std::move(shortcuts)); return WordProperty(wordCodePoints.toVector(), &unigramProperty, &ngrams); } diff --git a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h index 598122bf2..662bb8d4b 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h +++ b/native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h @@ -98,6 +98,10 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy { bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView wordCodePoints); + bool updateCounter(const PrevWordsInfo *const prevWordsInfo, + const CodePointArrayView wordCodePoints, const bool isValidWord, + const HistoricalInfo historicalInfo); + bool flush(const char *const filePath); bool flushWithGC(const char *const filePath); @@ -127,6 +131,8 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy { // prevent the dictionary from overflowing. static const int MARGIN_TO_REFUSE_DYNAMIC_OPERATIONS; static const int MIN_DICT_SIZE_TO_REFUSE_DYNAMIC_OPERATIONS; + // TODO: Remove + static const int DUMMY_PROBABILITY_FOR_VALID_WORDS; const Ver4DictBuffers::Ver4DictBuffersPtr mBuffers; const HeaderPolicy *const mHeaderPolicy; diff --git a/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.cpp b/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.cpp index fed0ae77e..af4bc186a 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.cpp @@ -43,7 +43,7 @@ const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityT /* static */ const HistoricalInfo ForgettingCurveUtils::createUpdatedHistoricalInfo( const HistoricalInfo *const originalHistoricalInfo, const int newProbability, const HistoricalInfo *const newHistoricalInfo, const HeaderPolicy *const headerPolicy) { - const int timestamp = newHistoricalInfo->getTimeStamp(); + const int timestamp = newHistoricalInfo->getTimestamp(); if (newProbability != NOT_A_PROBABILITY && originalHistoricalInfo->getLevel() == 0) { // Add entry as a valid word. const int level = clampToVisibleEntryLevelRange(newHistoricalInfo->getLevel()); @@ -78,7 +78,7 @@ const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityT /* static */ int ForgettingCurveUtils::decodeProbability( const HistoricalInfo *const historicalInfo, const HeaderPolicy *const headerPolicy) { - const int elapsedTimeStepCount = getElapsedTimeStepCount(historicalInfo->getTimeStamp(), + const int elapsedTimeStepCount = getElapsedTimeStepCount(historicalInfo->getTimestamp(), headerPolicy->getForgettingCurveDurationToLevelDown()); return sProbabilityTable.getProbability( headerPolicy->getForgettingCurveProbabilityValuesTableId(), @@ -102,7 +102,7 @@ const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityT /* static */ bool ForgettingCurveUtils::needsToKeep(const HistoricalInfo *const historicalInfo, const HeaderPolicy *const headerPolicy) { return historicalInfo->getLevel() > 0 - || getElapsedTimeStepCount(historicalInfo->getTimeStamp(), + || getElapsedTimeStepCount(historicalInfo->getTimestamp(), headerPolicy->getForgettingCurveDurationToLevelDown()) < DISCARD_LEVEL_ZERO_ENTRY_TIME_STEP_COUNT_THRESHOLD; } @@ -110,12 +110,12 @@ const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityT /* static */ const HistoricalInfo ForgettingCurveUtils::createHistoricalInfoToSave( const HistoricalInfo *const originalHistoricalInfo, const HeaderPolicy *const headerPolicy) { - if (originalHistoricalInfo->getTimeStamp() == NOT_A_TIMESTAMP) { + if (originalHistoricalInfo->getTimestamp() == NOT_A_TIMESTAMP) { return HistoricalInfo(); } const int durationToLevelDownInSeconds = headerPolicy->getForgettingCurveDurationToLevelDown(); const int elapsedTimeStep = getElapsedTimeStepCount( - originalHistoricalInfo->getTimeStamp(), durationToLevelDownInSeconds); + originalHistoricalInfo->getTimestamp(), durationToLevelDownInSeconds); if (elapsedTimeStep <= MAX_ELAPSED_TIME_STEP_COUNT) { // No need to update historical info. return *originalHistoricalInfo; @@ -124,7 +124,7 @@ const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityT const int maxLevelDownAmonut = elapsedTimeStep / (MAX_ELAPSED_TIME_STEP_COUNT + 1); const int levelDownAmount = (maxLevelDownAmonut >= originalHistoricalInfo->getLevel()) ? originalHistoricalInfo->getLevel() : maxLevelDownAmonut; - const int adjustedTimestampInSeconds = originalHistoricalInfo->getTimeStamp() + + const int adjustedTimestampInSeconds = originalHistoricalInfo->getTimestamp() + levelDownAmount * durationToLevelDownInSeconds; return HistoricalInfo(adjustedTimestampInSeconds, originalHistoricalInfo->getLevel() - levelDownAmount, 0 /* count */); diff --git a/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.h b/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.h index 313eb6b64..10abb405a 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.h +++ b/native/jni/src/suggest/policyimpl/dictionary/utils/forgetting_curve_utils.h @@ -20,7 +20,7 @@ #include <vector> #include "defines.h" -#include "suggest/policyimpl/dictionary/utils/historical_info.h" +#include "suggest/core/dictionary/property/historical_info.h" namespace latinime { |