diff options
Diffstat (limited to 'native/jni/src')
7 files changed, 61 insertions, 10 deletions
diff --git a/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp b/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp index 425b07624..5ba71c168 100644 --- a/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp +++ b/native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp @@ -150,24 +150,26 @@ int BigramDictionary::getBigramListPositionForWord(const int *prevWord, const in return mDictionaryStructurePolicy->getBigramsPositionOfNode(pos); } -bool BigramDictionary::isValidBigram(const int *word0, int length0, const int *word1, +int BigramDictionary::getBigramProbability(const int *word0, int length0, const int *word1, int length1) const { int pos = getBigramListPositionForWord(word0, length0, false /* forceLowerCaseSearch */); // getBigramListPositionForWord returns 0 if this word isn't in the dictionary or has no bigrams - if (NOT_A_DICT_POS == pos) return false; + if (NOT_A_DICT_POS == pos) return NOT_A_PROBABILITY; int nextWordPos = mDictionaryStructurePolicy->getTerminalNodePositionOfWord(word1, length1, false /* forceLowerCaseSearch */); - if (NOT_A_DICT_POS == nextWordPos) return false; + if (NOT_A_DICT_POS == nextWordPos) return NOT_A_PROBABILITY; BinaryDictionaryBigramsIterator bigramsIt( mDictionaryStructurePolicy->getBigramsStructurePolicy(), pos); while (bigramsIt.hasNext()) { bigramsIt.next(); if (bigramsIt.getBigramPos() == nextWordPos) { - return true; + return mDictionaryStructurePolicy->getProbability( + mDictionaryStructurePolicy->getUnigramProbabilityOfPtNode(nextWordPos), + bigramsIt.getProbability()); } } - return false; + return NOT_A_PROBABILITY; } // TODO: Move functions related to bigram to here diff --git a/native/jni/src/suggest/core/dictionary/bigram_dictionary.h b/native/jni/src/suggest/core/dictionary/bigram_dictionary.h index 99b964c49..8af7ee75d 100644 --- a/native/jni/src/suggest/core/dictionary/bigram_dictionary.h +++ b/native/jni/src/suggest/core/dictionary/bigram_dictionary.h @@ -29,7 +29,7 @@ class BigramDictionary { int getPredictions(const int *word, int length, int *outBigramCodePoints, int *outBigramProbability, int *outputTypes) const; - bool isValidBigram(const int *word1, int length1, const int *word2, int length2) const; + int getBigramProbability(const int *word1, int length1, const int *word2, int length2) const; ~BigramDictionary(); private: diff --git a/native/jni/src/suggest/core/dictionary/dictionary.cpp b/native/jni/src/suggest/core/dictionary/dictionary.cpp index 033572201..ec1b63a12 100644 --- a/native/jni/src/suggest/core/dictionary/dictionary.cpp +++ b/native/jni/src/suggest/core/dictionary/dictionary.cpp @@ -93,8 +93,9 @@ int Dictionary::getProbability(const int *word, int length) const { return getDictionaryStructurePolicy()->getUnigramProbabilityOfPtNode(pos); } -bool Dictionary::isValidBigram(const int *word0, int length0, const int *word1, int length1) const { - return mBigramDictionary->isValidBigram(word0, length0, word1, length1); +int Dictionary::getBigramProbability(const int *word0, int length0, const int *word1, + int length1) const { + return mBigramDictionary->getBigramProbability(word0, length0, word1, length1); } void Dictionary::addUnigramWord(const int *const word, const int length, const int probability) { diff --git a/native/jni/src/suggest/core/dictionary/dictionary.h b/native/jni/src/suggest/core/dictionary/dictionary.h index 06e84bbfe..974447468 100644 --- a/native/jni/src/suggest/core/dictionary/dictionary.h +++ b/native/jni/src/suggest/core/dictionary/dictionary.h @@ -67,7 +67,7 @@ class Dictionary { int getProbability(const int *word, int length) const; - bool isValidBigram(const int *word0, int length0, const int *word1, int length1) const; + int getBigramProbability(const int *word0, int length0, const int *word1, int length1) const; void addUnigramWord(const int *const word, const int length, const int probability); diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp index 3cfbfd85b..2198a13c9 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_policy.cpp @@ -248,7 +248,9 @@ void DynamicPatriciaTriePolicy::flush(const char *const filePath) { AKLOGI("Warning: flush() is called for non-updatable dictionary."); return; } - // TODO: Implement. + DynamicPatriciaTrieWritingHelper writingHelper(&mBufferWithExtendableBuffer, + &mBigramListPolicy, &mShortcutListPolicy); + writingHelper.writeToDictFile(filePath, mBuffer->getBuffer(), mHeaderPolicy.getSize()); } void DynamicPatriciaTriePolicy::flushWithGC(const char *const filePath) { diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp index 311d31e5d..a67c0d94a 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp +++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.cpp @@ -16,6 +16,9 @@ #include "suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h" +#include <cstdio> +#include <cstring> + #include "suggest/policyimpl/dictionary/bigram/dynamic_bigram_list_policy.h" #include "suggest/policyimpl/dictionary/dynamic_patricia_trie_node_reader.h" #include "suggest/policyimpl/dictionary/dynamic_patricia_trie_reading_helper.h" @@ -27,6 +30,8 @@ namespace latinime { const int DynamicPatriciaTrieWritingHelper::CHILDREN_POSITION_FIELD_SIZE = 3; +const char *const DynamicPatriciaTrieWritingHelper::TEMP_FILE_SUFFIX_FOR_WRITING_DICT_FILE = + ".tmp"; bool DynamicPatriciaTrieWritingHelper::addUnigramWord( DynamicPatriciaTrieReadingHelper *const readingHelper, @@ -131,6 +136,41 @@ bool DynamicPatriciaTrieWritingHelper::removeBigramWords(const int word0Pos, con return mBigramPolicy->removeBigram(nodeReader.getBigramsPos(), word1Pos); } +void DynamicPatriciaTrieWritingHelper::writeToDictFile(const char *const fileName, + const uint8_t *const headerBuf, const int headerSize) { + const int tmpFileNameBufSize = strlen(fileName) + + strlen(TEMP_FILE_SUFFIX_FOR_WRITING_DICT_FILE) + 1; + char tmpFileName[tmpFileNameBufSize]; + snprintf(tmpFileName, tmpFileNameBufSize, "%s%s", fileName, + TEMP_FILE_SUFFIX_FOR_WRITING_DICT_FILE); + FILE *const file = fopen(tmpFileName, "wb"); + if (!file) { + return; + } + // Write header. + if (fwrite(headerBuf, headerSize, 1, file) < 1) { + fclose(file); + remove(tmpFileName); + return; + } + // Write data in original buffer. + if (fwrite(mBuffer->getBuffer(false /* usesAdditionalBuffer */), + mBuffer->getOriginalBufferSize(), 1, file) < 1) { + fclose(file); + remove(tmpFileName); + return; + } + // Write data in additional buffer. + if (fwrite(mBuffer->getBuffer(true /* usesAdditionalBuffer */), + mBuffer->getTailPosition() - mBuffer->getOriginalBufferSize(), 1, file) < 1) { + fclose(file); + remove(tmpFileName); + return; + } + fclose(file); + rename(tmpFileName, fileName); +} + bool DynamicPatriciaTrieWritingHelper::markNodeAsMovedAndSetPosition( const DynamicPatriciaTrieNodeReader *const originalNode, const int movedPos, const int bigramLinkedNodePos) { diff --git a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h index 20e35abcf..faf7a4e1b 100644 --- a/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h +++ b/native/jni/src/suggest/policyimpl/dictionary/dynamic_patricia_trie_writing_helper.h @@ -17,6 +17,8 @@ #ifndef LATINIME_DYNAMIC_PATRICIA_TRIE_WRITING_HELPER_H #define LATINIME_DYNAMIC_PATRICIA_TRIE_WRITING_HELPER_H +#include <stdint.h> + #include "defines.h" namespace latinime { @@ -46,10 +48,14 @@ class DynamicPatriciaTrieWritingHelper { // Remove a bigram relation from word0Pos to word1Pos. bool removeBigramWords(const int word0Pos, const int word1Pos); + void writeToDictFile(const char *const fileName, const uint8_t *const headerBuf, + const int headerSize); + private: DISALLOW_IMPLICIT_CONSTRUCTORS(DynamicPatriciaTrieWritingHelper); static const int CHILDREN_POSITION_FIELD_SIZE; + static const char *const TEMP_FILE_SUFFIX_FOR_WRITING_DICT_FILE; BufferWithExtendableBuffer *const mBuffer; DynamicBigramListPolicy *const mBigramPolicy; |