Merge "Move word flags to language model dict content." (c912957f) · Commits · e / os / android_packages_inputmethods_LatinIME

native/jni/src/suggest/core/dictionary/property/unigram_property.h

+5 −0

Original line number	Diff line number	Diff line
		@@ -71,6 +71,11 @@ class UnigramProperty {
		return mIsBlacklisted;
		}

		bool isPossiblyOffensive() const {
		// TODO: Have dedicated flag.
		return mProbability == 0;
		}

		bool hasShortcuts() const {
		return !mShortcuts.empty();
		}

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.cpp

+12 −5

Original line number	Diff line number	Diff line
		@@ -38,7 +38,7 @@ bool LanguageModelDictContent::runGC(
		0 /* nextLevelBitmapEntryIndex */, outNgramCount);
		}

		int LanguageModelDictContent::getWordProbability(const WordIdArrayView prevWordIds,
		const WordAttributes LanguageModelDictContent::getWordAttributes(const WordIdArrayView prevWordIds,
		const int wordId, const HeaderPolicy *const headerPolicy) const {
		int bitmapEntryIndices[MAX_PREV_WORD_COUNT_FOR_N_GRAM + 1];
		bitmapEntryIndices[0] = mTrieMap.getRootBitmapEntryIndex();
		@@ -60,17 +60,24 @@ int LanguageModelDictContent::getWordProbability(const WordIdArrayView prevWordI
		}
		const ProbabilityEntry probabilityEntry =
		ProbabilityEntry::decode(result.mValue, mHasHistoricalInfo);
		int probability = NOT_A_PROBABILITY;
		if (mHasHistoricalInfo) {
		const int probability = ForgettingCurveUtils::decodeProbability(
		const int rawProbability = ForgettingCurveUtils::decodeProbability(
		probabilityEntry.getHistoricalInfo(), headerPolicy)
		+ ForgettingCurveUtils::getProbabilityBiasForNgram(i + 1 /* n */);
		return std::min(probability, MAX_PROBABILITY);
		probability = std::min(rawProbability, MAX_PROBABILITY);
		} else {
		return probabilityEntry.getProbability();
		probability = probabilityEntry.getProbability();
		}
		// TODO: Some flags in unigramProbabilityEntry should be overwritten by flags in
		// probabilityEntry.
		const ProbabilityEntry unigramProbabilityEntry = getProbabilityEntry(wordId);
		return WordAttributes(probability, unigramProbabilityEntry.isNotAWord(),
		unigramProbabilityEntry.isBlacklisted(),
		unigramProbabilityEntry.isPossiblyOffensive());
		}
		// Cannot find the word.
		return NOT_A_PROBABILITY;
		return WordAttributes();
		}

		ProbabilityEntry LanguageModelDictContent::getNgramProbabilityEntry(

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/language_model_dict_content.h

+2 −1

Original line number	Diff line number	Diff line
		@@ -21,6 +21,7 @@
		#include <vector>

		#include "defines.h"
		#include "suggest/core/dictionary/word_attributes.h"
		#include "suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h"
		#include "suggest/policyimpl/dictionary/structure/v4/content/terminal_position_lookup_table.h"
		#include "suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.h"
		@@ -128,7 +129,7 @@ class LanguageModelDictContent {
		const LanguageModelDictContent *const originalContent,
		int *const outNgramCount);

		int getWordProbability(const WordIdArrayView prevWordIds, const int wordId,
		const WordAttributes getWordAttributes(const WordIdArrayView prevWordIds, const int wordId,
		const HeaderPolicy *const headerPolicy) const;

		ProbabilityEntry getProbabilityEntry(const int wordId) const {

native/jni/src/suggest/policyimpl/dictionary/structure/v4/content/probability_entry.h

+27 −3

Original line number	Diff line number	Diff line
		@@ -49,7 +49,9 @@ class ProbabilityEntry {

		// Create from unigram property.
		ProbabilityEntry(const UnigramProperty *const unigramProperty)
		: mFlags(createFlags(unigramProperty->representsBeginningOfSentence())),
		: mFlags(createFlags(unigramProperty->representsBeginningOfSentence(),
		unigramProperty->isNotAWord(), unigramProperty->isBlacklisted(),
		unigramProperty->isPossiblyOffensive())),
		mProbability(unigramProperty->getProbability()),
		mHistoricalInfo(unigramProperty->getTimestamp(), unigramProperty->getLevel(),
		unigramProperty->getCount()) {}
		@@ -85,6 +87,18 @@ class ProbabilityEntry {
		return (mFlags & Ver4DictConstants::FLAG_REPRESENTS_BEGINNING_OF_SENTENCE) != 0;
		}

		bool isNotAWord() const {
		return (mFlags & Ver4DictConstants::FLAG_NOT_A_WORD) != 0;
		}

		bool isBlacklisted() const {
		return (mFlags & Ver4DictConstants::FLAG_BLACKLISTED) != 0;
		}

		bool isPossiblyOffensive() const {
		return (mFlags & Ver4DictConstants::FLAG_POSSIBLY_OFFENSIVE) != 0;
		}

		uint64_t encode(const bool hasHistoricalInfo) const {
		uint64_t encodedEntry = static_cast<uint64_t>(mFlags);
		if (hasHistoricalInfo) {
		@@ -142,10 +156,20 @@ class ProbabilityEntry {
		(encodedEntry >> (pos * CHAR_BIT)) & ((1ull << (size * CHAR_BIT)) - 1));
		}

		static uint8_t createFlags(const bool representsBeginningOfSentence) {
		static uint8_t createFlags(const bool representsBeginningOfSentence,
		const bool isNotAWord, const bool isBlacklisted, const bool isPossiblyOffensive) {
		uint8_t flags = 0;
		if (representsBeginningOfSentence) {
		flags ^= Ver4DictConstants::FLAG_REPRESENTS_BEGINNING_OF_SENTENCE;
		flags \|= Ver4DictConstants::FLAG_REPRESENTS_BEGINNING_OF_SENTENCE;
		}
		if (isNotAWord) {
		flags \|= Ver4DictConstants::FLAG_NOT_A_WORD;
		}
		if (isBlacklisted) {
		flags \|= Ver4DictConstants::FLAG_BLACKLISTED;
		}
		if (isPossiblyOffensive) {
		flags \|= Ver4DictConstants::FLAG_POSSIBLY_OFFENSIVE;
		}
		return flags;
		}

native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_dict_constants.cpp

+3 −0

Original line number	Diff line number	Diff line
		@@ -54,6 +54,9 @@ const int Ver4DictConstants::WORD_COUNT_FIELD_SIZE = 1;

		const uint8_t Ver4DictConstants::FLAG_REPRESENTS_BEGINNING_OF_SENTENCE = 0x1;
		const uint8_t Ver4DictConstants::FLAG_NOT_A_VALID_ENTRY = 0x2;
		const uint8_t Ver4DictConstants::FLAG_NOT_A_WORD = 0x4;
		const uint8_t Ver4DictConstants::FLAG_BLACKLISTED = 0x8;
		const uint8_t Ver4DictConstants::FLAG_POSSIBLY_OFFENSIVE = 0x10;

		const int Ver4DictConstants::SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE = 64;
		const int Ver4DictConstants::SHORTCUT_ADDRESS_TABLE_DATA_SIZE = 4;