nyancrimew/LatinIME
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

Merge "Rename PrevWordsInfo to NgramContext."

Browse source
This commit is contained in:
Keisuke Kuroyanagi 2014-10-09 12:47:15 +00:00 committed by Android (Google) Code Review
commit 10fa30e380
15 changed files with 114 additions and 114 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

30
native/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
View file

@ -242,15 +242,15 @@ static void latinime_BinaryDictionary_getSuggestions(JNIEnv *env, jclass clazz,
env->GetFloatArrayRegion(inOutWeightOfLangModelVsSpatialModel, 0, 1 /* len */,
&weightOfLangModelVsSpatialModel);
SuggestionResults suggestionResults(MAX_RESULTS);
const PrevWordsInfo prevWordsInfo = JniDataUtils::constructPrevWordsInfo(env,
const NgramContext ngramContext = JniDataUtils::constructNgramContext(env,
prevWordCodePointArrays, isBeginningOfSentenceArray, prevWordCount);
if (givenSuggestOptions.isGesture() || inputSize > 0) {
// TODO: Use SuggestionResults to return suggestions.
dictionary->getSuggestions(pInfo, traverseSession, xCoordinates, yCoordinates,
times, pointerIds, inputCodePoints, inputSize, &prevWordsInfo,
times, pointerIds, inputCodePoints, inputSize, &ngramContext,
&givenSuggestOptions, weightOfLangModelVsSpatialModel, &suggestionResults);
} else {
dictionary->getPredictions(&prevWordsInfo, &suggestionResults);
dictionary->getPredictions(&ngramContext, &suggestionResults);
}
if (DEBUG_DICT) {
suggestionResults.dumpSuggestions();
@ -289,10 +289,10 @@ static jint latinime_BinaryDictionary_getNgramProbability(JNIEnv *env, jclass cl
const jsize wordLength = env->GetArrayLength(word);
int wordCodePoints[wordLength];
env->GetIntArrayRegion(word, 0, wordLength, wordCodePoints);
const PrevWordsInfo prevWordsInfo = JniDataUtils::constructPrevWordsInfo(env,
const NgramContext ngramContext = JniDataUtils::constructNgramContext(env,
prevWordCodePointArrays, isBeginningOfSentenceArray,
env->GetArrayLength(prevWordCodePointArrays));
return dictionary->getNgramProbability(&prevWordsInfo,
return dictionary->getNgramProbability(&ngramContext,
CodePointArrayView(wordCodePoints, wordLength));
}
@ -402,7 +402,7 @@ static bool latinime_BinaryDictionary_addNgramEntry(JNIEnv *env, jclass clazz, j
if (!dictionary) {
return false;
}
const PrevWordsInfo prevWordsInfo = JniDataUtils::constructPrevWordsInfo(env,
const NgramContext ngramContext = JniDataUtils::constructNgramContext(env,
prevWordCodePointArrays, isBeginningOfSentenceArray,
env->GetArrayLength(prevWordCodePointArrays));
jsize wordLength = env->GetArrayLength(word);
@ -411,7 +411,7 @@ static bool latinime_BinaryDictionary_addNgramEntry(JNIEnv *env, jclass clazz, j
// Use 1 for count to indicate the ngram has inputted.
const NgramProperty ngramProperty(CodePointArrayView(wordCodePoints, wordLength).toVector(),
probability, HistoricalInfo(timestamp, 0 /* level */, 1 /* count */));
return dictionary->addNgramEntry(&prevWordsInfo, &ngramProperty);
return dictionary->addNgramEntry(&ngramContext, &ngramProperty);
}
static bool latinime_BinaryDictionary_removeNgramEntry(JNIEnv *env, jclass clazz, jlong dict,
@ -421,13 +421,13 @@ static bool latinime_BinaryDictionary_removeNgramEntry(JNIEnv *env, jclass clazz
if (!dictionary) {
return false;
}
const PrevWordsInfo prevWordsInfo = JniDataUtils::constructPrevWordsInfo(env,
const NgramContext ngramContext = JniDataUtils::constructNgramContext(env,
prevWordCodePointArrays, isBeginningOfSentenceArray,
env->GetArrayLength(prevWordCodePointArrays));
jsize codePointCount = env->GetArrayLength(word);
int wordCodePoints[codePointCount];
env->GetIntArrayRegion(word, 0, codePointCount, wordCodePoints);
return dictionary->removeNgramEntry(&prevWordsInfo,
return dictionary->removeNgramEntry(&ngramContext,
CodePointArrayView(wordCodePoints, codePointCount));
}
@ -439,14 +439,14 @@ static bool latinime_BinaryDictionary_updateEntriesForWordWithNgramContext(JNIEn
if (!dictionary) {
return false;
}
const PrevWordsInfo prevWordsInfo = JniDataUtils::constructPrevWordsInfo(env,
const NgramContext ngramContext = JniDataUtils::constructNgramContext(env,
prevWordCodePointArrays, isBeginningOfSentenceArray,
env->GetArrayLength(prevWordCodePointArrays));
jsize codePointCount = env->GetArrayLength(word);
int wordCodePoints[codePointCount];
env->GetIntArrayRegion(word, 0, codePointCount, wordCodePoints);
const HistoricalInfo historicalInfo(timestamp, 0 /* level */, count);
return dictionary->updateEntriesForWordWithNgramContext(&prevWordsInfo,
return dictionary->updateEntriesForWordWithNgramContext(&ngramContext,
CodePointArrayView(wordCodePoints, codePointCount), isValidWord == JNI_TRUE,
historicalInfo);
}
@ -529,9 +529,9 @@ static int latinime_BinaryDictionary_addMultipleDictionaryEntries(JNIEnv *env, j
const NgramProperty ngramProperty(
CodePointArrayView(word1CodePoints, word1Length).toVector(),
bigramProbability, HistoricalInfo(timestamp, 0 /* level */, 1 /* count */));
const PrevWordsInfo prevWordsInfo(word0CodePoints, word0Length,
const NgramContext ngramContext(word0CodePoints, word0Length,
false /* isBeginningOfSentence */);
dictionary->addNgramEntry(&prevWordsInfo, &ngramProperty);
dictionary->addNgramEntry(&ngramContext, &ngramProperty);
}
if (dictionary->needsToRunGC(true /* mindsBlockByGC */)) {
return i + 1;
@ -641,10 +641,10 @@ static bool latinime_BinaryDictionary_migrateNative(JNIEnv *env, jclass clazz, j
return false;
}
}
const PrevWordsInfo prevWordsInfo(wordCodePoints, wordCodePointCount,
const NgramContext ngramContext(wordCodePoints, wordCodePointCount,
wordProperty.getUnigramProperty()->representsBeginningOfSentence());
for (const NgramProperty &ngramProperty : *wordProperty.getNgramProperties()) {
if (!dictionaryStructureWithBufferPolicy->addNgramEntry(&prevWordsInfo,
if (!dictionaryStructureWithBufferPolicy->addNgramEntry(&ngramContext,
&ngramProperty)) {
LogUtils::logToJava(env, "Cannot add ngram to the new dict.");
return false;

8
native/jni/com_android_inputmethod_latin_DicTraverseSession.cpp
View file

@ -40,14 +40,14 @@ static void latinime_initDicTraverseSession(JNIEnv *env, jclass clazz, jlong tra
}
Dictionary *dict = reinterpret_cast<Dictionary *>(dictionary);
if (!previousWord) {
PrevWordsInfo prevWordsInfo;
ts->init(dict, &prevWordsInfo, 0 /* suggestOptions */);
NgramContext emptyNgramContext;
ts->init(dict, &emptyNgramContext, 0 /* suggestOptions */);
return;
}
int prevWord[previousWordLength];
env->GetIntArrayRegion(previousWord, 0, previousWordLength, prevWord);
PrevWordsInfo prevWordsInfo(prevWord, previousWordLength, false /* isStartOfSentence */);
ts->init(dict, &prevWordsInfo, 0 /* suggestOptions */);
NgramContext ngramContext(prevWord, previousWordLength, false /* isStartOfSentence */);
ts->init(dict, &ngramContext, 0 /* suggestOptions */);
}
static void latinime_releaseDicTraverseSession(JNIEnv *env, jclass clazz, jlong traverseSession) {

38
native/jni/src/suggest/core/dictionary/dictionary.cpp
View file

@ -46,11 +46,11 @@ Dictionary::Dictionary(JNIEnv *env, DictionaryStructureWithBufferPolicy::Structu
void Dictionary::getSuggestions(ProximityInfo *proximityInfo, DicTraverseSession *traverseSession,
int *xcoordinates, int *ycoordinates, int *times, int *pointerIds, int *inputCodePoints,
int inputSize, const PrevWordsInfo *const prevWordsInfo,
int inputSize, const NgramContext *const ngramContext,
const SuggestOptions *const suggestOptions, const float weightOfLangModelVsSpatialModel,
SuggestionResults *const outSuggestionResults) const {
TimeKeeper::setCurrentTime();
traverseSession->init(this, prevWordsInfo, suggestOptions);
traverseSession->init(this, ngramContext, suggestOptions);
const auto &suggest = suggestOptions->isGesture() ? mGestureSuggest : mTypingSuggest;
suggest->getSuggestions(proximityInfo, traverseSession, xcoordinates,
ycoordinates, times, pointerIds, inputCodePoints, inputSize,
@ -58,10 +58,10 @@ void Dictionary::getSuggestions(ProximityInfo *proximityInfo, DicTraverseSession
}
Dictionary::NgramListenerForPrediction::NgramListenerForPrediction(
const PrevWordsInfo *const prevWordsInfo, const WordIdArrayView prevWordIds,
const NgramContext *const ngramContext, const WordIdArrayView prevWordIds,
SuggestionResults *const suggestionResults,
const DictionaryStructureWithBufferPolicy *const dictStructurePolicy)
: mPrevWordsInfo(prevWordsInfo), mPrevWordIds(prevWordIds),
: mNgramContext(ngramContext), mPrevWordIds(prevWordIds),
mSuggestionResults(suggestionResults), mDictStructurePolicy(dictStructurePolicy) {}
void Dictionary::NgramListenerForPrediction::onVisitEntry(const int ngramProbability,
@ -69,7 +69,7 @@ void Dictionary::NgramListenerForPrediction::onVisitEntry(const int ngramProbabi
if (targetWordId == NOT_A_WORD_ID) {
return;
}
if (mPrevWordsInfo->isNthPrevWordBeginningOfSentence(1 /* n */)
if (mNgramContext->isNthPrevWordBeginningOfSentence(1 /* n */)
&& ngramProbability == NOT_A_PROBABILITY) {
return;
}
@ -85,20 +85,20 @@ void Dictionary::NgramListenerForPrediction::onVisitEntry(const int ngramProbabi
wordAttributes.getProbability());
}
void Dictionary::getPredictions(const PrevWordsInfo *const prevWordsInfo,
void Dictionary::getPredictions(const NgramContext *const ngramContext,
SuggestionResults *const outSuggestionResults) const {
TimeKeeper::setCurrentTime();
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds(
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(
mDictionaryStructureWithBufferPolicy.get(), &prevWordIdArray,
true /* tryLowerCaseSearch */);
NgramListenerForPrediction listener(prevWordsInfo, prevWordIds, outSuggestionResults,
NgramListenerForPrediction listener(ngramContext, prevWordIds, outSuggestionResults,
mDictionaryStructureWithBufferPolicy.get());
mDictionaryStructureWithBufferPolicy->iterateNgramEntries(prevWordIds, &listener);
}
int Dictionary::getProbability(const CodePointArrayView codePoints) const {
return getNgramProbability(nullptr /* prevWordsInfo */, codePoints);
return getNgramProbability(nullptr /* ngramContext */, codePoints);
}
int Dictionary::getMaxProbabilityOfExactMatches(const CodePointArrayView codePoints) const {
@ -107,18 +107,18 @@ int Dictionary::getMaxProbabilityOfExactMatches(const CodePointArrayView codePoi
mDictionaryStructureWithBufferPolicy.get(), codePoints);
}
int Dictionary::getNgramProbability(const PrevWordsInfo *const prevWordsInfo,
int Dictionary::getNgramProbability(const NgramContext *const ngramContext,
const CodePointArrayView codePoints) const {
TimeKeeper::setCurrentTime();
const int wordId = mDictionaryStructureWithBufferPolicy->getWordId(codePoints,
false /* forceLowerCaseSearch */);
if (wordId == NOT_A_WORD_ID) return NOT_A_PROBABILITY;
if (!prevWordsInfo) {
if (!ngramContext) {
return getDictionaryStructurePolicy()->getProbabilityOfWord(WordIdArrayView(), wordId);
}
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds
(mDictionaryStructureWithBufferPolicy.get(), &prevWordIdArray,
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(
mDictionaryStructureWithBufferPolicy.get(), &prevWordIdArray,
true /* tryLowerCaseSearch */);
return getDictionaryStructurePolicy()->getProbabilityOfWord(prevWordIds, wordId);
}
@ -140,23 +140,23 @@ bool Dictionary::removeUnigramEntry(const CodePointArrayView codePoints) {
return mDictionaryStructureWithBufferPolicy->removeUnigramEntry(codePoints);
}
bool Dictionary::addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Dictionary::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
TimeKeeper::setCurrentTime();
return mDictionaryStructureWithBufferPolicy->addNgramEntry(prevWordsInfo, ngramProperty);
return mDictionaryStructureWithBufferPolicy->addNgramEntry(ngramContext, ngramProperty);
}
bool Dictionary::removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Dictionary::removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView codePoints) {
TimeKeeper::setCurrentTime();
return mDictionaryStructureWithBufferPolicy->removeNgramEntry(prevWordsInfo, codePoints);
return mDictionaryStructureWithBufferPolicy->removeNgramEntry(ngramContext, codePoints);
}
bool Dictionary::updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
bool Dictionary::updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView codePoints, const bool isValidWord,
const HistoricalInfo historicalInfo) {
TimeKeeper::setCurrentTime();
return mDictionaryStructureWithBufferPolicy->updateEntriesForWordWithNgramContext(prevWordsInfo,
return mDictionaryStructureWithBufferPolicy->updateEntriesForWordWithNgramContext(ngramContext,
codePoints, isValidWord, historicalInfo);
}

18
native/jni/src/suggest/core/dictionary/dictionary.h
View file

@ -33,7 +33,7 @@ namespace latinime {
class DictionaryStructureWithBufferPolicy;
class DicTraverseSession;
class PrevWordsInfo;
class NgramContext;
class ProximityInfo;
class SuggestionResults;
class SuggestOptions;
@ -66,18 +66,18 @@ class Dictionary {
void getSuggestions(ProximityInfo *proximityInfo, DicTraverseSession *traverseSession,
int *xcoordinates, int *ycoordinates, int *times, int *pointerIds, int *inputCodePoints,
int inputSize, const PrevWordsInfo *const prevWordsInfo,
int inputSize, const NgramContext *const ngramContext,
const SuggestOptions *const suggestOptions, const float weightOfLangModelVsSpatialModel,
SuggestionResults *const outSuggestionResults) const;
void getPredictions(const PrevWordsInfo *const prevWordsInfo,
void getPredictions(const NgramContext *const ngramContext,
SuggestionResults *const outSuggestionResults) const;
int getProbability(const CodePointArrayView codePoints) const;
int getMaxProbabilityOfExactMatches(const CodePointArrayView codePoints) const;
int getNgramProbability(const PrevWordsInfo *const prevWordsInfo,
int getNgramProbability(const NgramContext *const ngramContext,
const CodePointArrayView codePoints) const;
bool addUnigramEntry(const CodePointArrayView codePoints,
@ -85,13 +85,13 @@ class Dictionary {
bool removeUnigramEntry(const CodePointArrayView codePoints);
bool addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView codePoints);
bool updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
bool updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView codePoints, const bool isValidWord,
const HistoricalInfo historicalInfo);
@ -123,7 +123,7 @@ class Dictionary {
class NgramListenerForPrediction : public NgramListener {
public:
NgramListenerForPrediction(const PrevWordsInfo *const prevWordsInfo,
NgramListenerForPrediction(const NgramContext *const ngramContext,
const WordIdArrayView prevWordIds, SuggestionResults *const suggestionResults,
const DictionaryStructureWithBufferPolicy *const dictStructurePolicy);
virtual void onVisitEntry(const int ngramProbability, const int targetWordId);
@ -131,7 +131,7 @@ class Dictionary {
private:
DISALLOW_IMPLICIT_CONSTRUCTORS(NgramListenerForPrediction);
const PrevWordsInfo *const mPrevWordsInfo;
const NgramContext *const mNgramContext;
const WordIdArrayView mPrevWordIds;
SuggestionResults *const mSuggestionResults;
const DictionaryStructureWithBufferPolicy *const mDictStructurePolicy;

6
native/jni/src/suggest/core/dictionary/dictionary_utils.cpp
View file

@ -33,10 +33,10 @@ namespace latinime {
std::vector<DicNode> current;
std::vector<DicNode> next;
// No prev words information.
PrevWordsInfo emptyPrevWordsInfo;
// No ngram context.
NgramContext emptyNgramContext;
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = emptyPrevWordsInfo.getPrevWordIds(
const WordIdArrayView prevWordIds = emptyNgramContext.getPrevWordIds(
dictionaryStructurePolicy, &prevWordIdArray, false /* tryLowerCaseSearch */);
current.emplace_back();
DicNodeUtils::initAsRoot(dictionaryStructurePolicy, prevWordIds, &current.front());

8
native/jni/src/suggest/core/policy/dictionary_structure_with_buffer_policy.h
View file

@ -33,7 +33,7 @@ class DicNodeVector;
class DictionaryHeaderStructurePolicy;
class MultiBigramMap;
class NgramListener;
class PrevWordsInfo;
class NgramContext;
class UnigramProperty;
/*
@ -81,15 +81,15 @@ class DictionaryStructureWithBufferPolicy {
virtual bool removeUnigramEntry(const CodePointArrayView wordCodePoints) = 0;
// Returns whether the update was success or not.
virtual bool addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
virtual bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) = 0;
// Returns whether the update was success or not.
virtual bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
virtual bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints) = 0;
// Returns whether the update was success or not.
virtual bool updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
virtual bool updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints, const bool isValidWord,
const HistoricalInfo historicalInfo) = 0;

4
native/jni/src/suggest/core/session/dic_traverse_session.cpp
View file

@ -30,12 +30,12 @@ const int DicTraverseSession::DICTIONARY_SIZE_THRESHOLD_TO_USE_LARGE_CACHE_FOR_S
256 * 1024;
void DicTraverseSession::init(const Dictionary *const dictionary,
const PrevWordsInfo *const prevWordsInfo, const SuggestOptions *const suggestOptions) {
const NgramContext *const ngramContext, const SuggestOptions *const suggestOptions) {
mDictionary = dictionary;
mMultiWordCostMultiplier = getDictionaryStructurePolicy()->getHeaderStructurePolicy()
->getMultiWordCostMultiplier();
mSuggestOptions = suggestOptions;
mPrevWordIdCount = prevWordsInfo->getPrevWordIds(getDictionaryStructurePolicy(),
mPrevWordIdCount = ngramContext->getPrevWordIds(getDictionaryStructurePolicy(),
&mPrevWordIdArray, true /* tryLowerCaseSearch */).size();
}

4
native/jni/src/suggest/core/session/dic_traverse_session.h
View file

@ -30,7 +30,7 @@ namespace latinime {
class Dictionary;
class DictionaryStructureWithBufferPolicy;
class PrevWordsInfo;
class NgramContext;
class ProximityInfo;
class SuggestOptions;
@ -61,7 +61,7 @@ class DicTraverseSession {
// Non virtual inline destructor -- never inherit this class
AK_FORCE_INLINE ~DicTraverseSession() {}
void init(const Dictionary *dictionary, const PrevWordsInfo *const prevWordsInfo,
void init(const Dictionary *dictionary, const NgramContext *const ngramContext,
const SuggestOptions *const suggestOptions);
// TODO: Remove and merge into init
void setupForGetSuggestions(const ProximityInfo *pInfo, const int *inputCodePoints,

30
native/jni/src/suggest/core/session/prev_words_info.h
View file

@ -14,8 +14,8 @@
* limitations under the License.
*/
#ifndef LATINIME_PREV_WORDS_INFO_H
#define LATINIME_PREV_WORDS_INFO_H
#ifndef LATINIME_NGRAM_CONTEXT_H
#define LATINIME_NGRAM_CONTEXT_H
#include <array>
@ -27,25 +27,25 @@
namespace latinime {
// Rename to NgramContext.
class PrevWordsInfo {
class NgramContext {
public:
// No prev word information.
PrevWordsInfo() : mPrevWordCount(0) {
NgramContext() : mPrevWordCount(0) {
clear();
}
PrevWordsInfo(const PrevWordsInfo &prevWordsInfo)
: mPrevWordCount(prevWordsInfo.mPrevWordCount) {
NgramContext(const NgramContext &ngramContext)
: mPrevWordCount(ngramContext.mPrevWordCount) {
for (size_t i = 0; i < mPrevWordCount; ++i) {
mPrevWordCodePointCount[i] = prevWordsInfo.mPrevWordCodePointCount[i];
memmove(mPrevWordCodePoints[i], prevWordsInfo.mPrevWordCodePoints[i],
mPrevWordCodePointCount[i] = ngramContext.mPrevWordCodePointCount[i];
memmove(mPrevWordCodePoints[i], ngramContext.mPrevWordCodePoints[i],
sizeof(mPrevWordCodePoints[i][0]) * mPrevWordCodePointCount[i]);
mIsBeginningOfSentence[i] = prevWordsInfo.mIsBeginningOfSentence[i];
mIsBeginningOfSentence[i] = ngramContext.mIsBeginningOfSentence[i];
}
}
// Construct from previous words.
PrevWordsInfo(const int prevWordCodePoints[][MAX_WORD_LENGTH],
NgramContext(const int prevWordCodePoints[][MAX_WORD_LENGTH],
const int *const prevWordCodePointCount, const bool *const isBeginningOfSentence,
const size_t prevWordCount)
: mPrevWordCount(std::min(NELEMS(mPrevWordCodePoints), prevWordCount)) {
@ -62,7 +62,7 @@ class PrevWordsInfo {
}
// Construct from a previous word.
PrevWordsInfo(const int *const prevWordCodePoints, const int prevWordCodePointCount,
NgramContext(const int *const prevWordCodePoints, const int prevWordCodePointCount,
const bool isBeginningOfSentence) : mPrevWordCount(1) {
clear();
if (prevWordCodePointCount > MAX_WORD_LENGTH || !prevWordCodePoints) {
@ -79,8 +79,8 @@ class PrevWordsInfo {
}
// TODO: Remove.
const PrevWordsInfo getTrimmedPrevWordsInfo(const size_t maxPrevWordCount) const {
return PrevWordsInfo(mPrevWordCodePoints, mPrevWordCodePointCount, mIsBeginningOfSentence,
const NgramContext getTrimmedNgramContext(const size_t maxPrevWordCount) const {
return NgramContext(mPrevWordCodePoints, mPrevWordCodePointCount, mIsBeginningOfSentence,
std::min(mPrevWordCount, maxPrevWordCount));
}
@ -123,7 +123,7 @@ class PrevWordsInfo {
}
private:
DISALLOW_ASSIGNMENT_OPERATOR(PrevWordsInfo);
DISALLOW_ASSIGNMENT_OPERATOR(NgramContext);
static int getWordId(const DictionaryStructureWithBufferPolicy *const dictStructurePolicy,
const int *const wordCodePoints, const int wordCodePointCount,
@ -166,4 +166,4 @@ class PrevWordsInfo {
bool mIsBeginningOfSentence[MAX_PREV_WORD_COUNT_FOR_N_GRAM];
};
} // namespace latinime
#endif // LATINIME_PREV_WORDS_INFO_H
#endif // LATINIME_NGRAM_CONTEXT_H

28
native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.cpp
View file

@ -338,7 +338,7 @@ bool Ver4PatriciaTriePolicy::removeUnigramEntry(const CodePointArrayView wordCod
return mNodeWriter.suppressUnigramEntry(&ptNodeParams);
}
bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: addNgramEntry() is called for non-updatable dictionary.");
@ -349,8 +349,8 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
mDictBuffer->getTailPosition());
return false;
}
if (!prevWordsInfo->isValid()) {
AKLOGE("prev words info is not valid for adding n-gram entry to the dictionary.");
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for adding n-gram entry to the dictionary.");
return false;
}
if (ngramProperty->getTargetCodePoints()->size() > MAX_WORD_LENGTH) {
@ -359,23 +359,23 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
return false;
}
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds(this, &prevWordIdArray,
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(this, &prevWordIdArray,
false /* tryLowerCaseSearch */);
if (prevWordIds.empty()) {
return false;
}
if (prevWordIds[0] == NOT_A_WORD_ID) {
if (prevWordsInfo->isNthPrevWordBeginningOfSentence(1 /* n */)) {
if (ngramContext->isNthPrevWordBeginningOfSentence(1 /* n */)) {
const UnigramProperty beginningOfSentenceUnigramProperty(
true /* representsBeginningOfSentence */, true /* isNotAWord */,
false /* isBlacklisted */, MAX_PROBABILITY /* probability */, HistoricalInfo());
if (!addUnigramEntry(prevWordsInfo->getNthPrevWordCodePoints(1 /* n */),
if (!addUnigramEntry(ngramContext->getNthPrevWordCodePoints(1 /* n */),
&beginningOfSentenceUnigramProperty)) {
AKLOGE("Cannot add unigram entry for the beginning-of-sentence.");
return false;
}
// Refresh word ids.
prevWordsInfo->getPrevWordIds(this, &prevWordIdArray, false /* tryLowerCaseSearch */);
ngramContext->getPrevWordIds(this, &prevWordIdArray, false /* tryLowerCaseSearch */);
} else {
return false;
}
@ -399,7 +399,7 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
}
}
bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Ver4PatriciaTriePolicy::removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: removeNgramEntry() is called for non-updatable dictionary.");
@ -410,8 +410,8 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
mDictBuffer->getTailPosition());
return false;
}
if (!prevWordsInfo->isValid()) {
AKLOGE("prev words info is not valid for removing n-gram entry form the dictionary.");
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for removing n-gram entry form the dictionary.");
return false;
}
if (wordCodePoints.size() > MAX_WORD_LENGTH) {
@ -419,7 +419,7 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
wordCodePoints.size());
}
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds(this, &prevWordIdArray,
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(this, &prevWordIdArray,
false /* tryLowerCaseSerch */);
if (prevWordIds.firstOrDefault(NOT_A_WORD_ID) == NOT_A_WORD_ID) {
return false;
@ -441,7 +441,7 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
bool Ver4PatriciaTriePolicy::updateEntriesForWordWithNgramContext(
const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView wordCodePoints,
const NgramContext *const ngramContext, const CodePointArrayView wordCodePoints,
const bool isValidWord, const HistoricalInfo historicalInfo) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: updateEntriesForWordWithNgramContext() is called for non-updatable "
@ -455,11 +455,11 @@ bool Ver4PatriciaTriePolicy::updateEntriesForWordWithNgramContext(
AKLOGE("Cannot update unigarm entry in updateEntriesForWordWithNgramContext().");
return false;
}
const int probabilityForNgram = prevWordsInfo->isNthPrevWordBeginningOfSentence(1 /* n */)
const int probabilityForNgram = ngramContext->isNthPrevWordBeginningOfSentence(1 /* n */)
? NOT_A_PROBABILITY : probability;
const NgramProperty ngramProperty(wordCodePoints.toVector(), probabilityForNgram,
historicalInfo);
if (!addNgramEntry(prevWordsInfo, &ngramProperty)) {
if (!addNgramEntry(ngramContext, &ngramProperty)) {
AKLOGE("Cannot update unigarm entry in updateEntriesForWordWithNgramContext().");
return false;
}

6
native/jni/src/suggest/policyimpl/dictionary/structure/backward/v402/ver4_patricia_trie_policy.h
View file

@ -112,13 +112,13 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
bool removeUnigramEntry(const CodePointArrayView wordCodePoints);
bool addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints);
bool updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
bool updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints, const bool isValidWord,
const HistoricalInfo historicalInfo);

6
native/jni/src/suggest/policyimpl/dictionary/structure/v2/patricia_trie_policy.h
View file

@ -93,21 +93,21 @@ class PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
return false;
}
bool addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
// This method should not be called for non-updatable dictionary.
AKLOGI("Warning: addNgramEntry() is called for non-updatable dictionary.");
return false;
}
bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints) {
// This method should not be called for non-updatable dictionary.
AKLOGI("Warning: removeNgramEntry() is called for non-updatable dictionary.");
return false;
}
bool updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
bool updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints, const bool isValidWord,
const HistoricalInfo historicalInfo) {
// This method should not be called for non-updatable dictionary.

32
native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.cpp
View file

@ -266,7 +266,7 @@ bool Ver4PatriciaTriePolicy::removeUnigramEntry(const CodePointArrayView wordCod
return true;
}
bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: addNgramEntry() is called for non-updatable dictionary.");
@ -277,8 +277,8 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
mDictBuffer->getTailPosition());
return false;
}
if (!prevWordsInfo->isValid()) {
AKLOGE("prev words info is not valid for adding n-gram entry to the dictionary.");
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for adding n-gram entry to the dictionary.");
return false;
}
if (ngramProperty->getTargetCodePoints()->size() > MAX_WORD_LENGTH) {
@ -287,7 +287,7 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
return false;
}
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds(this, &prevWordIdArray,
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(this, &prevWordIdArray,
false /* tryLowerCaseSearch */);
if (prevWordIds.empty()) {
return false;
@ -296,19 +296,19 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
if (prevWordIds[i] != NOT_A_WORD_ID) {
continue;
}
if (!prevWordsInfo->isNthPrevWordBeginningOfSentence(i + 1 /* n */)) {
if (!ngramContext->isNthPrevWordBeginningOfSentence(i + 1 /* n */)) {
return false;
}
const UnigramProperty beginningOfSentenceUnigramProperty(
true /* representsBeginningOfSentence */, true /* isNotAWord */,
false /* isBlacklisted */, MAX_PROBABILITY /* probability */, HistoricalInfo());
if (!addUnigramEntry(prevWordsInfo->getNthPrevWordCodePoints(1 /* n */),
if (!addUnigramEntry(ngramContext->getNthPrevWordCodePoints(1 /* n */),
&beginningOfSentenceUnigramProperty)) {
AKLOGE("Cannot add unigram entry for the beginning-of-sentence.");
return false;
}
// Refresh word ids.
prevWordsInfo->getPrevWordIds(this, &prevWordIdArray, false /* tryLowerCaseSearch */);
ngramContext->getPrevWordIds(this, &prevWordIdArray, false /* tryLowerCaseSearch */);
}
const int wordId = getWordId(CodePointArrayView(*ngramProperty->getTargetCodePoints()),
false /* forceLowerCaseSearch */);
@ -326,7 +326,7 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const PrevWordsInfo *const prevWordsI
}
}
bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool Ver4PatriciaTriePolicy::removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: removeNgramEntry() is called for non-updatable dictionary.");
@ -337,8 +337,8 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
mDictBuffer->getTailPosition());
return false;
}
if (!prevWordsInfo->isValid()) {
AKLOGE("prev words info is not valid for removing n-gram entry form the dictionary.");
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for removing n-gram entry form the dictionary.");
return false;
}
if (wordCodePoints.size() > MAX_WORD_LENGTH) {
@ -346,7 +346,7 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
wordCodePoints.size());
}
WordIdArray<MAX_PREV_WORD_COUNT_FOR_N_GRAM> prevWordIdArray;
const WordIdArrayView prevWordIds = prevWordsInfo->getPrevWordIds(this, &prevWordIdArray,
const WordIdArrayView prevWordIds = ngramContext->getPrevWordIds(this, &prevWordIdArray,
false /* tryLowerCaseSerch */);
if (prevWordIds.empty() || prevWordIds.contains(NOT_A_WORD_ID)) {
return false;
@ -364,7 +364,7 @@ bool Ver4PatriciaTriePolicy::removeNgramEntry(const PrevWordsInfo *const prevWor
}
bool Ver4PatriciaTriePolicy::updateEntriesForWordWithNgramContext(
const PrevWordsInfo *const prevWordsInfo, const CodePointArrayView wordCodePoints,
const NgramContext *const ngramContext, const CodePointArrayView wordCodePoints,
const bool isValidWord, const HistoricalInfo historicalInfo) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: updateEntriesForWordWithNgramContext() is called for non-updatable "
@ -379,13 +379,13 @@ bool Ver4PatriciaTriePolicy::updateEntriesForWordWithNgramContext(
AKLOGE("Cannot update unigarm entry in updateEntriesForWordWithNgramContext().");
return false;
}
const int probabilityForNgram = prevWordsInfo->isNthPrevWordBeginningOfSentence(1 /* n */)
const int probabilityForNgram = ngramContext->isNthPrevWordBeginningOfSentence(1 /* n */)
? NOT_A_PROBABILITY : probability;
const NgramProperty ngramProperty(wordCodePoints.toVector(), probabilityForNgram,
historicalInfo);
for (size_t i = 1; i <= prevWordsInfo->getPrevWordCount(); ++i) {
const PrevWordsInfo trimmedPrevWordsInfo(prevWordsInfo->getTrimmedPrevWordsInfo(i));
if (!addNgramEntry(&trimmedPrevWordsInfo, &ngramProperty)) {
for (size_t i = 1; i <= ngramContext->getPrevWordCount(); ++i) {
const NgramContext trimmedNgramContext(ngramContext->getTrimmedNgramContext(i));
if (!addNgramEntry(&trimmedNgramContext, &ngramProperty)) {
AKLOGE("Cannot update ngram entry in updateEntriesForWordWithNgramContext().");
return false;
}

6
native/jni/src/suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h
View file

@ -92,13 +92,13 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
bool removeUnigramEntry(const CodePointArrayView wordCodePoints);
bool addNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool removeNgramEntry(const PrevWordsInfo *const prevWordsInfo,
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints);
bool updateEntriesForWordWithNgramContext(const PrevWordsInfo *const prevWordsInfo,
bool updateEntriesForWordWithNgramContext(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints, const bool isValidWord,
const HistoricalInfo historicalInfo);

4
native/jni/src/utils/jni_data_utils.h
View file

@ -96,7 +96,7 @@ class JniDataUtils {
}
}
static PrevWordsInfo constructPrevWordsInfo(JNIEnv *env, jobjectArray prevWordCodePointArrays,
static NgramContext constructNgramContext(JNIEnv *env, jobjectArray prevWordCodePointArrays,
jbooleanArray isBeginningOfSentenceArray, const size_t prevWordCount) {
int prevWordCodePoints[MAX_PREV_WORD_COUNT_FOR_N_GRAM][MAX_WORD_LENGTH];
int prevWordCodePointCount[MAX_PREV_WORD_COUNT_FOR_N_GRAM];
@ -119,7 +119,7 @@ class JniDataUtils {
&isBeginningOfSentenceBoolean);
isBeginningOfSentence[i] = isBeginningOfSentenceBoolean == JNI_TRUE;
}
return PrevWordsInfo(prevWordCodePoints, prevWordCodePointCount, isBeginningOfSentence,
return NgramContext(prevWordCodePoints, prevWordCodePointCount, isBeginningOfSentence,
prevWordCount);
}