Make NgramProperty have NgramContext.

Bug: 14425059
Change-Id: I210acb816b122857dbbe1ee4dd6a35c5335bf2bf
main
Keisuke Kuroyanagi 2014-10-21 17:12:32 +09:00
parent f87bb77a91
commit 88bb28c132
12 changed files with 48 additions and 43 deletions

View File

@ -409,9 +409,10 @@ static bool latinime_BinaryDictionary_addNgramEntry(JNIEnv *env, jclass clazz, j
int wordCodePoints[wordLength];
env->GetIntArrayRegion(word, 0, wordLength, wordCodePoints);
// Use 1 for count to indicate the ngram has inputted.
const NgramProperty ngramProperty(CodePointArrayView(wordCodePoints, wordLength).toVector(),
const NgramProperty ngramProperty(ngramContext,
CodePointArrayView(wordCodePoints, wordLength).toVector(),
probability, HistoricalInfo(timestamp, 0 /* level */, 1 /* count */));
return dictionary->addNgramEntry(&ngramContext, &ngramProperty);
return dictionary->addNgramEntry(&ngramProperty);
}
static bool latinime_BinaryDictionary_removeNgramEntry(JNIEnv *env, jclass clazz, jlong dict,
@ -526,12 +527,12 @@ static int latinime_BinaryDictionary_addMultipleDictionaryEntries(JNIEnv *env, j
if (word0) {
jint bigramProbability = env->GetIntField(languageModelParam, bigramProbabilityFieldId);
// Use 1 for count to indicate the bigram has inputted.
const NgramProperty ngramProperty(
CodePointArrayView(word1CodePoints, word1Length).toVector(),
bigramProbability, HistoricalInfo(timestamp, 0 /* level */, 1 /* count */));
const NgramContext ngramContext(word0CodePoints, word0Length,
false /* isBeginningOfSentence */);
dictionary->addNgramEntry(&ngramContext, &ngramProperty);
const NgramProperty ngramProperty(ngramContext,
CodePointArrayView(word1CodePoints, word1Length).toVector(),
bigramProbability, HistoricalInfo(timestamp, 0 /* level */, 1 /* count */));
dictionary->addNgramEntry(&ngramProperty);
}
if (dictionary->needsToRunGC(true /* mindsBlockByGC */)) {
return i + 1;
@ -641,11 +642,8 @@ static bool latinime_BinaryDictionary_migrateNative(JNIEnv *env, jclass clazz, j
return false;
}
}
const NgramContext ngramContext(wordCodePoints, wordCodePointCount,
wordProperty.getUnigramProperty()->representsBeginningOfSentence());
for (const NgramProperty &ngramProperty : *wordProperty.getNgramProperties()) {
if (!dictionaryStructureWithBufferPolicy->addNgramEntry(&ngramContext,
&ngramProperty)) {
if (!dictionaryStructureWithBufferPolicy->addNgramEntry(&ngramProperty)) {
LogUtils::logToJava(env, "Cannot add ngram to the new dict.");
return false;
}

View File

@ -140,10 +140,9 @@ bool Dictionary::removeUnigramEntry(const CodePointArrayView codePoints) {
return mDictionaryStructureWithBufferPolicy->removeUnigramEntry(codePoints);
}
bool Dictionary::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
bool Dictionary::addNgramEntry(const NgramProperty *const ngramProperty) {
TimeKeeper::setCurrentTime();
return mDictionaryStructureWithBufferPolicy->addNgramEntry(ngramContext, ngramProperty);
return mDictionaryStructureWithBufferPolicy->addNgramEntry(ngramProperty);
}
bool Dictionary::removeNgramEntry(const NgramContext *const ngramContext,

View File

@ -85,8 +85,7 @@ class Dictionary {
bool removeUnigramEntry(const CodePointArrayView codePoints);
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool addNgramEntry(const NgramProperty *const ngramProperty);
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView codePoints);

View File

@ -21,15 +21,20 @@
#include "defines.h"
#include "suggest/core/dictionary/property/historical_info.h"
#include "suggest/core/session/ngram_context.h"
namespace latinime {
class NgramProperty {
public:
NgramProperty(const std::vector<int> &&targetCodePoints, const int probability,
const HistoricalInfo historicalInfo)
: mTargetCodePoints(std::move(targetCodePoints)), mProbability(probability),
mHistoricalInfo(historicalInfo) {}
NgramProperty(const NgramContext &ngramContext, const std::vector<int> &&targetCodePoints,
const int probability, const HistoricalInfo historicalInfo)
: mNgramContext(ngramContext), mTargetCodePoints(std::move(targetCodePoints)),
mProbability(probability), mHistoricalInfo(historicalInfo) {}
const NgramContext *getNgramContext() const {
return &mNgramContext;
}
const std::vector<int> *getTargetCodePoints() const {
return &mTargetCodePoints;
@ -48,6 +53,7 @@ class NgramProperty {
DISALLOW_DEFAULT_CONSTRUCTOR(NgramProperty);
DISALLOW_ASSIGNMENT_OPERATOR(NgramProperty);
const NgramContext mNgramContext;
const std::vector<int> mTargetCodePoints;
const int mProbability;
const HistoricalInfo mHistoricalInfo;

View File

@ -34,9 +34,9 @@ class WordProperty {
: mCodePoints(), mUnigramProperty(), mNgrams() {}
WordProperty(const std::vector<int> &&codePoints, const UnigramProperty *const unigramProperty,
const std::vector<NgramProperty> *const bigrams)
const std::vector<NgramProperty> *const ngrams)
: mCodePoints(std::move(codePoints)), mUnigramProperty(*unigramProperty),
mNgrams(*bigrams) {}
mNgrams(*ngrams) {}
void outputProperties(JNIEnv *const env, jintArray outCodePoints, jbooleanArray outFlags,
jintArray outProbabilityInfo, jobject outBigramTargets, jobject outBigramProbabilities,

View File

@ -40,7 +40,6 @@ class UnigramProperty;
* This class abstracts the structure of dictionaries.
* Implement this policy to support additional dictionaries.
*/
// TODO: Use word id instead of terminal PtNode position.
class DictionaryStructureWithBufferPolicy {
public:
typedef std::unique_ptr<DictionaryStructureWithBufferPolicy> StructurePolicyPtr;
@ -81,8 +80,7 @@ class DictionaryStructureWithBufferPolicy {
virtual bool removeUnigramEntry(const CodePointArrayView wordCodePoints) = 0;
// Returns whether the update was success or not.
virtual bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) = 0;
virtual bool addNgramEntry(const NgramProperty *const ngramProperty) = 0;
// Returns whether the update was success or not.
virtual bool removeNgramEntry(const NgramContext *const ngramContext,
@ -106,7 +104,6 @@ class DictionaryStructureWithBufferPolicy {
virtual void getProperty(const char *const query, const int queryLength, char *const outResult,
const int maxResultLength) = 0;
// Used for testing.
virtual const WordProperty getWordProperty(const CodePointArrayView wordCodePoints) const = 0;
// Method to iterate all words in the dictionary.

View File

@ -344,8 +344,7 @@ bool Ver4PatriciaTriePolicy::removeUnigramEntry(const CodePointArrayView wordCod
return mNodeWriter.suppressUnigramEntry(&ptNodeParams);
}
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramProperty *const ngramProperty) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: addNgramEntry() is called for non-updatable dictionary.");
return false;
@ -355,6 +354,7 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContex
mDictBuffer->getTailPosition());
return false;
}
const NgramContext *const ngramContext = ngramProperty->getNgramContext();
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for adding n-gram entry to the dictionary.");
return false;
@ -463,9 +463,9 @@ bool Ver4PatriciaTriePolicy::updateEntriesForWordWithNgramContext(
}
const int probabilityForNgram = ngramContext->isNthPrevWordBeginningOfSentence(1 /* n */)
? NOT_A_PROBABILITY : probability;
const NgramProperty ngramProperty(wordCodePoints.toVector(), probabilityForNgram,
const NgramProperty ngramProperty(*ngramContext, wordCodePoints.toVector(), probabilityForNgram,
historicalInfo);
if (!addNgramEntry(ngramContext, &ngramProperty)) {
if (!addNgramEntry(&ngramProperty)) {
AKLOGE("Cannot update unigarm entry in updateEntriesForWordWithNgramContext().");
return false;
}
@ -585,6 +585,8 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty(
bigramEntry.getHistoricalInfo(), mHeaderPolicy) :
bigramEntry.getProbability();
ngrams.emplace_back(
NgramContext(wordCodePoints.data(), wordCodePoints.size(),
ptNodeParams.representsBeginningOfSentence()),
CodePointArrayView(bigramWord1CodePoints, codePointCount).toVector(),
probability, *historicalInfo);
}

View File

@ -113,8 +113,7 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
bool removeUnigramEntry(const CodePointArrayView wordCodePoints);
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool addNgramEntry(const NgramProperty *const ngramProperty);
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints);

View File

@ -451,6 +451,8 @@ const WordProperty PatriciaTriePolicy::getWordProperty(
bigramWord1CodePoints, &word1Probability);
const int probability = getProbability(word1Probability, bigramsIt.getProbability());
ngrams.emplace_back(
NgramContext(wordCodePoints.data(), wordCodePoints.size(),
ptNodeParams.representsBeginningOfSentence()),
CodePointArrayView(bigramWord1CodePoints, word1CodePointCount).toVector(),
probability, HistoricalInfo());
}

View File

@ -93,8 +93,7 @@ class PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
return false;
}
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
bool addNgramEntry(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;

View File

@ -264,8 +264,7 @@ bool Ver4PatriciaTriePolicy::removeUnigramEntry(const CodePointArrayView wordCod
return true;
}
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty) {
bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramProperty *const ngramProperty) {
if (!mBuffers->isUpdatable()) {
AKLOGI("Warning: addNgramEntry() is called for non-updatable dictionary.");
return false;
@ -275,6 +274,7 @@ bool Ver4PatriciaTriePolicy::addNgramEntry(const NgramContext *const ngramContex
mDictBuffer->getTailPosition());
return false;
}
const NgramContext *const ngramContext = ngramProperty->getNgramContext();
if (!ngramContext->isValid()) {
AKLOGE("Ngram context is not valid for adding n-gram entry to the dictionary.");
return false;
@ -451,7 +451,8 @@ bool Ver4PatriciaTriePolicy::needsToRunGC(const bool mindsBlockByGC) const {
// Needs to reduce dictionary size.
return true;
} else if (mHeaderPolicy->isDecayingDict()) {
return ForgettingCurveUtils::needsToDecay(mindsBlockByGC, mEntryCounters.getEntryCounts(), mHeaderPolicy);
return ForgettingCurveUtils::needsToDecay(mindsBlockByGC, mEntryCounters.getEntryCounts(),
mHeaderPolicy);
}
return false;
}
@ -501,12 +502,16 @@ const WordProperty Ver4PatriciaTriePolicy::getWordProperty(
prevWordIds)) {
const int codePointCount = getCodePointsAndReturnCodePointCount(entry.getWordId(),
MAX_WORD_LENGTH, bigramWord1CodePoints);
const ProbabilityEntry probabilityEntry = entry.getProbabilityEntry();
const HistoricalInfo *const historicalInfo = probabilityEntry.getHistoricalInfo();
const int probability = probabilityEntry.hasHistoricalInfo() ?
const ProbabilityEntry ngramProbabilityEntry = entry.getProbabilityEntry();
const HistoricalInfo *const historicalInfo = ngramProbabilityEntry.getHistoricalInfo();
const int probability = ngramProbabilityEntry.hasHistoricalInfo() ?
ForgettingCurveUtils::decodeProbability(historicalInfo, mHeaderPolicy) :
probabilityEntry.getProbability();
ngrams.emplace_back(CodePointArrayView(bigramWord1CodePoints, codePointCount).toVector(),
ngramProbabilityEntry.getProbability();
ngrams.emplace_back(
NgramContext(
wordCodePoints.data(), wordCodePoints.size(),
probabilityEntry.representsBeginningOfSentence()),
CodePointArrayView(bigramWord1CodePoints, codePointCount).toVector(),
probability, *historicalInfo);
}
// Fetch shortcut information.

View File

@ -92,8 +92,7 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
bool removeUnigramEntry(const CodePointArrayView wordCodePoints);
bool addNgramEntry(const NgramContext *const ngramContext,
const NgramProperty *const ngramProperty);
bool addNgramEntry(const NgramProperty *const ngramProperty);
bool removeNgramEntry(const NgramContext *const ngramContext,
const CodePointArrayView wordCodePoints);