Merge "Add a jni method to fetch unigram information."
This commit is contained in:
commit
fc16c320c6
13 changed files with 403 additions and 2 deletions
|
@ -26,6 +26,7 @@ import com.android.inputmethod.latin.settings.NativeSuggestOptions;
|
|||
import com.android.inputmethod.latin.utils.CollectionUtils;
|
||||
import com.android.inputmethod.latin.utils.JniUtils;
|
||||
import com.android.inputmethod.latin.utils.StringUtils;
|
||||
import com.android.inputmethod.latin.utils.UnigramProperty;
|
||||
|
||||
import java.io.File;
|
||||
import java.util.ArrayList;
|
||||
|
@ -59,6 +60,19 @@ public final class BinaryDictionary extends Dictionary {
|
|||
|
||||
public static final int NOT_A_VALID_TIMESTAMP = -1;
|
||||
|
||||
// Format to get unigram flags from native side via getUnigramPropertyNative().
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_OUTPUT_FLAG_COUNT = 4;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_IS_NOT_A_WORD_INDEX = 0;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_IS_BLACKLISTED_INDEX = 1;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_HAS_BIGRAMS_INDEX = 2;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_HAS_SHORTCUTS_INDEX = 3;
|
||||
|
||||
// Format to get unigram historical info from native side via getUnigramPropertyNative().
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_OUTPUT_HISTORICAL_INFO_COUNT = 3;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_TIMESTAMP_INDEX = 0;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_LEVEL_INDEX = 1;
|
||||
private static final int FORMAT_UNIGRAM_PROPERTY_COUNT_INDEX = 2;
|
||||
|
||||
private long mNativeDict;
|
||||
private final Locale mLocale;
|
||||
private final long mDictSize;
|
||||
|
@ -128,6 +142,10 @@ public final class BinaryDictionary extends Dictionary {
|
|||
private static native int getFormatVersionNative(long dict);
|
||||
private static native int getProbabilityNative(long dict, int[] word);
|
||||
private static native int getBigramProbabilityNative(long dict, int[] word0, int[] word1);
|
||||
private static native void getUnigramPropertyNative(long dict, int[] word,
|
||||
int[] outCodePoints, boolean[] outFlags, int[] outProbability,
|
||||
int[] outHistoricalInfo, ArrayList<int[]> outShortcutTargets,
|
||||
ArrayList<Integer> outShortcutProbabilities);
|
||||
private static native int getSuggestionsNative(long dict, long proximityInfo,
|
||||
long traverseSession, int[] xCoordinates, int[] yCoordinates, int[] times,
|
||||
int[] pointerIds, int[] inputCodePoints, int inputSize, int commitPoint,
|
||||
|
@ -285,6 +303,32 @@ public final class BinaryDictionary extends Dictionary {
|
|||
return getBigramProbabilityNative(mNativeDict, codePoints0, codePoints1);
|
||||
}
|
||||
|
||||
@UsedForTesting
|
||||
public UnigramProperty getUnigramProperty(final String word) {
|
||||
if (TextUtils.isEmpty(word)) {
|
||||
return null;
|
||||
}
|
||||
final int[] codePoints = StringUtils.toCodePointArray(word);
|
||||
final int[] outCodePoints = new int[MAX_WORD_LENGTH];
|
||||
final boolean[] outFlags = new boolean[FORMAT_UNIGRAM_PROPERTY_OUTPUT_FLAG_COUNT];
|
||||
final int[] outProbability = new int[1];
|
||||
final int[] outHistoricalInfo =
|
||||
new int[FORMAT_UNIGRAM_PROPERTY_OUTPUT_HISTORICAL_INFO_COUNT];
|
||||
final ArrayList<int[]> outShortcutTargets = CollectionUtils.newArrayList();
|
||||
final ArrayList<Integer> outShortcutProbabilities = CollectionUtils.newArrayList();
|
||||
getUnigramPropertyNative(mNativeDict, codePoints, outCodePoints, outFlags, outProbability,
|
||||
outHistoricalInfo, outShortcutTargets, outShortcutProbabilities);
|
||||
return new UnigramProperty(codePoints,
|
||||
outFlags[FORMAT_UNIGRAM_PROPERTY_IS_NOT_A_WORD_INDEX],
|
||||
outFlags[FORMAT_UNIGRAM_PROPERTY_IS_BLACKLISTED_INDEX],
|
||||
outFlags[FORMAT_UNIGRAM_PROPERTY_HAS_BIGRAMS_INDEX],
|
||||
outFlags[FORMAT_UNIGRAM_PROPERTY_HAS_SHORTCUTS_INDEX], outProbability[0],
|
||||
outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_TIMESTAMP_INDEX],
|
||||
outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_LEVEL_INDEX],
|
||||
outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_COUNT_INDEX],
|
||||
outShortcutTargets, outShortcutProbabilities);
|
||||
}
|
||||
|
||||
// Add a unigram entry to binary dictionary with unigram attributes in native code.
|
||||
public void addUnigramWord(final String word, final int probability,
|
||||
final String shortcutTarget, final int shortcutProbability, final boolean isNotAWord,
|
||||
|
|
|
@ -0,0 +1,82 @@
|
|||
/*
|
||||
* Copyright (C) 2013 The Android Open Source Project
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
|
||||
package com.android.inputmethod.latin.utils;
|
||||
|
||||
import com.android.inputmethod.annotations.UsedForTesting;
|
||||
import com.android.inputmethod.latin.BinaryDictionary;
|
||||
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
|
||||
|
||||
import java.util.ArrayList;
|
||||
|
||||
// This has information that belong to a unigram. This class has some detailed attributes such as
|
||||
// historical information but they have to be checked only for testing purpose.
|
||||
@UsedForTesting
|
||||
public class UnigramProperty {
|
||||
public final String mCodePoints;
|
||||
public final boolean mIsNotAWord;
|
||||
public final boolean mIsBlacklisted;
|
||||
public final boolean mHasBigrams;
|
||||
public final boolean mHasShortcuts;
|
||||
public final int mProbability;
|
||||
// mTimestamp, mLevel and mCount are historical info. These values are depend on the
|
||||
// implementation in native code; thus, we must not use them and have any assumptions about
|
||||
// them except for tests.
|
||||
public final int mTimestamp;
|
||||
public final int mLevel;
|
||||
public final int mCount;
|
||||
public final ArrayList<WeightedString> mShortcutTargets = CollectionUtils.newArrayList();
|
||||
|
||||
private static int getCodePointCount(final int[] codePoints) {
|
||||
for (int i = 0; i < codePoints.length; i++) {
|
||||
if (codePoints[i] == 0) {
|
||||
return i;
|
||||
}
|
||||
}
|
||||
return codePoints.length;
|
||||
}
|
||||
|
||||
// This represents invalid unigram when the probability is BinaryDictionary.NOT_A_PROBABILITY.
|
||||
public UnigramProperty(final int[] codePoints, final boolean isNotAWord,
|
||||
final boolean isBlacklisted, final boolean hasBigram,
|
||||
final boolean hasShortcuts, final int probability, final int timestamp,
|
||||
final int level, final int count, final ArrayList<int[]> shortcutTargets,
|
||||
final ArrayList<Integer> shortcutProbabilities) {
|
||||
mCodePoints = new String(codePoints, 0 /* offset */, getCodePointCount(codePoints));
|
||||
mIsNotAWord = isNotAWord;
|
||||
mIsBlacklisted = isBlacklisted;
|
||||
mHasBigrams = hasBigram;
|
||||
mHasShortcuts = hasShortcuts;
|
||||
mProbability = probability;
|
||||
mTimestamp = timestamp;
|
||||
mLevel = level;
|
||||
mCount = count;
|
||||
final int shortcutTargetCount = shortcutTargets.size();
|
||||
for (int i = 0; i < shortcutTargetCount; i++) {
|
||||
final int[] shortcutTargetCodePointArray = shortcutTargets.get(i);
|
||||
final String shortcutTargetString = new String(shortcutTargetCodePointArray,
|
||||
0 /* offset */, getCodePointCount(shortcutTargetCodePointArray));
|
||||
mShortcutTargets.add(
|
||||
new WeightedString(shortcutTargetString, shortcutProbabilities.get(i)));
|
||||
}
|
||||
}
|
||||
|
||||
@UsedForTesting
|
||||
public boolean isValid() {
|
||||
return mProbability != BinaryDictionary.NOT_A_PROBABILITY;
|
||||
}
|
||||
}
|
|
@ -58,7 +58,8 @@ LATIN_IME_CORE_SRC_FILES := \
|
|||
dictionary.cpp \
|
||||
digraph_utils.cpp \
|
||||
error_type_utils.cpp \
|
||||
multi_bigram_map.cpp) \
|
||||
multi_bigram_map.cpp \
|
||||
unigram_property.cpp) \
|
||||
$(addprefix suggest/core/layout/, \
|
||||
additional_proximity_chars.cpp \
|
||||
proximity_info.cpp \
|
||||
|
|
|
@ -24,6 +24,7 @@
|
|||
#include "jni.h"
|
||||
#include "jni_common.h"
|
||||
#include "suggest/core/dictionary/dictionary.h"
|
||||
#include "suggest/core/dictionary/unigram_property.h"
|
||||
#include "suggest/core/suggest_options.h"
|
||||
#include "suggest/policyimpl/dictionary/structure/dictionary_structure_with_buffer_policy_factory.h"
|
||||
#include "suggest/policyimpl/dictionary/utils/dict_file_writing_utils.h"
|
||||
|
@ -258,6 +259,21 @@ static jint latinime_BinaryDictionary_getBigramProbability(JNIEnv *env, jclass c
|
|||
word1Length);
|
||||
}
|
||||
|
||||
static void latinime_BinaryDictionary_getUnigramProperty(JNIEnv *env, jclass clazz,
|
||||
jlong dict, jintArray word, jintArray outCodePoints, jbooleanArray outFlags,
|
||||
jintArray outProbability, jintArray outHistoricalInfo, jobject outShortcutTargets,
|
||||
jobject outShortcutProbabilities) {
|
||||
Dictionary *dictionary = reinterpret_cast<Dictionary *>(dict);
|
||||
if (!dictionary) return;
|
||||
const jsize wordLength = env->GetArrayLength(word);
|
||||
int wordCodePoints[wordLength];
|
||||
env->GetIntArrayRegion(word, 0, wordLength, wordCodePoints);
|
||||
const UnigramProperty unigramProperty = dictionary->getUnigramProperty(
|
||||
wordCodePoints, wordLength);
|
||||
unigramProperty.outputProperties(env, outCodePoints, outFlags, outProbability,
|
||||
outHistoricalInfo, outShortcutTargets, outShortcutProbabilities);
|
||||
}
|
||||
|
||||
static jfloat latinime_BinaryDictionary_calcNormalizedScore(JNIEnv *env, jclass clazz,
|
||||
jintArray before, jintArray after, jint score) {
|
||||
jsize beforeLength = env->GetArrayLength(before);
|
||||
|
@ -333,7 +349,6 @@ static void latinime_BinaryDictionary_removeBigramWords(JNIEnv *env, jclass claz
|
|||
word1Length);
|
||||
}
|
||||
|
||||
|
||||
// Returns how many language model params are processed.
|
||||
static int latinime_BinaryDictionary_addMultipleDictionaryEntries(JNIEnv *env, jclass clazz,
|
||||
jlong dict, jobjectArray languageModelParams, jint startIndex) {
|
||||
|
@ -493,6 +508,11 @@ static const JNINativeMethod sMethods[] = {
|
|||
const_cast<char *>("(J[I[I)I"),
|
||||
reinterpret_cast<void *>(latinime_BinaryDictionary_getBigramProbability)
|
||||
},
|
||||
{
|
||||
const_cast<char *>("getUnigramPropertyNative"),
|
||||
const_cast<char *>("(J[I[I[Z[I[ILjava/util/ArrayList;Ljava/util/ArrayList;)V"),
|
||||
reinterpret_cast<void *>(latinime_BinaryDictionary_getUnigramProperty)
|
||||
},
|
||||
{
|
||||
const_cast<char *>("calcNormalizedScoreNative"),
|
||||
const_cast<char *>("([I[II)F"),
|
||||
|
|
|
@ -143,6 +143,13 @@ void Dictionary::getProperty(const char *const query, const int queryLength, cha
|
|||
maxResultLength);
|
||||
}
|
||||
|
||||
const UnigramProperty Dictionary::getUnigramProperty(const int *const codePoints,
|
||||
const int codePointCount) {
|
||||
TimeKeeper::setCurrentTime();
|
||||
return mDictionaryStructureWithBufferPolicy.get()->getUnigramProperty(
|
||||
codePoints, codePointCount);
|
||||
}
|
||||
|
||||
void Dictionary::logDictionaryInfo(JNIEnv *const env) const {
|
||||
int dictionaryIdCodePointBuffer[HEADER_ATTRIBUTE_BUFFER_SIZE];
|
||||
int versionStringCodePointBuffer[HEADER_ATTRIBUTE_BUFFER_SIZE];
|
||||
|
|
|
@ -22,6 +22,7 @@
|
|||
#include "defines.h"
|
||||
#include "jni.h"
|
||||
#include "suggest/core/dictionary/bigram_dictionary.h"
|
||||
#include "suggest/core/dictionary/unigram_property.h"
|
||||
#include "suggest/core/policy/dictionary_header_structure_policy.h"
|
||||
#include "suggest/core/policy/dictionary_structure_with_buffer_policy.h"
|
||||
#include "suggest/core/suggest_interface.h"
|
||||
|
@ -33,6 +34,7 @@ class DictionaryStructureWithBufferPolicy;
|
|||
class DicTraverseSession;
|
||||
class ProximityInfo;
|
||||
class SuggestOptions;
|
||||
class UnigramProperty;
|
||||
|
||||
class Dictionary {
|
||||
public:
|
||||
|
@ -92,6 +94,8 @@ class Dictionary {
|
|||
void getProperty(const char *const query, const int queryLength, char *const outResult,
|
||||
const int maxResultLength);
|
||||
|
||||
const UnigramProperty getUnigramProperty(const int *const codePoints, const int codePointCount);
|
||||
|
||||
const DictionaryStructureWithBufferPolicy *getDictionaryStructurePolicy() const {
|
||||
return mDictionaryStructureWithBufferPolicy.get();
|
||||
}
|
||||
|
|
52
native/jni/src/suggest/core/dictionary/unigram_property.cpp
Normal file
52
native/jni/src/suggest/core/dictionary/unigram_property.cpp
Normal file
|
@ -0,0 +1,52 @@
|
|||
/*
|
||||
* Copyright (C) 2013 The Android Open Source Project
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#include "suggest/core/dictionary/unigram_property.h"
|
||||
|
||||
namespace latinime {
|
||||
|
||||
void UnigramProperty::outputProperties(JNIEnv *const env, jintArray outCodePoints,
|
||||
jbooleanArray outFlags, jintArray outProbability, jintArray outHistoricalInfo,
|
||||
jobject outShortcutTargets, jobject outShortcutProbabilities) const {
|
||||
env->SetIntArrayRegion(outCodePoints, 0 /* start */, mCodePointCount, mCodePoints);
|
||||
jboolean flags[] = {mIsNotAWord, mIsBlacklisted, mHasBigrams, mHasShortcuts};
|
||||
env->SetBooleanArrayRegion(outFlags, 0 /* start */, NELEMS(flags), flags);
|
||||
env->SetIntArrayRegion(outProbability, 0 /* start */, 1 /* len */, &mProbability);
|
||||
int historicalInfo[] = {mTimestamp, mLevel, mCount};
|
||||
env->SetIntArrayRegion(outHistoricalInfo, 0 /* start */, NELEMS(historicalInfo),
|
||||
historicalInfo);
|
||||
|
||||
jclass integerClass = env->FindClass("java/lang/Integer");
|
||||
jmethodID intToIntegerConstructorId = env->GetMethodID(integerClass, "<init>", "(I)V");
|
||||
jclass arrayListClass = env->FindClass("java/util/ArrayList");
|
||||
jmethodID addMethodId = env->GetMethodID(arrayListClass, "add", "(Ljava/lang/Object;)Z");
|
||||
const int shortcutTargetCount = mShortcutTargets.size();
|
||||
for (int i = 0; i < shortcutTargetCount; ++i) {
|
||||
jintArray shortcutTargetCodePointArray = env->NewIntArray(mShortcutTargets[i].size());
|
||||
env->SetIntArrayRegion(shortcutTargetCodePointArray, 0 /* start */,
|
||||
mShortcutTargets[i].size(), &mShortcutTargets[i][0]);
|
||||
env->CallVoidMethod(outShortcutTargets, addMethodId, shortcutTargetCodePointArray);
|
||||
env->DeleteLocalRef(shortcutTargetCodePointArray);
|
||||
jobject integerProbability = env->NewObject(integerClass, intToIntegerConstructorId,
|
||||
mShortcutProbabilities[i]);
|
||||
env->CallVoidMethod(outShortcutProbabilities, addMethodId, integerProbability);
|
||||
env->DeleteLocalRef(integerProbability);
|
||||
}
|
||||
env->DeleteLocalRef(integerClass);
|
||||
env->DeleteLocalRef(arrayListClass);
|
||||
}
|
||||
|
||||
} // namespace latinime
|
87
native/jni/src/suggest/core/dictionary/unigram_property.h
Normal file
87
native/jni/src/suggest/core/dictionary/unigram_property.h
Normal file
|
@ -0,0 +1,87 @@
|
|||
/*
|
||||
* Copyright (C) 2013 The Android Open Source Project
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef LATINIME_UNIGRAM_PROPERTY_H
|
||||
#define LATINIME_UNIGRAM_PROPERTY_H
|
||||
|
||||
#include <vector>
|
||||
|
||||
#include "defines.h"
|
||||
#include "jni.h"
|
||||
|
||||
namespace latinime {
|
||||
|
||||
// This class is used for returning information belonging to a unigram to java side.
|
||||
class UnigramProperty {
|
||||
public:
|
||||
// Invalid unigram.
|
||||
UnigramProperty()
|
||||
: mCodePoints(), mCodePointCount(0), mIsNotAWord(false), mIsBlacklisted(false),
|
||||
mHasBigrams(false), mHasShortcuts(false), mProbability(NOT_A_PROBABILITY),
|
||||
mTimestamp(0), mLevel(0), mCount(0), mShortcutTargets(),
|
||||
mShortcutProbabilities() {}
|
||||
|
||||
UnigramProperty(const UnigramProperty &unigramProperty)
|
||||
: mCodePoints(), mCodePointCount(unigramProperty.mCodePointCount),
|
||||
mIsNotAWord(unigramProperty.mIsNotAWord),
|
||||
mIsBlacklisted(unigramProperty.mIsBlacklisted),
|
||||
mHasBigrams(unigramProperty.mHasBigrams),
|
||||
mHasShortcuts(unigramProperty.mHasShortcuts),
|
||||
mProbability(unigramProperty.mProbability),
|
||||
mTimestamp(unigramProperty.mTimestamp), mLevel(unigramProperty.mLevel),
|
||||
mCount(unigramProperty.mCount), mShortcutTargets(unigramProperty.mShortcutTargets),
|
||||
mShortcutProbabilities(unigramProperty.mShortcutProbabilities) {
|
||||
memcpy(mCodePoints, unigramProperty.mCodePoints, sizeof(mCodePoints));
|
||||
}
|
||||
|
||||
UnigramProperty(const int *const codePoints, const int codePointCount,
|
||||
const bool isNotAWord, const bool isBlacklisted, const bool hasBigrams,
|
||||
const bool hasShortcuts, const int probability, const int timestamp,
|
||||
const int level, const int count,
|
||||
const std::vector<std::vector<int> > *const shortcutTargets,
|
||||
const std::vector<int> *const shortcutProbabilities)
|
||||
: mCodePoints(), mCodePointCount(codePointCount),
|
||||
mIsNotAWord(isNotAWord), mIsBlacklisted(isBlacklisted), mHasBigrams(hasBigrams),
|
||||
mHasShortcuts(hasShortcuts), mProbability(probability), mTimestamp(timestamp),
|
||||
mLevel(level), mCount(count), mShortcutTargets(*shortcutTargets),
|
||||
mShortcutProbabilities(*shortcutProbabilities) {
|
||||
memcpy(mCodePoints, codePoints, sizeof(mCodePoints));
|
||||
}
|
||||
|
||||
void outputProperties(JNIEnv *const env, jintArray outCodePoints, jbooleanArray outFlags,
|
||||
jintArray outProbability, jintArray outHistoricalInfo, jobject outShortcutTargets,
|
||||
jobject outShortcutProbabilities) const;
|
||||
|
||||
private:
|
||||
DISALLOW_ASSIGNMENT_OPERATOR(UnigramProperty);
|
||||
|
||||
int mCodePoints[MAX_WORD_LENGTH];
|
||||
int mCodePointCount;
|
||||
bool mIsNotAWord;
|
||||
bool mIsBlacklisted;
|
||||
bool mHasBigrams;
|
||||
bool mHasShortcuts;
|
||||
int mProbability;
|
||||
// Historical information
|
||||
int mTimestamp;
|
||||
int mLevel;
|
||||
int mCount;
|
||||
// Shortcut
|
||||
std::vector<std::vector<int> > mShortcutTargets;
|
||||
std::vector<int> mShortcutProbabilities;
|
||||
};
|
||||
} // namespace latinime
|
||||
#endif // LATINIME_UNIGRAM_PROPERTY_H
|
|
@ -18,6 +18,7 @@
|
|||
#define LATINIME_DICTIONARY_STRUCTURE_POLICY_H
|
||||
|
||||
#include "defines.h"
|
||||
#include "suggest/core/dictionary/unigram_property.h"
|
||||
#include "utils/exclusive_ownership_pointer.h"
|
||||
|
||||
namespace latinime {
|
||||
|
@ -90,6 +91,10 @@ class DictionaryStructureWithBufferPolicy {
|
|||
virtual void getProperty(const char *const query, const int queryLength, char *const outResult,
|
||||
const int maxResultLength) = 0;
|
||||
|
||||
// Used for testing.
|
||||
virtual const UnigramProperty getUnigramProperty(const int *const codePonts,
|
||||
const int codePointCount) const = 0;
|
||||
|
||||
protected:
|
||||
DictionaryStructureWithBufferPolicy() {}
|
||||
|
||||
|
|
|
@ -123,6 +123,12 @@ class PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
|
|||
}
|
||||
}
|
||||
|
||||
const UnigramProperty getUnigramProperty(const int *const codePoints,
|
||||
const int codePointCount) const {
|
||||
// getUnigramProperty is not supported.
|
||||
return UnigramProperty();
|
||||
}
|
||||
|
||||
private:
|
||||
DISALLOW_IMPLICIT_CONSTRUCTORS(PatriciaTriePolicy);
|
||||
|
||||
|
|
|
@ -16,8 +16,11 @@
|
|||
|
||||
#include "suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_policy.h"
|
||||
|
||||
#include <vector>
|
||||
|
||||
#include "suggest/core/dicnode/dic_node.h"
|
||||
#include "suggest/core/dicnode/dic_node_vector.h"
|
||||
#include "suggest/core/dictionary/unigram_property.h"
|
||||
#include "suggest/policyimpl/dictionary/structure/v3/dynamic_patricia_trie_reading_helper.h"
|
||||
#include "suggest/policyimpl/dictionary/structure/v4/ver4_patricia_trie_node_reader.h"
|
||||
#include "suggest/policyimpl/dictionary/utils/forgetting_curve_utils.h"
|
||||
|
@ -290,4 +293,42 @@ void Ver4PatriciaTriePolicy::getProperty(const char *const query, const int quer
|
|||
}
|
||||
}
|
||||
|
||||
const UnigramProperty Ver4PatriciaTriePolicy::getUnigramProperty(const int *const codePoints,
|
||||
const int codePointCount) const {
|
||||
const int ptNodePos = getTerminalPtNodePositionOfWord(codePoints, codePointCount,
|
||||
false /* forceLowerCaseSearch */);
|
||||
if (ptNodePos == NOT_A_DICT_POS) {
|
||||
AKLOGE("fetchUnigramProperty is called for invalid word.");
|
||||
return UnigramProperty();
|
||||
}
|
||||
const PtNodeParams ptNodeParams = mNodeReader.fetchNodeInfoInBufferFromPtNodePos(ptNodePos);
|
||||
const ProbabilityEntry probabilityEntry =
|
||||
mBuffers.get()->getProbabilityDictContent()->getProbabilityEntry(
|
||||
ptNodeParams.getTerminalId());
|
||||
// Fetch shortcut information.
|
||||
std::vector<std::vector<int> > shortcutTargets;
|
||||
std::vector<int> shortcutProbabilities;
|
||||
if (ptNodeParams.hasShortcutTargets()) {
|
||||
int shortcutTarget[MAX_WORD_LENGTH];
|
||||
const ShortcutDictContent *const shortcutDictContent =
|
||||
mBuffers.get()->getShortcutDictContent();
|
||||
bool hasNext = true;
|
||||
int shortcutPos = getShortcutPositionOfPtNode(ptNodePos);
|
||||
while (hasNext) {
|
||||
int shortcutTargetLength = 0;
|
||||
int shortcutProbability = NOT_A_PROBABILITY;
|
||||
shortcutDictContent->getShortcutEntryAndAdvancePosition(MAX_WORD_LENGTH, shortcutTarget,
|
||||
&shortcutTargetLength, &shortcutProbability, &hasNext, &shortcutPos);
|
||||
std::vector<int> target(shortcutTarget, shortcutTarget + shortcutTargetLength);
|
||||
shortcutTargets.push_back(target);
|
||||
shortcutProbabilities.push_back(shortcutProbability);
|
||||
}
|
||||
}
|
||||
return UnigramProperty(ptNodeParams.getCodePoints(), ptNodeParams.getCodePointCount(),
|
||||
ptNodeParams.isNotAWord(), ptNodeParams.isBlacklisted(), ptNodeParams.hasBigrams(),
|
||||
ptNodeParams.hasShortcutTargets(), ptNodeParams.getProbability(),
|
||||
probabilityEntry.getTimeStamp(), probabilityEntry.getLevel(),
|
||||
probabilityEntry.getCount(), &shortcutTargets, &shortcutProbabilities);
|
||||
}
|
||||
|
||||
} // namespace latinime
|
||||
|
|
|
@ -107,6 +107,9 @@ class Ver4PatriciaTriePolicy : public DictionaryStructureWithBufferPolicy {
|
|||
void getProperty(const char *const query, const int queryLength, char *const outResult,
|
||||
const int maxResultLength);
|
||||
|
||||
const UnigramProperty getUnigramProperty(const int *const codePoints,
|
||||
const int codePointCount) const;
|
||||
|
||||
private:
|
||||
DISALLOW_IMPLICIT_CONSTRUCTORS(Ver4PatriciaTriePolicy);
|
||||
|
||||
|
|
|
@ -24,6 +24,7 @@ import android.util.Pair;
|
|||
import com.android.inputmethod.latin.BinaryDictionary.LanguageModelParam;
|
||||
import com.android.inputmethod.latin.makedict.CodePointUtils;
|
||||
import com.android.inputmethod.latin.makedict.FormatSpec;
|
||||
import com.android.inputmethod.latin.utils.UnigramProperty;
|
||||
|
||||
import java.io.File;
|
||||
import java.io.IOException;
|
||||
|
@ -824,4 +825,52 @@ public class BinaryDictionaryTests extends AndroidTestCase {
|
|||
assertEquals(probability, binaryDictionary.getBigramProbability(word0, word1));
|
||||
}
|
||||
}
|
||||
|
||||
public void testGetUnigramProperties() {
|
||||
testGetUnigramProperties(4 /* formatVersion */);
|
||||
}
|
||||
|
||||
private void testGetUnigramProperties(final int formatVersion) {
|
||||
final long seed = System.currentTimeMillis();
|
||||
final Random random = new Random(seed);
|
||||
final int ITERATION_COUNT = 1000;
|
||||
final int codePointSetSize = 20;
|
||||
final int[] codePointSet = CodePointUtils.generateCodePointSet(codePointSetSize, random);
|
||||
|
||||
File dictFile = null;
|
||||
try {
|
||||
dictFile = createEmptyDictionaryAndGetFile("TestBinaryDictionary", formatVersion);
|
||||
} catch (IOException e) {
|
||||
fail("IOException while writing an initial dictionary : " + e);
|
||||
}
|
||||
final BinaryDictionary binaryDictionary = new BinaryDictionary(dictFile.getAbsolutePath(),
|
||||
0 /* offset */, dictFile.length(), true /* useFullEditDistance */,
|
||||
Locale.getDefault(), TEST_LOCALE, true /* isUpdatable */);
|
||||
|
||||
final UnigramProperty invalidUnigramProperty =
|
||||
binaryDictionary.getUnigramProperty("dummyWord");
|
||||
assertFalse(invalidUnigramProperty.isValid());
|
||||
|
||||
for (int i = 0; i < ITERATION_COUNT; i++) {
|
||||
final String word = CodePointUtils.generateWord(random, codePointSet);
|
||||
final int unigramProbability = random.nextInt(0xFF);
|
||||
final boolean isNotAWord = random.nextBoolean();
|
||||
final boolean isBlacklisted = random.nextBoolean();
|
||||
// TODO: Add tests for shortcut.
|
||||
// TODO: Add tests for historical info.
|
||||
binaryDictionary.addUnigramWord(word, unigramProbability,
|
||||
null /* shortcutTarget */, BinaryDictionary.NOT_A_PROBABILITY,
|
||||
isNotAWord, isBlacklisted, BinaryDictionary.NOT_A_VALID_TIMESTAMP);
|
||||
final UnigramProperty unigramProperty =
|
||||
binaryDictionary.getUnigramProperty(word);
|
||||
assertEquals(word, unigramProperty.mCodePoints);
|
||||
assertTrue(unigramProperty.isValid());
|
||||
assertEquals(isNotAWord, unigramProperty.mIsNotAWord);
|
||||
assertEquals(isBlacklisted, unigramProperty.mIsBlacklisted);
|
||||
assertEquals(false, unigramProperty.mHasBigrams);
|
||||
assertEquals(false, unigramProperty.mHasShortcuts);
|
||||
assertEquals(unigramProbability, unigramProperty.mProbability);
|
||||
assertTrue(unigramProperty.mShortcutTargets.isEmpty());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
Loading…
Reference in a new issue