Merge "Extend jni interface to support time stamp and shortcut."
This commit is contained in:
commit
fd15afae46
2 changed files with 42 additions and 8 deletions
|
@ -57,6 +57,8 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
@UsedForTesting
|
@UsedForTesting
|
||||||
public static final String MAX_BIGRAM_COUNT_QUERY = "MAX_BIGRAM_COUNT";
|
public static final String MAX_BIGRAM_COUNT_QUERY = "MAX_BIGRAM_COUNT";
|
||||||
|
|
||||||
|
private static final int NOT_A_VALID_TIME_STAMP = -1;
|
||||||
|
|
||||||
private long mNativeDict;
|
private long mNativeDict;
|
||||||
private final Locale mLocale;
|
private final Locale mLocale;
|
||||||
private final long mDictSize;
|
private final long mDictSize;
|
||||||
|
@ -133,9 +135,11 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
int[] outputAutoCommitFirstWordConfidence);
|
int[] outputAutoCommitFirstWordConfidence);
|
||||||
private static native float calcNormalizedScoreNative(int[] before, int[] after, int score);
|
private static native float calcNormalizedScoreNative(int[] before, int[] after, int score);
|
||||||
private static native int editDistanceNative(int[] before, int[] after);
|
private static native int editDistanceNative(int[] before, int[] after);
|
||||||
private static native void addUnigramWordNative(long dict, int[] word, int probability);
|
private static native void addUnigramWordNative(long dict, int[] word, int probability,
|
||||||
|
int[] shortcutTarget, int shortcutProbability, boolean isNotAWord,
|
||||||
|
boolean isBlacklisted, int timeStamp);
|
||||||
private static native void addBigramWordsNative(long dict, int[] word0, int[] word1,
|
private static native void addBigramWordsNative(long dict, int[] word0, int[] word1,
|
||||||
int probability);
|
int probability, int timeStamp);
|
||||||
private static native void removeBigramWordsNative(long dict, int[] word0, int[] word1);
|
private static native void removeBigramWordsNative(long dict, int[] word0, int[] word1);
|
||||||
private static native int addMultipleDictionaryEntriesNative(long dict,
|
private static native int addMultipleDictionaryEntriesNative(long dict,
|
||||||
LanguageModelParam[] languageModelParams, int startIndex);
|
LanguageModelParam[] languageModelParams, int startIndex);
|
||||||
|
@ -282,7 +286,10 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
final int[] codePoints = StringUtils.toCodePointArray(word);
|
final int[] codePoints = StringUtils.toCodePointArray(word);
|
||||||
addUnigramWordNative(mNativeDict, codePoints, probability);
|
final int[] shortcutTarget = new int[0];
|
||||||
|
addUnigramWordNative(mNativeDict, codePoints, probability, shortcutTarget,
|
||||||
|
NOT_A_PROBABILITY, false /* isNotAWord */, false /* isBlacklisted */,
|
||||||
|
NOT_A_VALID_TIME_STAMP);
|
||||||
}
|
}
|
||||||
|
|
||||||
// Add a bigram entry to binary dictionary in native code.
|
// Add a bigram entry to binary dictionary in native code.
|
||||||
|
@ -292,7 +299,8 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
}
|
}
|
||||||
final int[] codePoints0 = StringUtils.toCodePointArray(word0);
|
final int[] codePoints0 = StringUtils.toCodePointArray(word0);
|
||||||
final int[] codePoints1 = StringUtils.toCodePointArray(word1);
|
final int[] codePoints1 = StringUtils.toCodePointArray(word1);
|
||||||
addBigramWordsNative(mNativeDict, codePoints0, codePoints1, probability);
|
addBigramWordsNative(mNativeDict, codePoints0, codePoints1, probability,
|
||||||
|
NOT_A_VALID_TIME_STAMP);
|
||||||
}
|
}
|
||||||
|
|
||||||
// Remove a bigram entry form binary dictionary in native code.
|
// Remove a bigram entry form binary dictionary in native code.
|
||||||
|
@ -308,15 +316,25 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
public static class LanguageModelParam {
|
public static class LanguageModelParam {
|
||||||
public final int[] mWord0;
|
public final int[] mWord0;
|
||||||
public final int[] mWord1;
|
public final int[] mWord1;
|
||||||
|
public final int[] mShortcutTarget;
|
||||||
public final int mUnigramProbability;
|
public final int mUnigramProbability;
|
||||||
public final int mBigramProbability;
|
public final int mBigramProbability;
|
||||||
|
public final int mShortcutProbability;
|
||||||
|
public final boolean mIsNotAWord;
|
||||||
|
public final boolean mIsBlacklisted;
|
||||||
|
public final int mTimeStamp;
|
||||||
|
|
||||||
// Constructor for unigram.
|
// Constructor for unigram.
|
||||||
public LanguageModelParam(final String word, final int unigramProbability) {
|
public LanguageModelParam(final String word, final int unigramProbability) {
|
||||||
mWord0 = null;
|
mWord0 = null;
|
||||||
mWord1 = StringUtils.toCodePointArray(word);
|
mWord1 = StringUtils.toCodePointArray(word);
|
||||||
|
mShortcutTarget = null;
|
||||||
mUnigramProbability = unigramProbability;
|
mUnigramProbability = unigramProbability;
|
||||||
mBigramProbability = NOT_A_PROBABILITY;
|
mBigramProbability = NOT_A_PROBABILITY;
|
||||||
|
mShortcutProbability = NOT_A_PROBABILITY;
|
||||||
|
mIsNotAWord = false;
|
||||||
|
mIsBlacklisted = false;
|
||||||
|
mTimeStamp = NOT_A_VALID_TIME_STAMP;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Constructor for unigram and bigram.
|
// Constructor for unigram and bigram.
|
||||||
|
@ -324,8 +342,13 @@ public final class BinaryDictionary extends Dictionary {
|
||||||
final int unigramProbability, final int bigramProbability) {
|
final int unigramProbability, final int bigramProbability) {
|
||||||
mWord0 = StringUtils.toCodePointArray(word0);
|
mWord0 = StringUtils.toCodePointArray(word0);
|
||||||
mWord1 = StringUtils.toCodePointArray(word1);
|
mWord1 = StringUtils.toCodePointArray(word1);
|
||||||
|
mShortcutTarget = null;
|
||||||
mUnigramProbability = unigramProbability;
|
mUnigramProbability = unigramProbability;
|
||||||
mBigramProbability = bigramProbability;
|
mBigramProbability = bigramProbability;
|
||||||
|
mShortcutProbability = NOT_A_PROBABILITY;
|
||||||
|
mIsNotAWord = false;
|
||||||
|
mIsBlacklisted = false;
|
||||||
|
mTimeStamp = NOT_A_VALID_TIME_STAMP;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
@ -277,7 +277,8 @@ static jint latinime_BinaryDictionary_editDistance(JNIEnv *env, jclass clazz, ji
|
||||||
}
|
}
|
||||||
|
|
||||||
static void latinime_BinaryDictionary_addUnigramWord(JNIEnv *env, jclass clazz, jlong dict,
|
static void latinime_BinaryDictionary_addUnigramWord(JNIEnv *env, jclass clazz, jlong dict,
|
||||||
jintArray word, jint probability) {
|
jintArray word, jint probability, jintArray shortcutTarget, jint shortuctProbability,
|
||||||
|
jboolean isNotAWord, jboolean isBlacklisted, jint timeStamp) {
|
||||||
Dictionary *dictionary = reinterpret_cast<Dictionary *>(dict);
|
Dictionary *dictionary = reinterpret_cast<Dictionary *>(dict);
|
||||||
if (!dictionary) {
|
if (!dictionary) {
|
||||||
return;
|
return;
|
||||||
|
@ -285,11 +286,16 @@ static void latinime_BinaryDictionary_addUnigramWord(JNIEnv *env, jclass clazz,
|
||||||
jsize wordLength = env->GetArrayLength(word);
|
jsize wordLength = env->GetArrayLength(word);
|
||||||
int codePoints[wordLength];
|
int codePoints[wordLength];
|
||||||
env->GetIntArrayRegion(word, 0, wordLength, codePoints);
|
env->GetIntArrayRegion(word, 0, wordLength, codePoints);
|
||||||
|
jsize shortcutLength = shortcutTarget ? env->GetArrayLength(shortcutTarget) : 0;
|
||||||
|
int shortcutTargetCodePoints[wordLength];
|
||||||
|
if (shortcutTarget) {
|
||||||
|
env->GetIntArrayRegion(shortcutTarget, 0, shortcutLength, shortcutTargetCodePoints);
|
||||||
|
}
|
||||||
dictionary->addUnigramWord(codePoints, wordLength, probability);
|
dictionary->addUnigramWord(codePoints, wordLength, probability);
|
||||||
}
|
}
|
||||||
|
|
||||||
static void latinime_BinaryDictionary_addBigramWords(JNIEnv *env, jclass clazz, jlong dict,
|
static void latinime_BinaryDictionary_addBigramWords(JNIEnv *env, jclass clazz, jlong dict,
|
||||||
jintArray word0, jintArray word1, jint probability) {
|
jintArray word0, jintArray word1, jint probability, jint timeStamp) {
|
||||||
Dictionary *dictionary = reinterpret_cast<Dictionary *>(dict);
|
Dictionary *dictionary = reinterpret_cast<Dictionary *>(dict);
|
||||||
if (!dictionary) {
|
if (!dictionary) {
|
||||||
return;
|
return;
|
||||||
|
@ -335,6 +341,8 @@ static int latinime_BinaryDictionary_addMultipleDictionaryEntries(JNIEnv *env, j
|
||||||
jobject languageModelParam = env->GetObjectArrayElement(languageModelParams, 0);
|
jobject languageModelParam = env->GetObjectArrayElement(languageModelParams, 0);
|
||||||
jclass languageModelParamClass = env->GetObjectClass(languageModelParam);
|
jclass languageModelParamClass = env->GetObjectClass(languageModelParam);
|
||||||
env->DeleteLocalRef(languageModelParam);
|
env->DeleteLocalRef(languageModelParam);
|
||||||
|
|
||||||
|
// TODO: Support shortcut, timestamp and flags.
|
||||||
jfieldID word0FieldId = env->GetFieldID(languageModelParamClass, "mWord0", "[I");
|
jfieldID word0FieldId = env->GetFieldID(languageModelParamClass, "mWord0", "[I");
|
||||||
jfieldID word1FieldId = env->GetFieldID(languageModelParamClass, "mWord1", "[I");
|
jfieldID word1FieldId = env->GetFieldID(languageModelParamClass, "mWord1", "[I");
|
||||||
jfieldID unigramProbabilityFieldId =
|
jfieldID unigramProbabilityFieldId =
|
||||||
|
@ -345,6 +353,9 @@ static int latinime_BinaryDictionary_addMultipleDictionaryEntries(JNIEnv *env, j
|
||||||
|
|
||||||
for (int i = startIndex; i < languageModelParamCount; ++i) {
|
for (int i = startIndex; i < languageModelParamCount; ++i) {
|
||||||
jobject languageModelParam = env->GetObjectArrayElement(languageModelParams, i);
|
jobject languageModelParam = env->GetObjectArrayElement(languageModelParams, i);
|
||||||
|
// languageModelParam is a set of params for word1; thus, word1 cannot be null. On the
|
||||||
|
// other hand, word0 can be null and then it means the set of params doesn't contain bigram
|
||||||
|
// information.
|
||||||
jintArray word0 = static_cast<jintArray>(
|
jintArray word0 = static_cast<jintArray>(
|
||||||
env->GetObjectField(languageModelParam, word0FieldId));
|
env->GetObjectField(languageModelParam, word0FieldId));
|
||||||
jsize word0Length = word0 ? env->GetArrayLength(word0) : 0;
|
jsize word0Length = word0 ? env->GetArrayLength(word0) : 0;
|
||||||
|
@ -459,12 +470,12 @@ static const JNINativeMethod sMethods[] = {
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
const_cast<char *>("addUnigramWordNative"),
|
const_cast<char *>("addUnigramWordNative"),
|
||||||
const_cast<char *>("(J[II)V"),
|
const_cast<char *>("(J[II[IIZZI)V"),
|
||||||
reinterpret_cast<void *>(latinime_BinaryDictionary_addUnigramWord)
|
reinterpret_cast<void *>(latinime_BinaryDictionary_addUnigramWord)
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
const_cast<char *>("addBigramWordsNative"),
|
const_cast<char *>("addBigramWordsNative"),
|
||||||
const_cast<char *>("(J[I[II)V"),
|
const_cast<char *>("(J[I[III)V"),
|
||||||
reinterpret_cast<void *>(latinime_BinaryDictionary_addBigramWords)
|
reinterpret_cast<void *>(latinime_BinaryDictionary_addBigramWords)
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
|
|
Loading…
Reference in a new issue