Merge "Extend jni methods and enable Beginning-of-Sentence."

4 files changed, 39 insertions, 21 deletions

diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
index e7ab02ac1..b77540622 100644
--- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java
+++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
@@ -191,7 +191,8 @@ public final class BinaryDictionary extends Dictionary {
     private static native void closeNative(long dict);
     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 int getBigramProbabilityNative(long dict, int[] word0,
+            boolean isBeginningOfSentence, int[] word1);
     private static native void getWordPropertyNative(long dict, int[] word,
             int[] outCodePoints, boolean[] outFlags, int[] outProbabilityInfo,
             ArrayList<int[]> outBigramTargets, ArrayList<int[]> outBigramProbabilityInfo,
@@ -200,15 +201,17 @@ public final class BinaryDictionary extends Dictionary {
     private static native void getSuggestionsNative(long dict, long proximityInfo,
             long traverseSession, int[] xCoordinates, int[] yCoordinates, int[] times,
             int[] pointerIds, int[] inputCodePoints, int inputSize, int[] suggestOptions,
-            int[] prevWordCodePointArray, int[] outputSuggestionCount, int[] outputCodePoints,
-            int[] outputScores, int[] outputIndices, int[] outputTypes,
-            int[] outputAutoCommitFirstWordConfidence, float[] inOutLanguageWeight);
+            int[] prevWordCodePointArray, boolean isBeginningOfSentence,
+            int[] outputSuggestionCount, int[] outputCodePoints, int[] outputScores,
+            int[] outputIndices, int[] outputTypes, int[] outputAutoCommitFirstWordConfidence,
+            float[] inOutLanguageWeight);
     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,
-            int probability, int timestamp);
-    private static native void removeBigramWordsNative(long dict, int[] word0, int[] word1);
+            int[] shortcutTarget, int shortcutProbability, boolean isBeginningOfSentence,
+            boolean isNotAWord, boolean isBlacklisted, int timestamp);
+    private static native void addBigramWordsNative(long dict, int[] word0,
+            boolean isBeginningOfSentence, int[] word1, int probability, int timestamp);
+    private static native void removeBigramWordsNative(long dict, int[] word0,
+            boolean isBeginningOfSentence, int[] word1);
     private static native int addMultipleDictionaryEntriesNative(long dict,
             LanguageModelParam[] languageModelParams, int startIndex);
     private static native String getPropertyNative(long dict, String query);
@@ -301,7 +304,8 @@ public final class BinaryDictionary extends Dictionary {
                 getTraverseSession(sessionId).getSession(), inputPointers.getXCoordinates(),
                 inputPointers.getYCoordinates(), inputPointers.getTimes(),
                 inputPointers.getPointerIds(), mInputCodePoints, inputSize,
-                mNativeSuggestOptions.getOptions(), prevWordCodePointArray, mOutputSuggestionCount,
+                mNativeSuggestOptions.getOptions(), prevWordCodePointArray,
+                prevWordsInfo.mIsBeginningOfSentence, mOutputSuggestionCount,
                 mOutputCodePoints, mOutputScores, mSpaceIndices, mOutputTypes,
                 mOutputAutoCommitFirstWordConfidence, mInputOutputLanguageWeight);
         if (inOutLanguageWeight != null) {
@@ -364,12 +368,13 @@ public final class BinaryDictionary extends Dictionary {
     }
 
     public int getNgramProbability(final PrevWordsInfo prevWordsInfo, final String word) {
-        if (TextUtils.isEmpty(prevWordsInfo.mPrevWord) || TextUtils.isEmpty(word)) {
+        if (!prevWordsInfo.isValid() || TextUtils.isEmpty(word)) {
             return NOT_A_PROBABILITY;
         }
         final int[] codePoints0 = StringUtils.toCodePointArray(prevWordsInfo.mPrevWord);
         final int[] codePoints1 = StringUtils.toCodePointArray(word);
-        return getBigramProbabilityNative(mNativeDict, codePoints0, codePoints1);
+        return getBigramProbabilityNative(mNativeDict, codePoints0,
+                prevWordsInfo.mIsBeginningOfSentence, codePoints1);
     }
 
     public WordProperty getWordProperty(final String word) {
@@ -420,16 +425,17 @@ public final class BinaryDictionary extends Dictionary {
 
     // Add a unigram entry to binary dictionary with unigram attributes in native code.
     public void addUnigramEntry(final String word, final int probability,
-            final String shortcutTarget, final int shortcutProbability, final boolean isNotAWord,
+            final String shortcutTarget, final int shortcutProbability,
+            final boolean isBeginningOfSentence, final boolean isNotAWord,
             final boolean isBlacklisted, final int timestamp) {
-        if (TextUtils.isEmpty(word)) {
+        if (word == null || (word.isEmpty() && !isBeginningOfSentence)) {
             return;
         }
         final int[] codePoints = StringUtils.toCodePointArray(word);
         final int[] shortcutTargetCodePoints = (shortcutTarget != null) ?
                 StringUtils.toCodePointArray(shortcutTarget) : null;
         addUnigramWordNative(mNativeDict, codePoints, probability, shortcutTargetCodePoints,
-                shortcutProbability, isNotAWord, isBlacklisted, timestamp);
+                shortcutProbability, isBeginningOfSentence, isNotAWord, isBlacklisted, timestamp);
         mHasUpdated = true;
     }
 
@@ -437,23 +443,25 @@ public final class BinaryDictionary extends Dictionary {
     public void addNgramEntry(final PrevWordsInfo prevWordsInfo, final String word,
             final int probability,
             final int timestamp) {
-        if (TextUtils.isEmpty(prevWordsInfo.mPrevWord) || TextUtils.isEmpty(word)) {
+        if (!prevWordsInfo.isValid() || TextUtils.isEmpty(word)) {
             return;
         }
         final int[] codePoints0 = StringUtils.toCodePointArray(prevWordsInfo.mPrevWord);
         final int[] codePoints1 = StringUtils.toCodePointArray(word);
-        addBigramWordsNative(mNativeDict, codePoints0, codePoints1, probability, timestamp);
+        addBigramWordsNative(mNativeDict, codePoints0, prevWordsInfo.mIsBeginningOfSentence,
+                codePoints1, probability, timestamp);
         mHasUpdated = true;
     }
 
     // Remove an n-gram entry from the binary dictionary in native code.
     public void removeNgramEntry(final PrevWordsInfo prevWordsInfo, final String word) {
-        if (TextUtils.isEmpty(prevWordsInfo.mPrevWord) || TextUtils.isEmpty(word)) {
+        if (!prevWordsInfo.isValid() || TextUtils.isEmpty(word)) {
             return;
         }
         final int[] codePoints0 = StringUtils.toCodePointArray(prevWordsInfo.mPrevWord);
         final int[] codePoints1 = StringUtils.toCodePointArray(word);
-        removeBigramWordsNative(mNativeDict, codePoints0, codePoints1);
+        removeBigramWordsNative(mNativeDict, codePoints0, prevWordsInfo.mIsBeginningOfSentence,
+                codePoints1);
         mHasUpdated = true;
     }
 
diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
index d67253c3b..334ebb37d 100644
--- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
+++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java
@@ -292,7 +292,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary {
             final String shortcutTarget, final int shortcutFreq, final boolean isNotAWord,
             final boolean isBlacklisted, final int timestamp) {
         mBinaryDictionary.addUnigramEntry(word, frequency, shortcutTarget, shortcutFreq,
-                isNotAWord, isBlacklisted, timestamp);
+                false /* isBeginningOfSentence */, isNotAWord, isBlacklisted, timestamp);
     }
 
     /**
diff --git a/java/src/com/android/inputmethod/latin/PrevWordsInfo.java b/java/src/com/android/inputmethod/latin/PrevWordsInfo.java
index ecc8947db..3494d16f7 100644
--- a/java/src/com/android/inputmethod/latin/PrevWordsInfo.java
+++ b/java/src/com/android/inputmethod/latin/PrevWordsInfo.java
@@ -20,6 +20,8 @@ import android.util.Log;
 
 // TODO: Support multiple previous words for n-gram.
 public class PrevWordsInfo {
+    public static final PrevWordsInfo BEGINNING_OF_SENTENCE = new PrevWordsInfo();
+
     // The previous word. May be null after resetting and before starting a new composing word, or
     // when there is no context like at the start of text for example. It can also be set to null
     // externally when the user enters a separator that does not let bigrams across, like a period
@@ -32,7 +34,7 @@ public class PrevWordsInfo {
 
     // Beginning of sentence.
     public PrevWordsInfo() {
-        mPrevWord = null;
+        mPrevWord = "";
         mIsBeginningOfSentence = true;
     }
 
@@ -40,4 +42,8 @@ public class PrevWordsInfo {
         mPrevWord = prevWord;
         mIsBeginningOfSentence = false;
     }
+
+    public boolean isValid() {
+        return mPrevWord != null;
+    }
 }
diff --git a/java/src/com/android/inputmethod/latin/makedict/WordProperty.java b/java/src/com/android/inputmethod/latin/makedict/WordProperty.java
index 853392200..ed832510c 100644
--- a/java/src/com/android/inputmethod/latin/makedict/WordProperty.java
+++ b/java/src/com/android/inputmethod/latin/makedict/WordProperty.java
@@ -35,6 +35,8 @@ public final class WordProperty implements Comparable<WordProperty> {
     public final ProbabilityInfo mProbabilityInfo;
     public final ArrayList<WeightedString> mShortcutTargets;
     public final ArrayList<WeightedString> mBigrams;
+    // TODO: Support mIsBeginningOfSentence.
+    public final boolean mIsBeginningOfSentence;
     public final boolean mIsNotAWord;
     public final boolean mIsBlacklistEntry;
     public final boolean mHasShortcuts;
@@ -51,6 +53,7 @@ public final class WordProperty implements Comparable<WordProperty> {
         mProbabilityInfo = probabilityInfo;
         mShortcutTargets = shortcutTargets;
         mBigrams = bigrams;
+        mIsBeginningOfSentence = false;
         mIsNotAWord = isNotAWord;
         mIsBlacklistEntry = isBlacklistEntry;
         mHasBigrams = bigrams != null && !bigrams.isEmpty();
@@ -77,6 +80,7 @@ public final class WordProperty implements Comparable<WordProperty> {
         mProbabilityInfo = createProbabilityInfoFromArray(probabilityInfo);
         mShortcutTargets = CollectionUtils.newArrayList();
         mBigrams = CollectionUtils.newArrayList();
+        mIsBeginningOfSentence = false;
         mIsNotAWord = isNotAWord;
         mIsBlacklistEntry = isBlacklisted;
         mHasShortcuts = hasShortcuts;