Add a JNI to get the frequency

Bug: 4192129
Change-Id: I3f220f5a10114d4eb23956148076cf76220bda0f

4 files changed, 19 insertions, 6 deletions

diff --git a/native/jni/src/dictionary.cpp b/native/jni/src/dictionary.cpp
index 65d0f73a3..1fb02478b 100644
--- a/native/jni/src/dictionary.cpp
+++ b/native/jni/src/dictionary.cpp
@@ -55,8 +55,8 @@ Dictionary::~Dictionary() {
     delete mBigramDictionary;
 }
 
-bool Dictionary::isValidWord(const int32_t *word, int length) {
-    return mUnigramDictionary->isValidWord(word, length);
+int Dictionary::getFrequency(const int32_t *word, int length) {
+    return mUnigramDictionary->getFrequency(word, length);
 }
 
 bool Dictionary::isValidBigram(const int32_t *word1, int length1, const int32_t *word2,
diff --git a/native/jni/src/dictionary.h b/native/jni/src/dictionary.h
index 87891ee4d..9f2367904 100644
--- a/native/jni/src/dictionary.h
+++ b/native/jni/src/dictionary.h
@@ -52,7 +52,7 @@ class Dictionary {
                 maxWordLength, maxBigrams);
     }
 
-    bool isValidWord(const int32_t *word, int length);
+    int getFrequency(const int32_t *word, int length);
     bool isValidBigram(const int32_t *word1, int length1, const int32_t *word2, int length2);
     void *getDict() { return (void *)mDict; }
     int getDictSize() { return mDictSize; }
diff --git a/native/jni/src/unigram_dictionary.cpp b/native/jni/src/unigram_dictionary.cpp
index 828582848..efe9c4fe3 100644
--- a/native/jni/src/unigram_dictionary.cpp
+++ b/native/jni/src/unigram_dictionary.cpp
@@ -747,8 +747,21 @@ int UnigramDictionary::getMostFrequentWordLikeInner(const uint16_t * const inWor
     return maxFreq;
 }
 
-bool UnigramDictionary::isValidWord(const int32_t* const inWord, const int length) const {
-    return NOT_VALID_WORD != BinaryFormat::getTerminalPosition(DICT_ROOT, inWord, length);
+int UnigramDictionary::getFrequency(const int32_t* const inWord, const int length) const {
+    const uint8_t* const root = DICT_ROOT;
+    int pos = BinaryFormat::getTerminalPosition(root, inWord, length);
+    if (NOT_VALID_WORD == pos) {
+        return NOT_A_PROBABILITY;
+    }
+    const uint8_t flags = BinaryFormat::getFlagsAndForwardPointer(root, &pos);
+    const bool hasMultipleChars = (0 != (FLAG_HAS_MULTIPLE_CHARS & flags));
+    if (hasMultipleChars) {
+        pos = BinaryFormat::skipOtherCharacters(root, pos);
+    } else {
+        BinaryFormat::getCharCodeAndForwardPointer(DICT_ROOT, &pos);
+    }
+    const int unigramFreq = BinaryFormat::readFrequencyWithoutMovingPointer(root, pos);
+    return unigramFreq;
 }
 
 // TODO: remove this function.
diff --git a/native/jni/src/unigram_dictionary.h b/native/jni/src/unigram_dictionary.h
index b9233518f..b70894004 100644
--- a/native/jni/src/unigram_dictionary.h
+++ b/native/jni/src/unigram_dictionary.h
@@ -72,7 +72,7 @@ class UnigramDictionary {
 
     UnigramDictionary(const uint8_t* const streamStart, int typedLetterMultipler,
             int fullWordMultiplier, int maxWordLength, int maxWords, const unsigned int flags);
-    bool isValidWord(const int32_t* const inWord, const int length) const;
+    int getFrequency(const int32_t* const inWord, const int length) const;
     int getBigramPosition(int pos, unsigned short *word, int offset, int length) const;
     int getSuggestions(ProximityInfo *proximityInfo, WordsPriorityQueuePool *queuePool,
             Correction *correction, const int *xcoordinates, const int *ycoordinates,