aboutsummaryrefslogtreecommitdiffstats
path: root/dictionary
diff options
context:
space:
mode:
Diffstat (limited to 'dictionary')
-rw-r--r--dictionary/jni/com_android_inputmethod_latin_BinaryDictionary.cpp36
-rw-r--r--dictionary/src/dictionary.cpp58
-rw-r--r--dictionary/src/dictionary.h8
3 files changed, 56 insertions, 46 deletions
diff --git a/dictionary/jni/com_android_inputmethod_latin_BinaryDictionary.cpp b/dictionary/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
index a1f410db2..c9e158c4c 100644
--- a/dictionary/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
+++ b/dictionary/jni/com_android_inputmethod_latin_BinaryDictionary.cpp
@@ -42,7 +42,7 @@ static jmethodID sAddWordMethod;
//
// helper function to throw an exception
//
-static void throwException(JNIEnv *env, const char* ex, const char* fmt, int data)
+static void throwException(JNIEnv *env, const char* ex, const char* fmt, int data)
{
if (jclass cls = env->FindClass(ex)) {
char msg[1000];
@@ -66,7 +66,7 @@ static jint latinime_BinaryDictionary_open
Asset *dictAsset = am->openNonAsset(resourcePath, Asset::ACCESS_BUFFER);
if (dictAsset == NULL) {
- LOGE("DICT: Couldn't get asset %s\n", resourcePath);
+ LOGE("DICT: Couldn't get asset %s\n", resourcePath);
env->ReleaseStringUTFChars(resourceString, resourcePath);
return 0;
}
@@ -79,15 +79,15 @@ static jint latinime_BinaryDictionary_open
}
Dictionary *dictionary = new Dictionary(dict, typedLetterMultiplier, fullWordMultiplier);
dictionary->setAsset(dictAsset);
-
+
env->ReleaseStringUTFChars(resourceString, resourcePath);
- return (jint) dictionary;
+ return (jint) dictionary;
}
static int latinime_BinaryDictionary_getSuggestions(
- JNIEnv *env, jobject object, jint dict, jintArray inputArray, jint arraySize,
- jcharArray outputArray, jintArray frequencyArray, jint maxWordLength, jint maxWords,
- jint maxAlternatives)
+ JNIEnv *env, jobject object, jint dict, jintArray inputArray, jint arraySize,
+ jcharArray outputArray, jintArray frequencyArray, jint maxWordLength, jint maxWords,
+ jint maxAlternatives, jint skipPos)
{
Dictionary *dictionary = (Dictionary*) dict;
if (dictionary == NULL)
@@ -96,9 +96,9 @@ static int latinime_BinaryDictionary_getSuggestions(
int *frequencies = env->GetIntArrayElements(frequencyArray, NULL);
int *inputCodes = env->GetIntArrayElements(inputArray, NULL);
jchar *outputChars = env->GetCharArrayElements(outputArray, NULL);
-
+
int count = dictionary->getSuggestions(inputCodes, arraySize, (unsigned short*) outputChars, frequencies,
- maxWordLength, maxWords, maxAlternatives);
+ maxWordLength, maxWords, maxAlternatives, skipPos);
env->ReleaseIntArrayElements(frequencyArray, frequencies, 0);
env->ReleaseIntArrayElements(inputArray, inputCodes, JNI_ABORT);
@@ -112,16 +112,16 @@ static jboolean latinime_BinaryDictionary_isValidWord
{
Dictionary *dictionary = (Dictionary*) dict;
if (dictionary == NULL) return (jboolean) false;
-
+
jchar *word = env->GetCharArrayElements(wordArray, NULL);
jboolean result = dictionary->isValidWord((unsigned short*) word, wordLength);
env->ReleaseCharArrayElements(wordArray, word, JNI_ABORT);
-
+
return result;
}
static void latinime_BinaryDictionary_close
- (JNIEnv *env, jobject object, jint dict)
+ (JNIEnv *env, jobject object, jint dict)
{
Dictionary *dictionary = (Dictionary*) dict;
((Asset*) dictionary->getAsset())->close();
@@ -131,10 +131,10 @@ static void latinime_BinaryDictionary_close
// ----------------------------------------------------------------------------
static JNINativeMethod gMethods[] = {
- {"openNative", "(Landroid/content/res/AssetManager;Ljava/lang/String;II)I",
+ {"openNative", "(Landroid/content/res/AssetManager;Ljava/lang/String;II)I",
(void*)latinime_BinaryDictionary_open},
{"closeNative", "(I)V", (void*)latinime_BinaryDictionary_close},
- {"getSuggestionsNative", "(I[II[C[IIII)I", (void*)latinime_BinaryDictionary_getSuggestions},
+ {"getSuggestionsNative", "(I[II[C[IIIII)I", (void*)latinime_BinaryDictionary_getSuggestions},
{"isValidWordNative", "(I[CI)Z", (void*)latinime_BinaryDictionary_isValidWord}
};
@@ -153,7 +153,7 @@ static int registerNativeMethods(JNIEnv* env, const char* className,
fprintf(stderr, "RegisterNatives failed for '%s'\n", className);
return JNI_FALSE;
}
-
+
return JNI_TRUE;
}
@@ -161,21 +161,21 @@ static int registerNatives(JNIEnv *env)
{
const char* const kClassPathName = "com/android/inputmethod/latin/BinaryDictionary";
jclass clazz;
-
+
clazz = env->FindClass("java/io/FileDescriptor");
if (clazz == NULL) {
LOGE("Can't find %s", "java/io/FileDescriptor");
return -1;
}
sDescriptorField = env->GetFieldID(clazz, "descriptor", "I");
-
+
clazz = env->FindClass("android/content/res/AssetManager");
if (clazz == NULL) {
LOGE("Can't find %s", "java/io/FileDescriptor");
return -1;
}
sAssetManagerNativeField = env->GetFieldID(clazz, "mObject", "I");
-
+
return registerNativeMethods(env,
kClassPathName, gMethods, sizeof(gMethods) / sizeof(gMethods[0]));
}
diff --git a/dictionary/src/dictionary.cpp b/dictionary/src/dictionary.cpp
index b37f4c926..cc711f419 100644
--- a/dictionary/src/dictionary.cpp
+++ b/dictionary/src/dictionary.cpp
@@ -49,11 +49,8 @@ Dictionary::~Dictionary()
}
int Dictionary::getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
- int maxWordLength, int maxWords, int maxAlternatives)
+ int maxWordLength, int maxWords, int maxAlternatives, int skipPos)
{
- memset(frequencies, 0, maxWords * sizeof(*frequencies));
- memset(outWords, 0, maxWords * maxWordLength * sizeof(*outWords));
-
mFrequencies = frequencies;
mOutputChars = outWords;
mInputCodes = codes;
@@ -62,8 +59,10 @@ int Dictionary::getSuggestions(int *codes, int codesSize, unsigned short *outWor
mMaxWordLength = maxWordLength;
mMaxWords = maxWords;
mWords = 0;
+ mSkipPos = skipPos;
+ mMaxEditDistance = mInputLength < 5 ? 2 : mInputLength / 2;
- getWordsRec(0, 0, mInputLength * 3, false, 1, 0);
+ getWordsRec(0, 0, mInputLength * 3, false, 1, 0, 0);
if (DEBUG_DICT) LOGI("Returning %d words", mWords);
return mWords;
@@ -110,7 +109,11 @@ bool
Dictionary::addWord(unsigned short *word, int length, int frequency)
{
word[length] = 0;
- if (DEBUG_DICT) LOGI("Found word = %s, freq = %d : \n", word, frequency);
+ if (DEBUG_DICT) {
+ char s[length + 1];
+ for (int i = 0; i <= length; i++) s[i] = word[i];
+ LOGI("Found word = %s, freq = %d : \n", s, frequency);
+ }
// Find the right insertion point
int insertAt = 0;
@@ -144,16 +147,14 @@ Dictionary::addWord(unsigned short *word, int length, int frequency)
}
unsigned short
-Dictionary::toLowerCase(unsigned short c, const int depth) {
+Dictionary::toLowerCase(unsigned short c) {
if (c < sizeof(BASE_CHARS) / sizeof(BASE_CHARS[0])) {
c = BASE_CHARS[c];
}
- if (depth == 0) {
- if (c >='A' && c <= 'Z') {
- c |= 32;
- } else if (c > 127) {
- c = u_tolower(c);
- }
+ if (c >='A' && c <= 'Z') {
+ c |= 32;
+ } else if (c > 127) {
+ c = u_tolower(c);
}
return c;
}
@@ -178,12 +179,16 @@ Dictionary::sameAsTyped(unsigned short *word, int length)
static char QUOTE = '\'';
void
-Dictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int snr, int inputIndex)
+Dictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int snr, int inputIndex,
+ int diffs)
{
// Optimization: Prune out words that are too long compared to how much was typed.
if (depth > maxDepth) {
return;
}
+ if (diffs > mMaxEditDistance) {
+ return;
+ }
int count = getCount(&pos);
int *currentChars = NULL;
if (mInputLength <= inputIndex) {
@@ -194,7 +199,7 @@ Dictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int s
for (int i = 0; i < count; i++) {
unsigned short c = getChar(&pos);
- unsigned short lowerC = toLowerCase(c, depth);
+ unsigned short lowerC = toLowerCase(c);
bool terminal = getTerminal(&pos);
int childrenAddress = getAddress(&pos);
int freq = 1;
@@ -207,38 +212,41 @@ Dictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int s
}
if (childrenAddress != 0) {
getWordsRec(childrenAddress, depth + 1, maxDepth,
- completion, snr, inputIndex);
+ completion, snr, inputIndex, diffs);
}
- } else if (c == QUOTE && currentChars[0] != QUOTE) {
- // Skip the ' and continue deeper
- mWord[depth] = QUOTE;
+ } else if (c == QUOTE && currentChars[0] != QUOTE || mSkipPos == depth) {
+ // Skip the ' or other letter and continue deeper
+ mWord[depth] = c;
if (childrenAddress != 0) {
- getWordsRec(childrenAddress, depth + 1, maxDepth, false, snr, inputIndex);
+ getWordsRec(childrenAddress, depth + 1, maxDepth, false, snr, inputIndex, diffs);
}
} else {
int j = 0;
while (currentChars[j] > 0) {
- int addedWeight = j == 0 ? mTypedLetterMultiplier : 1;
if (currentChars[j] == lowerC || currentChars[j] == c) {
+ int addedWeight = j == 0 ? mTypedLetterMultiplier : 1;
mWord[depth] = c;
if (mInputLength == inputIndex + 1) {
if (terminal) {
if (//INCLUDE_TYPED_WORD_IF_VALID ||
!sameAsTyped(mWord, depth + 1)) {
- addWord(mWord, depth + 1,
- (freq * snr * addedWeight * mFullWordMultiplier));
+ int finalFreq = freq * snr * addedWeight;
+ if (mSkipPos < 0) finalFreq *= mFullWordMultiplier;
+ addWord(mWord, depth + 1, finalFreq);
}
}
if (childrenAddress != 0) {
getWordsRec(childrenAddress, depth + 1,
- maxDepth, true, snr * addedWeight, inputIndex + 1);
+ maxDepth, true, snr * addedWeight, inputIndex + 1,
+ diffs + (j > 0));
}
} else if (childrenAddress != 0) {
getWordsRec(childrenAddress, depth + 1, maxDepth,
- false, snr * addedWeight, inputIndex + 1);
+ false, snr * addedWeight, inputIndex + 1, diffs + (j > 0));
}
}
j++;
+ if (mSkipPos >= 0) break;
}
}
}
diff --git a/dictionary/src/dictionary.h b/dictionary/src/dictionary.h
index b13e97795..8f195ca9a 100644
--- a/dictionary/src/dictionary.h
+++ b/dictionary/src/dictionary.h
@@ -32,7 +32,7 @@ class Dictionary {
public:
Dictionary(void *dict, int typedLetterMultipler, int fullWordMultiplier);
int getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
- int maxWordLength, int maxWords, int maxAlternatives);
+ int maxWordLength, int maxWords, int maxAlternatives, int skipPos);
bool isValidWord(unsigned short *word, int length);
void setAsset(void *asset) { mAsset = asset; }
void *getAsset() { return mAsset; }
@@ -49,9 +49,9 @@ private:
bool sameAsTyped(unsigned short *word, int length);
bool addWord(unsigned short *word, int length, int frequency);
- unsigned short toLowerCase(unsigned short c, int depth);
+ unsigned short toLowerCase(unsigned short c);
void getWordsRec(int pos, int depth, int maxDepth, bool completion, int frequency,
- int inputIndex);
+ int inputIndex, int diffs);
bool isValidWordRec(int pos, unsigned short *word, int offset, int length);
unsigned char *mDict;
@@ -66,6 +66,8 @@ private:
int mInputLength;
int mMaxAlternatives;
unsigned short mWord[128];
+ int mSkipPos;
+ int mMaxEditDistance;
int mFullWordMultiplier;
int mTypedLetterMultiplier;
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-23 00:11:26 +0000