aboutsummaryrefslogtreecommitdiffstats
path: root/native/jni/src
diff options
context:
space:
mode:
Diffstat (limited to 'native/jni/src')
-rw-r--r--native/jni/src/correction.cpp24
-rw-r--r--native/jni/src/correction.h2
-rw-r--r--native/jni/src/defines.h19
-rw-r--r--native/jni/src/proximity_info.cpp2
-rw-r--r--native/jni/src/proximity_info.h8
-rw-r--r--native/jni/src/proximity_info_state.cpp33
-rw-r--r--native/jni/src/proximity_info_state.h4
-rw-r--r--native/jni/src/proximity_info_state_utils.cpp6
-rw-r--r--native/jni/src/proximity_info_utils.h2
9 files changed, 69 insertions, 31 deletions
diff --git a/native/jni/src/correction.cpp b/native/jni/src/correction.cpp
index d4bd4aa00..b8690eb94 100644
--- a/native/jni/src/correction.cpp
+++ b/native/jni/src/correction.cpp
@@ -214,7 +214,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
bool incremented = false;
if (mLastCharExceeded && mInputIndex == mInputSize - 1) {
// TODO: Do not check the proximity if EditDistance exceeds the threshold
- const ProximityType matchId = mProximityInfoState.getMatchedProximityId(
+ const ProximityType matchId = mProximityInfoState.getProximityType(
mInputIndex, c, true, &proximityIndex);
if (isEquivalentChar(matchId)) {
mLastCharExceeded = false;
@@ -268,7 +268,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
bool secondTransposing = false;
if (mTransposedCount % 2 == 1) {
- if (isEquivalentChar(mProximityInfoState.getMatchedProximityId(
+ if (isEquivalentChar(mProximityInfoState.getProximityType(
mInputIndex - 1, c, false))) {
++mTransposedCount;
secondTransposing = true;
@@ -300,7 +300,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
ProximityType matchedProximityCharId = secondTransposing
? EQUIVALENT_CHAR
- : mProximityInfoState.getMatchedProximityId(
+ : mProximityInfoState.getProximityType(
mInputIndex, c, checkProximityChars, &proximityIndex);
if (UNRELATED_CHAR == matchedProximityCharId
@@ -308,7 +308,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
if (canTryCorrection && mOutputIndex > 0
&& mCorrectionStates[mOutputIndex].mProximityMatching
&& mCorrectionStates[mOutputIndex].mExceeding
- && isEquivalentChar(mProximityInfoState.getMatchedProximityId(
+ && isEquivalentChar(mProximityInfoState.getProximityType(
mInputIndex, mWord[mOutputIndex - 1], false))) {
if (DEBUG_CORRECTION
&& (INPUTLENGTH_FOR_DEBUG <= 0 || INPUTLENGTH_FOR_DEBUG == mInputSize)
@@ -327,7 +327,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
// Here, we are doing something equivalent to matchedProximityCharId,
// but we already know that "excessive char correction" just happened
// so that we just need to check "mProximityCount == 0".
- matchedProximityCharId = mProximityInfoState.getMatchedProximityId(
+ matchedProximityCharId = mProximityInfoState.getProximityType(
mInputIndex, c, mProximityCount == 0, &proximityIndex);
}
}
@@ -344,10 +344,10 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
if (mInputIndex < mInputSize - 1 && mOutputIndex > 0 && mTransposedCount > 0
&& !mCorrectionStates[mOutputIndex].mTransposing
&& mCorrectionStates[mOutputIndex - 1].mTransposing
- && isEquivalentChar(mProximityInfoState.getMatchedProximityId(
+ && isEquivalentChar(mProximityInfoState.getProximityType(
mInputIndex, mWord[mOutputIndex - 1], false))
&& isEquivalentChar(
- mProximityInfoState.getMatchedProximityId(mInputIndex + 1, c, false))) {
+ mProximityInfoState.getProximityType(mInputIndex + 1, c, false))) {
// Conversion t->e
// Example:
// occaisional -> occa sional
@@ -359,7 +359,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
&& !mCorrectionStates[mOutputIndex].mTransposing
&& mCorrectionStates[mOutputIndex - 1].mTransposing
&& isEquivalentChar(
- mProximityInfoState.getMatchedProximityId(mInputIndex - 1, c, false))) {
+ mProximityInfoState.getProximityType(mInputIndex - 1, c, false))) {
// Conversion t->s
// Example:
// chcolate -> chocolate
@@ -371,7 +371,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
&& mCorrectionStates[mOutputIndex].mProximityMatching
&& mCorrectionStates[mOutputIndex].mSkipping
&& isEquivalentChar(
- mProximityInfoState.getMatchedProximityId(mInputIndex - 1, c, false))) {
+ mProximityInfoState.getProximityType(mInputIndex - 1, c, false))) {
// Conversion p->s
// Note: This logic tries saving cases like contrst --> contrast -- "a" is one of
// proximity chars of "s", but it should rather be handled as a skipped char.
@@ -383,7 +383,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
&& mCorrectionStates[mOutputIndex].mSkipping
&& mCorrectionStates[mOutputIndex].mAdditionalProximityMatching
&& isProximityCharOrEquivalentChar(
- mProximityInfoState.getMatchedProximityId(mInputIndex + 1, c, false))) {
+ mProximityInfoState.getProximityType(mInputIndex + 1, c, false))) {
// Conversion s->a
incrementInputIndex();
--mSkippedCount;
@@ -392,7 +392,7 @@ Correction::CorrectionType Correction::processCharAndCalcState(const int c, cons
mDistances[mOutputIndex] = ADDITIONAL_PROXIMITY_CHAR_DISTANCE_INFO;
} else if ((mExceeding || mTransposing) && mInputIndex - 1 < mInputSize
&& isEquivalentChar(
- mProximityInfoState.getMatchedProximityId(mInputIndex + 1, c, false))) {
+ mProximityInfoState.getProximityType(mInputIndex + 1, c, false))) {
// 1.2. Excessive or transpose correction
if (mTransposing) {
++mTransposedCount;
@@ -614,7 +614,7 @@ inline static bool isUpperCase(unsigned short c) {
multiplyIntCapped(matchWeight, &finalFreq);
}
- if (proximityInfoState->getMatchedProximityId(0, word[0], true) == UNRELATED_CHAR) {
+ if (proximityInfoState->getProximityType(0, word[0], true) == UNRELATED_CHAR) {
multiplyRate(FIRST_CHAR_DIFFERENT_DEMOTION_RATE, &finalFreq);
}
diff --git a/native/jni/src/correction.h b/native/jni/src/correction.h
index 34f794d84..f0d62102f 100644
--- a/native/jni/src/correction.h
+++ b/native/jni/src/correction.h
@@ -119,7 +119,7 @@ class Correction {
// proximity info state
void initInputParams(const ProximityInfo *proximityInfo, const int *inputCodes,
const int inputSize, const int *xCoordinates, const int *yCoordinates) {
- mProximityInfoState.initInputParams(0, MAX_POINT_TO_KEY_LENGTH,
+ mProximityInfoState.initInputParams(0, static_cast<float>(MAX_VALUE_FOR_WEIGHTING),
proximityInfo, inputCodes, inputSize, xCoordinates, yCoordinates, 0, 0, false);
}
diff --git a/native/jni/src/defines.h b/native/jni/src/defines.h
index a84dfddec..11dc3d20e 100644
--- a/native/jni/src/defines.h
+++ b/native/jni/src/defines.h
@@ -341,8 +341,9 @@ static inline void prof_out(void) {
#define MAX_DEPTH_MULTIPLIER 3
#define FIRST_WORD_INDEX 0
-// Max Distance between point to key
-#define MAX_POINT_TO_KEY_LENGTH 10000000
+// Max value for length, distance and probability which are used in weighting
+// TODO: Remove
+#define MAX_VALUE_FOR_WEIGHTING 10000000
// The max number of the keys in one keyboard layout
#define MAX_KEY_COUNT_IN_A_KEYBOARD 64
@@ -411,4 +412,18 @@ typedef enum {
A_DOUBLE_LETTER,
A_STRONG_DOUBLE_LETTER
} DoubleLetterLevel;
+
+typedef enum {
+ CT_MATCH,
+ CT_OMISSION,
+ CT_INSERTION,
+ CT_TRANSPOSITION,
+ CT_SUBSTITUTION,
+ CT_SPACE_SUBSTITUTION,
+ CT_SPACE_OMISSION,
+ CT_COMPLETION,
+ CT_TERMINAL,
+ CT_NEW_WORD,
+ CT_NEW_WORD_BIGRAM,
+} CorrectionType;
#endif // LATINIME_DEFINES_H
diff --git a/native/jni/src/proximity_info.cpp b/native/jni/src/proximity_info.cpp
index 81eb0b37d..74b5e0131 100644
--- a/native/jni/src/proximity_info.cpp
+++ b/native/jni/src/proximity_info.cpp
@@ -204,6 +204,6 @@ int ProximityInfo::getKeyKeyDistanceG(const int keyId0, const int keyId1) const
if (keyId0 >= 0 && keyId1 >= 0) {
return mKeyKeyDistancesG[keyId0][keyId1];
}
- return MAX_POINT_TO_KEY_LENGTH;
+ return MAX_VALUE_FOR_WEIGHTING;
}
} // namespace latinime
diff --git a/native/jni/src/proximity_info.h b/native/jni/src/proximity_info.h
index 22bbdf165..57a175d2c 100644
--- a/native/jni/src/proximity_info.h
+++ b/native/jni/src/proximity_info.h
@@ -70,7 +70,7 @@ class ProximityInfo {
int getKeyCenterYOfKeyIdG(int keyId) const;
int getKeyKeyDistanceG(int keyId0, int keyId1) const;
- void AK_FORCE_INLINE initializeProximities(const int *const inputCodes,
+ AK_FORCE_INLINE void initializeProximities(const int *const inputCodes,
const int *const inputXCoordinates, const int *const inputYCoordinates,
const int inputSize, int *allInputCodes) const {
ProximityInfoUtils::initializeProximities(inputCodes, inputXCoordinates, inputYCoordinates,
@@ -79,10 +79,14 @@ class ProximityInfo {
KEY_COUNT, mLocaleStr, &mCodeToKeyMap, allInputCodes);
}
- int AK_FORCE_INLINE getKeyIndexOf(const int c) const {
+ AK_FORCE_INLINE int getKeyIndexOf(const int c) const {
return ProximityInfoUtils::getKeyIndexOf(KEY_COUNT, c, &mCodeToKeyMap);
}
+ AK_FORCE_INLINE bool isCodePointOnKeyboard(const int codePoint) const {
+ return getKeyIndexOf(codePoint) != NOT_AN_INDEX;
+ }
+
private:
DISALLOW_IMPLICIT_CONSTRUCTORS(ProximityInfo);
diff --git a/native/jni/src/proximity_info_state.cpp b/native/jni/src/proximity_info_state.cpp
index d13248caf..00e7ffc6c 100644
--- a/native/jni/src/proximity_info_state.cpp
+++ b/native/jni/src/proximity_info_state.cpp
@@ -170,7 +170,7 @@ float ProximityInfoState::getPointToKeyLength(
return 0.0f;
}
// If the char is not a key on the keyboard then return the max length.
- return MAX_POINT_TO_KEY_LENGTH;
+ return static_cast<float>(MAX_VALUE_FOR_WEIGHTING);
}
float ProximityInfoState::getPointToKeyLength_G(const int inputIndex, const int codePoint) const {
@@ -193,20 +193,20 @@ float ProximityInfoState::getPointToKeyByIdLength(const int inputIndex, const in
// the same position. We want to see if c is in it: if so, then the word contains at that position
// a character close to what the user typed.
// What the user typed is actually the first character of the array.
-// proximityIndex is a pointer to the variable where getMatchedProximityId returns the index of c
+// proximityIndex is a pointer to the variable where getProximityType returns the index of c
// in the proximity chars of the input index.
// Notice : accented characters do not have a proximity list, so they are alone in their list. The
// non-accented version of the character should be considered "close", but not the other keys close
// to the non-accented version.
-ProximityType ProximityInfoState::getMatchedProximityId(const int index, const int c,
+ProximityType ProximityInfoState::getProximityType(const int index, const int codePoint,
const bool checkProximityChars, int *proximityIndex) const {
const int *currentCodePoints = getProximityCodePointsAt(index);
const int firstCodePoint = currentCodePoints[0];
- const int baseLowerC = toBaseLowerCase(c);
+ const int baseLowerC = toBaseLowerCase(codePoint);
// The first char in the array is what user typed. If it matches right away, that means the
// user typed that same char for this pos.
- if (firstCodePoint == baseLowerC || firstCodePoint == c) {
+ if (firstCodePoint == baseLowerC || firstCodePoint == codePoint) {
return EQUIVALENT_CHAR;
}
@@ -222,7 +222,8 @@ ProximityType ProximityInfoState::getMatchedProximityId(const int index, const i
int j = 1;
while (j < MAX_PROXIMITY_CHARS_SIZE
&& currentCodePoints[j] > ADDITIONAL_PROXIMITY_CHAR_DELIMITER_CODE) {
- const bool matched = (currentCodePoints[j] == baseLowerC || currentCodePoints[j] == c);
+ const bool matched = (currentCodePoints[j] == baseLowerC
+ || currentCodePoints[j] == codePoint);
if (matched) {
if (proximityIndex) {
*proximityIndex = j;
@@ -236,7 +237,8 @@ ProximityType ProximityInfoState::getMatchedProximityId(const int index, const i
++j;
while (j < MAX_PROXIMITY_CHARS_SIZE
&& currentCodePoints[j] > ADDITIONAL_PROXIMITY_CHAR_DELIMITER_CODE) {
- const bool matched = (currentCodePoints[j] == baseLowerC || currentCodePoints[j] == c);
+ const bool matched = (currentCodePoints[j] == baseLowerC
+ || currentCodePoints[j] == codePoint);
if (matched) {
if (proximityIndex) {
*proximityIndex = j;
@@ -250,6 +252,21 @@ ProximityType ProximityInfoState::getMatchedProximityId(const int index, const i
return UNRELATED_CHAR;
}
+ProximityType ProximityInfoState::getProximityTypeG(const int index, const int codePoint) const {
+ if (!isUsed()) {
+ return UNRELATED_NOR_SUBSTITUTION_CHAR;
+ }
+ const int lowerCodePoint = toLowerCase(codePoint);
+ const int baseLowerCodePoint = toBaseCodePoint(lowerCodePoint);
+ for (int i = 0; i < static_cast<int>(mSampledSearchKeyVectors[index].size()); ++i) {
+ if (mSampledSearchKeyVectors[index][i] == lowerCodePoint
+ || mSampledSearchKeyVectors[index][i] == baseLowerCodePoint) {
+ return EQUIVALENT_CHAR;
+ }
+ }
+ return UNRELATED_NOR_SUBSTITUTION_CHAR;
+}
+
bool ProximityInfoState::isKeyInSerchKeysAfterIndex(const int index, const int keyId) const {
ASSERT(keyId >= 0 && index >= 0 && index < mSampledInputSize);
return mSampledSearchKeySets[index].test(keyId);
@@ -297,6 +314,6 @@ float ProximityInfoState::getProbability(const int index, const int keyIndex) co
if (it != mCharProbabilities[index].end()) {
return it->second;
}
- return static_cast<float>(MAX_POINT_TO_KEY_LENGTH);
+ return static_cast<float>(MAX_VALUE_FOR_WEIGHTING);
}
} // namespace latinime
diff --git a/native/jni/src/proximity_info_state.h b/native/jni/src/proximity_info_state.h
index c4cbd582d..dd1eb764f 100644
--- a/native/jni/src/proximity_info_state.h
+++ b/native/jni/src/proximity_info_state.h
@@ -152,9 +152,11 @@ class ProximityInfoState {
float getPointToKeyLength(const int inputIndex, const int codePoint, const float scale) const;
float getPointToKeyLength_G(const int inputIndex, const int codePoint) const;
- ProximityType getMatchedProximityId(const int index, const int c,
+ ProximityType getProximityType(const int index, const int codePoint,
const bool checkProximityChars, int *proximityIndex = 0) const;
+ ProximityType getProximityTypeG(const int index, const int codePoint) const;
+
const std::vector<int> *getSearchKeyVector(const int index) const {
return &mSampledSearchKeyVectors[index];
}
diff --git a/native/jni/src/proximity_info_state_utils.cpp b/native/jni/src/proximity_info_state_utils.cpp
index 2bf327fcc..2ca39f9e3 100644
--- a/native/jni/src/proximity_info_state_utils.cpp
+++ b/native/jni/src/proximity_info_state_utils.cpp
@@ -645,7 +645,7 @@ namespace latinime {
return min((*SampledDistanceCache_G)[index] * scale, maxPointToKeyLength);
}
// If the char is not a key on the keyboard then return the max length.
- return static_cast<float>(MAX_POINT_TO_KEY_LENGTH);
+ return static_cast<float>(MAX_VALUE_FOR_WEIGHTING);
}
/* static */ float ProximityInfoStateUtils::getPointToKeyByIdLength(const float maxPointToKeyLength,
@@ -678,7 +678,7 @@ namespace latinime {
const float currentAngle = getPointAngle(sampledInputXs, sampledInputYs, i);
const float speedRate = (*sampledSpeedRates)[i];
- float nearestKeyDistance = static_cast<float>(MAX_POINT_TO_KEY_LENGTH);
+ float nearestKeyDistance = static_cast<float>(MAX_VALUE_FOR_WEIGHTING);
for (int j = 0; j < keyCount; ++j) {
if ((*SampledNearKeySets)[i].test(j)) {
const float distance = getPointToKeyByIdLength(
@@ -1016,7 +1016,7 @@ namespace latinime {
float sumLogProbability = 0.0f;
// TODO: Current implementation is greedy algorithm. DP would be efficient for many cases.
for (int i = 0; i < sampledInputSize && index < MAX_WORD_LENGTH - 1; ++i) {
- float minLogProbability = static_cast<float>(MAX_POINT_TO_KEY_LENGTH);
+ float minLogProbability = static_cast<float>(MAX_VALUE_FOR_WEIGHTING);
int character = NOT_AN_INDEX;
for (hash_map_compat<int, float>::const_iterator it = (*charProbabilities)[i].begin();
it != (*charProbabilities)[i].end(); ++it) {
diff --git a/native/jni/src/proximity_info_utils.h b/native/jni/src/proximity_info_utils.h
index 51cafba2c..71c97e325 100644
--- a/native/jni/src/proximity_info_utils.h
+++ b/native/jni/src/proximity_info_utils.h
@@ -226,7 +226,7 @@ class ProximityInfoUtils {
const int *const keyYCoordinates, const int *const keyWidths, const int *keyHeights,
const int keyId, const int x, const int y) {
// NOT_A_ID is -1, but return whenever < 0 just in case
- if (keyId < 0) return MAX_POINT_TO_KEY_LENGTH;
+ if (keyId < 0) return MAX_VALUE_FOR_WEIGHTING;
const int left = keyXCoordinates[keyId];
const int top = keyYCoordinates[keyId];
const int right = left + keyWidths[keyId];
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-31 16:36:47 +0000