diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
index 6a6a0a4ee282e5e1cf865edcbfc2eabd520cb701..ab9edb1107539c31ecc84b224b02e14a3c69acf5 100644
--- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java
+++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java
@@ -61,6 +61,8 @@ public class BinaryDictionary extends Dictionary {
public static final Flag FLAG_REQUIRES_GERMAN_UMLAUT_PROCESSING =
new Flag(R.bool.config_require_umlaut_processing, 0x1);
+ public static final Flag FLAG_USE_FULL_EDIT_DISTANCE = new Flag(0x2);
+
// Can create a new flag from extravalue :
// public static final Flag FLAG_MYFLAG =
// new Flag("my_flag", 0x02);
diff --git a/java/src/com/android/inputmethod/latin/Flag.java b/java/src/com/android/inputmethod/latin/Flag.java
index 3cb8f7e177eee83a61a813fd3090d60f3bd551e5..4ba6c80f59a1b2de9fabd4b078f028fd89838932 100644
--- a/java/src/com/android/inputmethod/latin/Flag.java
+++ b/java/src/com/android/inputmethod/latin/Flag.java
@@ -25,8 +25,9 @@ public class Flag {
public final int mMask;
public final int mSource;
- static private final int SOURCE_CONFIG = 1;
- static private final int SOURCE_EXTRAVALUE = 2;
+ private static final int SOURCE_CONFIG = 1;
+ private static final int SOURCE_EXTRAVALUE = 2;
+ private static final int SOURCE_PARAM = 3;
public Flag(int resourceId, int mask) {
mName = null;
@@ -42,6 +43,13 @@ public class Flag {
mMask = mask;
}
+ public Flag(int mask) {
+ mName = null;
+ mResource = 0;
+ mSource = SOURCE_PARAM;
+ mMask = mask;
+ }
+
// If context/switcher are null, set all related flags in flagArray to on.
public static int initFlags(Flag[] flagArray, Context context, SubtypeSwitcher switcher) {
int flags = 0;
@@ -57,6 +65,9 @@ public class Flag {
switcher.currentSubtypeContainsExtraValueKey(entry.mName))
flags |= entry.mMask;
break;
+ case Flag.SOURCE_PARAM:
+ flags |= entry.mMask;
+ break;
}
}
return flags;
diff --git a/native/src/correction.cpp b/native/src/correction.cpp
index d5bfed017f34a5a3ff49d887dd00d44720f531aa..9a7e5f35da5b4ddeb740e8ade5e862c269220f09 100644
--- a/native/src/correction.cpp
+++ b/native/src/correction.cpp
@@ -62,7 +62,8 @@ void Correction::initCorrectionState(
}
void Correction::setCorrectionParams(const int skipPos, const int excessivePos,
- const int transposedPos, const int spaceProximityPos, const int missingSpacePos) {
+ const int transposedPos, const int spaceProximityPos, const int missingSpacePos,
+ const bool useFullEditDistance) {
// TODO: remove
mTransposedPos = transposedPos;
mExcessivePos = excessivePos;
@@ -74,6 +75,7 @@ void Correction::setCorrectionParams(const int skipPos, const int excessivePos,
mSpaceProximityPos = spaceProximityPos;
mMissingSpacePos = missingSpacePos;
+ mUseFullEditDistance = useFullEditDistance;
}
void Correction::checkState() {
@@ -439,7 +441,7 @@ inline static void multiplyIntCapped(const int multiplier, int *base) {
}
inline static int powerIntCapped(const int base, const int n) {
- if (n == 0) return 1;
+ if (n <= 0) return 1;
if (base == 2) {
return n < 31 ? 1 << n : S_INT_MAX;
} else {
@@ -529,6 +531,8 @@ int Correction::RankingAlgorithm::calculateFinalFreq(const int inputIndex, const
const int excessiveCount = correction->mExcessiveCount + correction->mTransposedCount % 2;
const int proximityMatchedCount = correction->mProximityCount;
const bool lastCharExceeded = correction->mLastCharExceeded;
+ const bool useFullEditDistance = correction->mUseFullEditDistance;
+ const int outputLength = outputIndex + 1;
if (skippedCount >= inputLength || inputLength == 0) {
return -1;
}
@@ -682,6 +686,12 @@ int Correction::RankingAlgorithm::calculateFinalFreq(const int inputIndex, const
multiplyIntCapped(fullWordMultiplier, &finalFreq);
}
+ if (useFullEditDistance && outputLength > inputLength + 1) {
+ const int diff = outputLength - inputLength - 1;
+ const int divider = diff < 31 ? 1 << diff : S_INT_MAX;
+ finalFreq = divider > finalFreq ? 1 : finalFreq / divider;
+ }
+
if (DEBUG_DICT_FULL) {
LOGI("calc: %d, %d", outputIndex, sameLength);
}
diff --git a/native/src/correction.h b/native/src/correction.h
index 41130ad77ae98a684e4ad075fa5e6fc1f3ff12ff..ddb98b9d8464ca30f9d4028aef6fbf9d97c630e6 100644
--- a/native/src/correction.h
+++ b/native/src/correction.h
@@ -44,7 +44,7 @@ public:
// TODO: remove
void setCorrectionParams(const int skipPos, const int excessivePos, const int transposedPos,
- const int spaceProximityPos, const int missingSpacePos);
+ const int spaceProximityPos, const int missingSpacePos, const bool useFullEditDistance);
void checkState();
bool initProcessState(const int index);
@@ -111,6 +111,7 @@ private:
const int FULL_WORD_MULTIPLIER;
const ProximityInfo *mProximityInfo;
+ bool mUseFullEditDistance;
int mMaxEditDistance;
int mMaxDepth;
int mInputLength;
diff --git a/native/src/unigram_dictionary.cpp b/native/src/unigram_dictionary.cpp
index 517dc843e73462a071747e715d82e96decb79953..1b798a8f17d106aa10ff889ac3c4c617c177fc9b 100644
--- a/native/src/unigram_dictionary.cpp
+++ b/native/src/unigram_dictionary.cpp
@@ -132,7 +132,8 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
memcpy(codesDest, codesSrc, remainingBytes);
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
- (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS + codesRemain, outWords, frequencies);
+ (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS + codesRemain, outWords, frequencies,
+ flags);
}
int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
@@ -146,7 +147,7 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, const int *x
codesSize, flags, codes, codesSize, 0, codesBuffer, outWords, frequencies);
} else { // Normal processing
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize,
- outWords, frequencies);
+ outWords, frequencies, flags);
}
PROF_START(20);
@@ -175,7 +176,7 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, const int *x
void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
const int *xcoordinates, const int *ycoordinates, const int *codes, const int codesSize,
- unsigned short *outWords, int *frequencies) {
+ unsigned short *outWords, int *frequencies, const int flags) {
PROF_OPEN;
PROF_START(0);
@@ -187,9 +188,10 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
mCorrection->initCorrection(mProximityInfo, mInputLength, maxDepth);
PROF_END(0);
+ const bool useFullEditDistance = USE_FULL_EDIT_DISTANCE & flags;
// TODO: remove
PROF_START(1);
- getSuggestionCandidates();
+ getSuggestionCandidates(useFullEditDistance);
PROF_END(1);
PROF_START(2);
@@ -212,7 +214,7 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
if (DEBUG_DICT) {
LOGI("--- Suggest missing space characters %d", i);
}
- getMissingSpaceWords(mInputLength, i, mCorrection);
+ getMissingSpaceWords(mInputLength, i, mCorrection, useFullEditDistance);
}
}
PROF_END(5);
@@ -231,7 +233,7 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
i, x, y, proximityInfo->hasSpaceProximity(x, y));
}
if (proximityInfo->hasSpaceProximity(x, y)) {
- getMistypedSpaceWords(mInputLength, i, mCorrection);
+ getMistypedSpaceWords(mInputLength, i, mCorrection, useFullEditDistance);
}
}
}
@@ -315,10 +317,10 @@ bool UnigramDictionary::addWord(unsigned short *word, int length, int frequency)
static const char QUOTE = '\'';
static const char SPACE = ' ';
-void UnigramDictionary::getSuggestionCandidates() {
+void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance) {
// TODO: Remove setCorrectionParams
mCorrection->setCorrectionParams(0, 0, 0,
- -1 /* spaceProximityPos */, -1 /* missingSpacePos */);
+ -1 /* spaceProximityPos */, -1 /* missingSpacePos */, useFullEditDistance);
int rootPosition = ROOT_POS;
// Get the number of children of root, then increment the position
int childCount = Dictionary::getCount(DICT_ROOT, &rootPosition);
@@ -349,16 +351,20 @@ void UnigramDictionary::getSuggestionCandidates() {
}
void UnigramDictionary::getMissingSpaceWords(
- const int inputLength, const int missingSpacePos, Correction *correction) {
+ const int inputLength, const int missingSpacePos, Correction *correction,
+ const bool useFullEditDistance) {
correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
- -1 /* transposedPos */, -1 /* spaceProximityPos */, missingSpacePos);
+ -1 /* transposedPos */, -1 /* spaceProximityPos */, missingSpacePos,
+ useFullEditDistance);
getSplitTwoWordsSuggestion(inputLength, correction);
}
void UnigramDictionary::getMistypedSpaceWords(
- const int inputLength, const int spaceProximityPos, Correction *correction) {
+ const int inputLength, const int spaceProximityPos, Correction *correction,
+ const bool useFullEditDistance) {
correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
- -1 /* transposedPos */, spaceProximityPos, -1 /* missingSpacePos */);
+ -1 /* transposedPos */, spaceProximityPos, -1 /* missingSpacePos */,
+ useFullEditDistance);
getSplitTwoWordsSuggestion(inputLength, correction);
}
diff --git a/native/src/unigram_dictionary.h b/native/src/unigram_dictionary.h
index 65746db8d3bba980a049a9c9429c79e2327e8011..ef9709a89a579edfa05a1565d9564de661a230c7 100644
--- a/native/src/unigram_dictionary.h
+++ b/native/src/unigram_dictionary.h
@@ -78,7 +78,7 @@ private:
void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize,
- unsigned short *outWords, int *frequencies);
+ unsigned short *outWords, int *frequencies, const int flags);
bool isDigraph(const int* codes, const int i, const int codesSize) const;
void getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
@@ -87,13 +87,13 @@ private:
void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize,
unsigned short *outWords, int *frequencies);
- void getSuggestionCandidates();
+ void getSuggestionCandidates(const bool useFullEditDistance);
bool addWord(unsigned short *word, int length, int frequency);
void getSplitTwoWordsSuggestion(const int inputLength, Correction *correction);
- void getMissingSpaceWords(
- const int inputLength, const int missingSpacePos, Correction *correction);
- void getMistypedSpaceWords(
- const int inputLength, const int spaceProximityPos, Correction *correction);
+ void getMissingSpaceWords(const int inputLength, const int missingSpacePos,
+ Correction *correction, const bool useFullEditDistance);
+ void getMistypedSpaceWords(const int inputLength, const int spaceProximityPos,
+ Correction *correction, const bool useFullEditDistance);
void onTerminal(const int freq, Correction *correction);
bool needsToSkipCurrentNode(const unsigned short c,
const int inputIndex, const int skipPos, const int depth);
@@ -122,7 +122,8 @@ private:
// or something very bad (like, the apocalypse) will happen.
// Please update both at the same time.
enum {
- REQUIRES_GERMAN_UMLAUT_PROCESSING = 0x1
+ REQUIRES_GERMAN_UMLAUT_PROCESSING = 0x1,
+ USE_FULL_EDIT_DISTANCE = 0x2
};
static const struct digraph_t { int first; int second; } GERMAN_UMLAUT_DIGRAPHS[];