Merge "Improve DicNode handling."
This commit is contained in:
commit
bd44c9815d
8 changed files with 35 additions and 45 deletions
|
@ -97,7 +97,7 @@ class DicNode {
|
||||||
|
|
||||||
DicNode(const DicNode &dicNode);
|
DicNode(const DicNode &dicNode);
|
||||||
DicNode &operator=(const DicNode &dicNode);
|
DicNode &operator=(const DicNode &dicNode);
|
||||||
virtual ~DicNode() {}
|
~DicNode() {}
|
||||||
|
|
||||||
// Init for copy
|
// Init for copy
|
||||||
void initByCopy(const DicNode *const dicNode) {
|
void initByCopy(const DicNode *const dicNode) {
|
||||||
|
|
|
@ -33,7 +33,7 @@ class DicNodeProperties {
|
||||||
mIsTerminal(false), mHasChildrenPtNodes(false), mIsBlacklistedOrNotAWord(false),
|
mIsTerminal(false), mHasChildrenPtNodes(false), mIsBlacklistedOrNotAWord(false),
|
||||||
mDepth(0), mLeavingDepth(0) {}
|
mDepth(0), mLeavingDepth(0) {}
|
||||||
|
|
||||||
virtual ~DicNodeProperties() {}
|
~DicNodeProperties() {}
|
||||||
|
|
||||||
// Should be called only once per DicNode is initialized.
|
// Should be called only once per DicNode is initialized.
|
||||||
void init(const int pos, const int childrenPos, const int nodeCodePoint, const int probability,
|
void init(const int pos, const int childrenPos, const int nodeCodePoint, const int probability,
|
||||||
|
|
|
@ -37,7 +37,18 @@ class DicNodeState {
|
||||||
mDicNodeStateScoring() {
|
mDicNodeStateScoring() {
|
||||||
}
|
}
|
||||||
|
|
||||||
virtual ~DicNodeState() {}
|
~DicNodeState() {}
|
||||||
|
|
||||||
|
DicNodeState &operator=(const DicNodeState& src) {
|
||||||
|
init(&src);
|
||||||
|
return *this;
|
||||||
|
}
|
||||||
|
|
||||||
|
DicNodeState(const DicNodeState& src)
|
||||||
|
: mDicNodeStateInput(), mDicNodeStateOutput(), mDicNodeStatePrevWord(),
|
||||||
|
mDicNodeStateScoring() {
|
||||||
|
init(&src);
|
||||||
|
}
|
||||||
|
|
||||||
// Init with prevWordPos
|
// Init with prevWordPos
|
||||||
void init(const int prevWordPos) {
|
void init(const int prevWordPos) {
|
||||||
|
@ -62,11 +73,6 @@ class DicNodeState {
|
||||||
mDicNodeStateOutput.addMergedNodeCodePoints(
|
mDicNodeStateOutput.addMergedNodeCodePoints(
|
||||||
mergedNodeCodePointCount, mergedNodeCodePoints);
|
mergedNodeCodePointCount, mergedNodeCodePoints);
|
||||||
}
|
}
|
||||||
|
|
||||||
private:
|
|
||||||
// Caution!!!
|
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
|
||||||
// for this class
|
|
||||||
};
|
};
|
||||||
} // namespace latinime
|
} // namespace latinime
|
||||||
#endif // LATINIME_DIC_NODE_STATE_H
|
#endif // LATINIME_DIC_NODE_STATE_H
|
||||||
|
|
|
@ -25,7 +25,7 @@ namespace latinime {
|
||||||
class DicNodeStateInput {
|
class DicNodeStateInput {
|
||||||
public:
|
public:
|
||||||
DicNodeStateInput() {}
|
DicNodeStateInput() {}
|
||||||
virtual ~DicNodeStateInput() {}
|
~DicNodeStateInput() {}
|
||||||
|
|
||||||
// TODO: Merge into DicNodeStatePrevWord::truncate
|
// TODO: Merge into DicNodeStatePrevWord::truncate
|
||||||
void truncate(const int commitPoint) {
|
void truncate(const int commitPoint) {
|
||||||
|
@ -89,9 +89,8 @@ class DicNodeStateInput {
|
||||||
}
|
}
|
||||||
|
|
||||||
private:
|
private:
|
||||||
// Caution!!!
|
DISALLOW_COPY_AND_ASSIGN(DicNodeStateInput);
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
|
||||||
// for this class
|
|
||||||
int mInputIndex[MAX_POINTER_COUNT_G];
|
int mInputIndex[MAX_POINTER_COUNT_G];
|
||||||
int mPrevCodePoint[MAX_POINTER_COUNT_G];
|
int mPrevCodePoint[MAX_POINTER_COUNT_G];
|
||||||
float mTerminalDiffCost[MAX_POINTER_COUNT_G];
|
float mTerminalDiffCost[MAX_POINTER_COUNT_G];
|
||||||
|
|
|
@ -27,11 +27,9 @@ namespace latinime {
|
||||||
|
|
||||||
class DicNodeStateOutput {
|
class DicNodeStateOutput {
|
||||||
public:
|
public:
|
||||||
DicNodeStateOutput() : mOutputtedCodePointCount(0) {
|
DicNodeStateOutput() : mOutputtedCodePointCount(0) {}
|
||||||
init();
|
|
||||||
}
|
|
||||||
|
|
||||||
virtual ~DicNodeStateOutput() {}
|
~DicNodeStateOutput() {}
|
||||||
|
|
||||||
void init() {
|
void init() {
|
||||||
mOutputtedCodePointCount = 0;
|
mOutputtedCodePointCount = 0;
|
||||||
|
@ -72,9 +70,8 @@ class DicNodeStateOutput {
|
||||||
int mCodePointsBuf[MAX_WORD_LENGTH];
|
int mCodePointsBuf[MAX_WORD_LENGTH];
|
||||||
|
|
||||||
private:
|
private:
|
||||||
// Caution!!!
|
DISALLOW_COPY_AND_ASSIGN(DicNodeStateOutput);
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
|
||||||
// for this class
|
|
||||||
uint16_t mOutputtedCodePointCount;
|
uint16_t mOutputtedCodePointCount;
|
||||||
};
|
};
|
||||||
} // namespace latinime
|
} // namespace latinime
|
||||||
|
|
|
@ -31,20 +31,9 @@ class DicNodeStatePrevWord {
|
||||||
public:
|
public:
|
||||||
AK_FORCE_INLINE DicNodeStatePrevWord()
|
AK_FORCE_INLINE DicNodeStatePrevWord()
|
||||||
: mPrevWordCount(0), mPrevWordLength(0), mPrevWordStart(0), mPrevWordProbability(0),
|
: mPrevWordCount(0), mPrevWordLength(0), mPrevWordStart(0), mPrevWordProbability(0),
|
||||||
mPrevWordPtNodePos(NOT_A_DICT_POS), mSecondWordFirstInputIndex(NOT_AN_INDEX) {
|
mPrevWordPtNodePos(NOT_A_DICT_POS), mSecondWordFirstInputIndex(NOT_AN_INDEX) {}
|
||||||
memset(mPrevWord, 0, sizeof(mPrevWord));
|
|
||||||
}
|
|
||||||
|
|
||||||
virtual ~DicNodeStatePrevWord() {}
|
~DicNodeStatePrevWord() {}
|
||||||
|
|
||||||
void init() {
|
|
||||||
mPrevWordLength = 0;
|
|
||||||
mPrevWordCount = 0;
|
|
||||||
mPrevWordStart = 0;
|
|
||||||
mPrevWordProbability = -1;
|
|
||||||
mPrevWordPtNodePos = NOT_A_DICT_POS;
|
|
||||||
mSecondWordFirstInputIndex = NOT_AN_INDEX;
|
|
||||||
}
|
|
||||||
|
|
||||||
void init(const int prevWordNodePos) {
|
void init(const int prevWordNodePos) {
|
||||||
mPrevWordLength = 0;
|
mPrevWordLength = 0;
|
||||||
|
@ -53,6 +42,7 @@ class DicNodeStatePrevWord {
|
||||||
mPrevWordProbability = -1;
|
mPrevWordProbability = -1;
|
||||||
mPrevWordPtNodePos = prevWordNodePos;
|
mPrevWordPtNodePos = prevWordNodePos;
|
||||||
mSecondWordFirstInputIndex = NOT_AN_INDEX;
|
mSecondWordFirstInputIndex = NOT_AN_INDEX;
|
||||||
|
mPrevWord[0] = 0;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Init by copy
|
// Init by copy
|
||||||
|
@ -141,9 +131,8 @@ class DicNodeStatePrevWord {
|
||||||
int mPrevWord[MAX_WORD_LENGTH];
|
int mPrevWord[MAX_WORD_LENGTH];
|
||||||
|
|
||||||
private:
|
private:
|
||||||
// Caution!!!
|
DISALLOW_COPY_AND_ASSIGN(DicNodeStatePrevWord);
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
|
||||||
// for this class
|
|
||||||
int16_t mPrevWordCount;
|
int16_t mPrevWordCount;
|
||||||
int16_t mPrevWordLength;
|
int16_t mPrevWordLength;
|
||||||
int16_t mPrevWordStart;
|
int16_t mPrevWordStart;
|
||||||
|
|
|
@ -37,7 +37,7 @@ class DicNodeStateScoring {
|
||||||
mNormalizedCompoundDistanceAfterFirstWord(MAX_VALUE_FOR_WEIGHTING) {
|
mNormalizedCompoundDistanceAfterFirstWord(MAX_VALUE_FOR_WEIGHTING) {
|
||||||
}
|
}
|
||||||
|
|
||||||
virtual ~DicNodeStateScoring() {}
|
~DicNodeStateScoring() {}
|
||||||
|
|
||||||
void init() {
|
void init() {
|
||||||
mEditCorrectionCount = 0;
|
mEditCorrectionCount = 0;
|
||||||
|
@ -175,9 +175,8 @@ class DicNodeStateScoring {
|
||||||
}
|
}
|
||||||
|
|
||||||
private:
|
private:
|
||||||
// Caution!!!
|
DISALLOW_COPY_AND_ASSIGN(DicNodeStateScoring);
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
|
||||||
// for this class
|
|
||||||
DoubleLetterLevel mDoubleLetterLevel;
|
DoubleLetterLevel mDoubleLetterLevel;
|
||||||
DigraphUtils::DigraphCodePointIndex mDigraphIndex;
|
DigraphUtils::DigraphCodePointIndex mDigraphIndex;
|
||||||
|
|
||||||
|
|
|
@ -248,17 +248,16 @@ void Suggest::processTerminalDicNode(
|
||||||
if (dicNode->shouldBeFilteredBySafetyNetForBigram()) {
|
if (dicNode->shouldBeFilteredBySafetyNetForBigram()) {
|
||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
if (!dicNode->hasMatchedOrProximityCodePoints()) {
|
||||||
|
return;
|
||||||
|
}
|
||||||
// Create a non-cached node here.
|
// Create a non-cached node here.
|
||||||
DicNode terminalDicNode;
|
DicNode terminalDicNode(*dicNode);
|
||||||
DicNodeUtils::initByCopy(dicNode, &terminalDicNode);
|
|
||||||
if (TRAVERSAL->needsToTraverseAllUserInput()
|
if (TRAVERSAL->needsToTraverseAllUserInput()
|
||||||
&& dicNode->getInputIndex(0) < traverseSession->getInputSize()) {
|
&& dicNode->getInputIndex(0) < traverseSession->getInputSize()) {
|
||||||
Weighting::addCostAndForwardInputIndex(WEIGHTING, CT_TERMINAL_INSERTION, traverseSession, 0,
|
Weighting::addCostAndForwardInputIndex(WEIGHTING, CT_TERMINAL_INSERTION, traverseSession, 0,
|
||||||
&terminalDicNode, traverseSession->getMultiBigramMap());
|
&terminalDicNode, traverseSession->getMultiBigramMap());
|
||||||
}
|
}
|
||||||
if (!dicNode->hasMatchedOrProximityCodePoints()) {
|
|
||||||
return;
|
|
||||||
}
|
|
||||||
Weighting::addCostAndForwardInputIndex(WEIGHTING, CT_TERMINAL, traverseSession, 0,
|
Weighting::addCostAndForwardInputIndex(WEIGHTING, CT_TERMINAL, traverseSession, 0,
|
||||||
&terminalDicNode, traverseSession->getMultiBigramMap());
|
&terminalDicNode, traverseSession->getMultiBigramMap());
|
||||||
traverseSession->getDicTraverseCache()->copyPushTerminal(&terminalDicNode);
|
traverseSession->getDicTraverseCache()->copyPushTerminal(&terminalDicNode);
|
||||||
|
@ -375,6 +374,7 @@ void Suggest::processDicNodeAsTransposition(DicTraverseSession *traverseSession,
|
||||||
DicNode *dicNode) const {
|
DicNode *dicNode) const {
|
||||||
const int16_t pointIndex = dicNode->getInputIndex(0);
|
const int16_t pointIndex = dicNode->getInputIndex(0);
|
||||||
DicNodeVector childDicNodes1;
|
DicNodeVector childDicNodes1;
|
||||||
|
DicNodeVector childDicNodes2;
|
||||||
DicNodeUtils::getAllChildDicNodes(dicNode, traverseSession->getDictionaryStructurePolicy(),
|
DicNodeUtils::getAllChildDicNodes(dicNode, traverseSession->getDictionaryStructurePolicy(),
|
||||||
&childDicNodes1);
|
&childDicNodes1);
|
||||||
const int childSize1 = childDicNodes1.getSizeAndLock();
|
const int childSize1 = childDicNodes1.getSizeAndLock();
|
||||||
|
@ -386,7 +386,7 @@ void Suggest::processDicNodeAsTransposition(DicTraverseSession *traverseSession,
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
if (childDicNodes1[i]->hasChildren()) {
|
if (childDicNodes1[i]->hasChildren()) {
|
||||||
DicNodeVector childDicNodes2;
|
childDicNodes2.clear();
|
||||||
DicNodeUtils::getAllChildDicNodes(childDicNodes1[i],
|
DicNodeUtils::getAllChildDicNodes(childDicNodes1[i],
|
||||||
traverseSession->getDictionaryStructurePolicy(), &childDicNodes2);
|
traverseSession->getDictionaryStructurePolicy(), &childDicNodes2);
|
||||||
const int childSize2 = childDicNodes2.getSizeAndLock();
|
const int childSize2 = childDicNodes2.getSizeAndLock();
|
||||||
|
|
Loading…
Reference in a new issue