nyancrimew
/
LatinIME
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Replace NOT_VALID_WORD with NOT_A_VALID_WORD_POS 

Bug: 6669677
Change-Id: If5fb58e8a98b04b767eaaf624cb6a40841b64f54
main
Keisuke Kuroynagi 2013-07-16 15:22:31 +09:00
parent 0d47f28448
commit 8a7129530b
9 changed files with 24 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
native/jni/src/defines.h
View File

@ -292,7 +292,7 @@ static inline void prof_out(void) {
// of the binary dictionary where a {key,value} string pair scheme is used. // of the binary dictionary where a {key,value} string pair scheme is used.
#define LARGEST_INT_DIGIT_COUNT 11 #define LARGEST_INT_DIGIT_COUNT 11
#define NOT_VALID_WORD (-99) #define NOT_A_VALID_WORD_POS (-99)
#define NOT_A_CODE_POINT (-1) #define NOT_A_CODE_POINT (-1)
#define NOT_A_DISTANCE (-1) #define NOT_A_DISTANCE (-1)
#define NOT_A_COORDINATE (-1) #define NOT_A_COORDINATE (-1)

6
native/jni/src/suggest/core/dicnode/dic_node.h
View File

@ -112,7 +112,7 @@ class DicNode {
mIsUsed = true; mIsUsed = true;
mIsCachedForNextSuggestion = false; mIsCachedForNextSuggestion = false;
mDicNodeProperties.init( mDicNodeProperties.init(
NOT_A_DICT_POS, rootGroupPos, NOT_A_DICT_POS /* attributesPos */, NOT_A_VALID_WORD_POS /* pos */, rootGroupPos, NOT_A_DICT_POS /* attributesPos */,
NOT_A_CODE_POINT /* nodeCodePoint */, NOT_A_PROBABILITY /* probability */, NOT_A_CODE_POINT /* nodeCodePoint */, NOT_A_PROBABILITY /* probability */,
false /* isTerminal */, true /* hasChildren */, false /* isTerminal */, true /* hasChildren */,
false /* isBlacklistedOrNotAWord */, 0 /* depth */, 0 /* terminalDepth */); false /* isBlacklistedOrNotAWord */, 0 /* depth */, 0 /* terminalDepth */);
@ -125,7 +125,7 @@ class DicNode {
mIsUsed = true; mIsUsed = true;
mIsCachedForNextSuggestion = dicNode->mIsCachedForNextSuggestion; mIsCachedForNextSuggestion = dicNode->mIsCachedForNextSuggestion;
mDicNodeProperties.init( mDicNodeProperties.init(
NOT_A_DICT_POS, rootGroupPos, NOT_A_DICT_POS /* attributesPos */, NOT_A_VALID_WORD_POS /* pos */, rootGroupPos, NOT_A_DICT_POS /* attributesPos */,
NOT_A_CODE_POINT /* nodeCodePoint */, NOT_A_PROBABILITY /* probability */, NOT_A_CODE_POINT /* nodeCodePoint */, NOT_A_PROBABILITY /* probability */,
false /* isTerminal */, true /* hasChildren */, false /* isTerminal */, true /* hasChildren */,
false /* isBlacklistedOrNotAWord */, 0 /* depth */, 0 /* terminalDepth */); false /* isBlacklistedOrNotAWord */, 0 /* depth */, 0 /* terminalDepth */);
@ -231,7 +231,7 @@ class DicNode {
} }
bool isFirstWord() const { bool isFirstWord() const {
return mDicNodeState.mDicNodeStatePrevWord.getPrevWordNodePos() == NOT_VALID_WORD; return mDicNodeState.mDicNodeStatePrevWord.getPrevWordNodePos() == NOT_A_VALID_WORD_POS;
} }
bool isCompletion(const int inputSize) const { bool isCompletion(const int inputSize) const {

4
native/jni/src/suggest/core/dicnode/dic_node_state_prevword.h
View File

@ -29,7 +29,7 @@ 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),
mPrevWordNodePos(0) { mPrevWordNodePos(NOT_A_VALID_WORD_POS) {
memset(mPrevWord, 0, sizeof(mPrevWord)); memset(mPrevWord, 0, sizeof(mPrevWord));
memset(mPrevSpacePositions, 0, sizeof(mPrevSpacePositions)); memset(mPrevSpacePositions, 0, sizeof(mPrevSpacePositions));
} }
@ -41,7 +41,7 @@ class DicNodeStatePrevWord {
mPrevWordCount = 0; mPrevWordCount = 0;
mPrevWordStart = 0; mPrevWordStart = 0;
mPrevWordProbability = -1; mPrevWordProbability = -1;
mPrevWordNodePos = NOT_VALID_WORD; mPrevWordNodePos = NOT_A_VALID_WORD_POS;
memset(mPrevSpacePositions, 0, sizeof(mPrevSpacePositions)); memset(mPrevSpacePositions, 0, sizeof(mPrevSpacePositions));
} }

5
native/jni/src/suggest/core/dicnode/dic_node_utils.cpp
View File

@ -179,8 +179,9 @@ namespace latinime {
const int unigramProbability = node->getProbability(); const int unigramProbability = node->getProbability();
const int wordPos = node->getPos(); const int wordPos = node->getPos();
const int prevWordPos = node->getPrevWordPos(); const int prevWordPos = node->getPrevWordPos();
if (NOT_VALID_WORD == wordPos || NOT_VALID_WORD == prevWordPos) { if (NOT_A_VALID_WORD_POS == wordPos || NOT_A_VALID_WORD_POS == prevWordPos) {
// Note: Normally wordPos comes from the dictionary and should never equal NOT_VALID_WORD. // Note: Normally wordPos comes from the dictionary and should never equal
// NOT_A_VALID_WORD_POS.
return ProbabilityUtils::backoff(unigramProbability); return ProbabilityUtils::backoff(unigramProbability);
} }
if (multiBigramMap) { if (multiBigramMap) {

4
native/jni/src/suggest/core/dictionary/bigram_dictionary.cpp
View File

@ -153,7 +153,7 @@ int BigramDictionary::getBigramListPositionForWord(const int *prevWord, const in
if (0 >= prevWordLength) return 0; if (0 >= prevWordLength) return 0;
int pos = mBinaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord( int pos = mBinaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord(
mBinaryDictionaryInfo, prevWord, prevWordLength, forceLowerCaseSearch); mBinaryDictionaryInfo, prevWord, prevWordLength, forceLowerCaseSearch);
if (NOT_VALID_WORD == pos) return 0; if (NOT_A_VALID_WORD_POS == pos) return 0;
return BinaryFormat::getBigramListPositionForWordPosition( return BinaryFormat::getBigramListPositionForWordPosition(
mBinaryDictionaryInfo->getDictRoot(), pos); mBinaryDictionaryInfo->getDictRoot(), pos);
} }
@ -181,7 +181,7 @@ bool BigramDictionary::isValidBigram(const int *word0, int length0, const int *w
if (0 == pos) return false; if (0 == pos) return false;
int nextWordPos = mBinaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord( int nextWordPos = mBinaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord(
mBinaryDictionaryInfo, word1, length1, false /* forceLowerCaseSearch */); mBinaryDictionaryInfo, word1, length1, false /* forceLowerCaseSearch */);
if (NOT_VALID_WORD == nextWordPos) return false; if (NOT_A_VALID_WORD_POS == nextWordPos) return false;
for (BinaryDictionaryBigramsIterator bigramsIt(mBinaryDictionaryInfo, pos); for (BinaryDictionaryBigramsIterator bigramsIt(mBinaryDictionaryInfo, pos);
bigramsIt.hasNext(); /* no-op */) { bigramsIt.hasNext(); /* no-op */) {

18
native/jni/src/suggest/core/dictionary/binary_format.h
View File

@ -255,7 +255,7 @@ inline bool BinaryFormat::hasChildrenInFlags(const uint8_t flags) {
} }
// This function gets the byte position of the last chargroup of the exact matching word in the // This function gets the byte position of the last chargroup of the exact matching word in the
// dictionary. If no match is found, it returns NOT_VALID_WORD. // dictionary. If no match is found, it returns NOT_A_VALID_WORD_POS.
AK_FORCE_INLINE int BinaryFormat::getTerminalPosition(const uint8_t *const root, AK_FORCE_INLINE int BinaryFormat::getTerminalPosition(const uint8_t *const root,
const int *const inWord, const int length, const bool forceLowerCaseSearch) { const int *const inWord, const int length, const bool forceLowerCaseSearch) {
int pos = 0; int pos = 0;
@ -264,22 +264,22 @@ AK_FORCE_INLINE int BinaryFormat::getTerminalPosition(const uint8_t *const root,
while (true) { while (true) {
// If we already traversed the tree further than the word is long, there means // If we already traversed the tree further than the word is long, there means
// there was no match (or we would have found it). // there was no match (or we would have found it).
if (wordPos >= length) return NOT_VALID_WORD; if (wordPos >= length) return NOT_A_VALID_WORD_POS;
int charGroupCount = BinaryFormat::getGroupCountAndForwardPointer(root, &pos); int charGroupCount = BinaryFormat::getGroupCountAndForwardPointer(root, &pos);
const int wChar = forceLowerCaseSearch const int wChar = forceLowerCaseSearch
? CharUtils::toLowerCase(inWord[wordPos]) : inWord[wordPos]; ? CharUtils::toLowerCase(inWord[wordPos]) : inWord[wordPos];
while (true) { while (true) {
// If there are no more character groups in this node, it means we could not // If there are no more character groups in this node, it means we could not
// find a matching character for this depth, therefore there is no match. // find a matching character for this depth, therefore there is no match.
if (0 >= charGroupCount) return NOT_VALID_WORD; if (0 >= charGroupCount) return NOT_A_VALID_WORD_POS;
const int charGroupPos = pos; const int charGroupPos = pos;
const uint8_t flags = BinaryFormat::getFlagsAndForwardPointer(root, &pos); const uint8_t flags = BinaryFormat::getFlagsAndForwardPointer(root, &pos);
int character = BinaryFormat::getCodePointAndForwardPointer(root, &pos); int character = BinaryFormat::getCodePointAndForwardPointer(root, &pos);
if (character == wChar) { if (character == wChar) {
// This is the correct node. Only one character group may start with the same // This is the correct node. Only one character group may start with the same
// char within a node, so either we found our match in this node, or there is // char within a node, so either we found our match in this node, or there is
// no match and we can return NOT_VALID_WORD. So we will check all the characters // no match and we can return NOT_A_VALID_WORD_POS. So we will check all the
// in this character group indeed does match. // characters in this character group indeed does match.
if (FLAG_HAS_MULTIPLE_CHARS & flags) { if (FLAG_HAS_MULTIPLE_CHARS & flags) {
character = BinaryFormat::getCodePointAndForwardPointer(root, &pos); character = BinaryFormat::getCodePointAndForwardPointer(root, &pos);
while (NOT_A_CODE_POINT != character) { while (NOT_A_CODE_POINT != character) {
@ -288,8 +288,8 @@ AK_FORCE_INLINE int BinaryFormat::getTerminalPosition(const uint8_t *const root,
// character that does not match, as explained above, it means the word is // character that does not match, as explained above, it means the word is
// not in the dictionary (by virtue of this chargroup being the only one to // not in the dictionary (by virtue of this chargroup being the only one to
// match the word on the first character, but not matching the whole word). // match the word on the first character, but not matching the whole word).
if (wordPos >= length) return NOT_VALID_WORD; if (wordPos >= length) return NOT_A_VALID_WORD_POS;
if (inWord[wordPos] != character) return NOT_VALID_WORD; if (inWord[wordPos] != character) return NOT_A_VALID_WORD_POS;
character = BinaryFormat::getCodePointAndForwardPointer(root, &pos); character = BinaryFormat::getCodePointAndForwardPointer(root, &pos);
} }
} }
@ -305,7 +305,7 @@ AK_FORCE_INLINE int BinaryFormat::getTerminalPosition(const uint8_t *const root,
pos = BinaryFormat::skipProbability(FLAG_IS_TERMINAL, pos); pos = BinaryFormat::skipProbability(FLAG_IS_TERMINAL, pos);
} }
if (FLAG_GROUP_ADDRESS_TYPE_NOADDRESS == (MASK_GROUP_ADDRESS_TYPE & flags)) { if (FLAG_GROUP_ADDRESS_TYPE_NOADDRESS == (MASK_GROUP_ADDRESS_TYPE & flags)) {
return NOT_VALID_WORD; return NOT_A_VALID_WORD_POS;
} }
// We have children and we are still shorter than the word we are searching for, so // We have children and we are still shorter than the word we are searching for, so
// we need to traverse children. Put the pointer on the children position, and // we need to traverse children. Put the pointer on the children position, and
@ -474,7 +474,7 @@ AK_FORCE_INLINE int BinaryFormat::getCodePointsAndProbabilityAndReturnCodePointC
AK_FORCE_INLINE int BinaryFormat::getBigramListPositionForWordPosition( AK_FORCE_INLINE int BinaryFormat::getBigramListPositionForWordPosition(
const uint8_t *const root, int position) { const uint8_t *const root, int position) {
if (NOT_VALID_WORD == position) return 0; if (NOT_A_VALID_WORD_POS == position) return 0;
const uint8_t flags = getFlagsAndForwardPointer(root, &position); const uint8_t flags = getFlagsAndForwardPointer(root, &position);
if (!(flags & FLAG_HAS_BIGRAMS)) return 0; if (!(flags & FLAG_HAS_BIGRAMS)) return 0;
if (flags & FLAG_HAS_MULTIPLE_CHARS) { if (flags & FLAG_HAS_MULTIPLE_CHARS) {

2
native/jni/src/suggest/core/dictionary/dictionary.cpp
View File

@ -89,7 +89,7 @@ int Dictionary::getProbability(const int *word, int length) const {
mBinaryDictionaryInfo.getStructurePolicy(); mBinaryDictionaryInfo.getStructurePolicy();
int pos = structurePolicy->getTerminalNodePositionOfWord(&mBinaryDictionaryInfo, word, length, int pos = structurePolicy->getTerminalNodePositionOfWord(&mBinaryDictionaryInfo, word, length,
false /* forceLowerCaseSearch */); false /* forceLowerCaseSearch */);
if (NOT_VALID_WORD == pos) { if (NOT_A_VALID_WORD_POS == pos) {
return NOT_A_PROBABILITY; return NOT_A_PROBABILITY;
} }
return structurePolicy->getUnigramProbability(&mBinaryDictionaryInfo, pos); return structurePolicy->getUnigramProbability(&mBinaryDictionaryInfo, pos);

4
native/jni/src/suggest/core/session/dic_traverse_session.cpp
View File

@ -32,13 +32,13 @@ void DicTraverseSession::init(const Dictionary *const dictionary, const int *pre
mMultiWordCostMultiplier = binaryDictionaryInfo->getHeader()->getMultiWordCostMultiplier(); mMultiWordCostMultiplier = binaryDictionaryInfo->getHeader()->getMultiWordCostMultiplier();
mSuggestOptions = suggestOptions; mSuggestOptions = suggestOptions;
if (!prevWord) { if (!prevWord) {
mPrevWordPos = NOT_VALID_WORD; mPrevWordPos = NOT_A_VALID_WORD_POS;
return; return;
} }
// TODO: merge following similar calls to getTerminalPosition into one case-insensitive call. // TODO: merge following similar calls to getTerminalPosition into one case-insensitive call.
mPrevWordPos = binaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord( mPrevWordPos = binaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord(
binaryDictionaryInfo, prevWord, prevWordLength, false /* forceLowerCaseSearch */); binaryDictionaryInfo, prevWord, prevWordLength, false /* forceLowerCaseSearch */);
if (mPrevWordPos == NOT_VALID_WORD) { if (mPrevWordPos == NOT_A_VALID_WORD_POS) {
// Check bigrams for lower-cased previous word if original was not found. Useful for // Check bigrams for lower-cased previous word if original was not found. Useful for
// auto-capitalized words like "The [current_word]". // auto-capitalized words like "The [current_word]".
mPrevWordPos = binaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord( mPrevWordPos = binaryDictionaryInfo->getStructurePolicy()->getTerminalNodePositionOfWord(

2
native/jni/src/suggest/core/session/dic_traverse_session.h
View File

@ -55,7 +55,7 @@ class DicTraverseSession {
} }
AK_FORCE_INLINE DicTraverseSession(JNIEnv *env, jstring localeStr) AK_FORCE_INLINE DicTraverseSession(JNIEnv *env, jstring localeStr)
: mPrevWordPos(NOT_VALID_WORD), mProximityInfo(0), : mPrevWordPos(NOT_A_VALID_WORD_POS), mProximityInfo(0),
mDictionary(0), mSuggestOptions(0), mDicNodesCache(), mMultiBigramMap(), mDictionary(0), mSuggestOptions(0), mDicNodesCache(), mMultiBigramMap(),
mInputSize(0), mPartiallyCommited(false), mMaxPointerCount(1), mInputSize(0), mPartiallyCommited(false), mMaxPointerCount(1),
mMultiWordCostMultiplier(1.0f) { mMultiWordCostMultiplier(1.0f) {