aboutsummaryrefslogtreecommitdiffstats
path: root/native/src/unigram_dictionary.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'native/src/unigram_dictionary.cpp')
-rw-r--r--native/src/unigram_dictionary.cpp300
1 files changed, 144 insertions, 156 deletions
diff --git a/native/src/unigram_dictionary.cpp b/native/src/unigram_dictionary.cpp
index 8eb5a9700..e95e03ce5 100644
--- a/native/src/unigram_dictionary.cpp
+++ b/native/src/unigram_dictionary.cpp
@@ -25,6 +25,7 @@
#include "unigram_dictionary.h"
#include "binary_format.h"
+#include "terminal_attributes.h"
namespace latinime {
@@ -48,19 +49,23 @@ UnigramDictionary::UnigramDictionary(const uint8_t* const streamStart, int typed
if (DEBUG_DICT) {
LOGI("UnigramDictionary - constructor");
}
- mCorrection = new Correction(typedLetterMultiplier, fullWordMultiplier);
}
UnigramDictionary::~UnigramDictionary() {
- delete mCorrection;
}
-static inline unsigned int getCodesBufferSize(const int* codes, const int codesSize,
+static inline unsigned int getCodesBufferSize(const int *codes, const int codesSize,
const int MAX_PROXIMITY_CHARS) {
return sizeof(*codes) * MAX_PROXIMITY_CHARS * codesSize;
}
-bool UnigramDictionary::isDigraph(const int* codes, const int i, const int codesSize) const {
+// TODO: This needs to take an const unsigned short* and not tinker with its contents
+static inline void addWord(
+ unsigned short *word, int length, int frequency, WordsPriorityQueue *queue) {
+ queue->push(frequency, word, length);
+}
+
+bool UnigramDictionary::isDigraph(const int *codes, const int i, const int codesSize) const {
// There can't be a digraph if we don't have at least 2 characters to examine
if (i + 2 > codesSize) return false;
@@ -86,9 +91,10 @@ bool UnigramDictionary::isDigraph(const int* codes, const int i, const int codes
// codesSrc is the current point in the user-input, original, content-unmodified buffer.
// codesRemain is the remaining size in codesSrc.
void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
- const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
- const int codesBufferSize, const int flags, const int* codesSrc, const int codesRemain,
- const int currentDepth, int* codesDest, unsigned short* outWords, int* frequencies) {
+ const int *xcoordinates, const int *ycoordinates, const int *codesBuffer,
+ const int codesBufferSize, const int flags, const int *codesSrc,
+ const int codesRemain, const int currentDepth, int *codesDest, Correction *correction,
+ WordsPriorityQueuePool *queuePool) {
if (currentDepth < MAX_UMLAUT_SEARCH_DEPTH) {
for (int i = 0; i < codesRemain; ++i) {
@@ -105,8 +111,8 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, codesBufferSize, flags,
codesSrc + (i + 1) * MAX_PROXIMITY_CHARS, codesRemain - i - 1,
- currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS, outWords,
- frequencies);
+ currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS, correction,
+ queuePool);
// Copy the second char of the digraph in place, then continue processing on
// the remaining part of the word.
@@ -114,9 +120,9 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
memcpy(codesDest + i * MAX_PROXIMITY_CHARS, codesSrc + i * MAX_PROXIMITY_CHARS,
BYTES_IN_ONE_CHAR);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
- codesBuffer, codesBufferSize, flags, codesSrc + i * MAX_PROXIMITY_CHARS,
- codesRemain - i, currentDepth + 1, codesDest + i * MAX_PROXIMITY_CHARS,
- outWords, frequencies);
+ codesBuffer, codesBufferSize, flags,
+ codesSrc + i * MAX_PROXIMITY_CHARS, codesRemain - i, currentDepth + 1,
+ codesDest + i * MAX_PROXIMITY_CHARS, correction, queuePool);
return;
}
}
@@ -132,40 +138,39 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
memcpy(codesDest, codesSrc, remainingBytes);
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
- (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS + codesRemain, outWords, frequencies,
- flags);
+ (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS + codesRemain, flags, correction,
+ queuePool);
}
-int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
+int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
+ WordsPriorityQueuePool *queuePool, Correction *correction, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize, const int flags,
unsigned short *outWords, int *frequencies) {
+ Correction* masterCorrection = correction;
if (REQUIRES_GERMAN_UMLAUT_PROCESSING & flags)
{ // Incrementally tune the word and try all possibilities
int codesBuffer[getCodesBufferSize(codes, codesSize, MAX_PROXIMITY_CHARS)];
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
- codesSize, flags, codes, codesSize, 0, codesBuffer, outWords, frequencies);
+ codesSize, flags, codes, codesSize, 0, codesBuffer, masterCorrection, queuePool);
} else { // Normal processing
- getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize,
- outWords, frequencies, flags);
+ getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize, flags,
+ masterCorrection, queuePool);
}
PROF_START(20);
- // Get the word count
- int suggestedWordsCount = 0;
- while (suggestedWordsCount < MAX_WORDS && mFrequencies[suggestedWordsCount] > 0) {
- suggestedWordsCount++;
- }
+ const int suggestedWordsCount =
+ queuePool->getMasterQueue()->outputSuggestions(frequencies, outWords);
if (DEBUG_DICT) {
LOGI("Returning %d words", suggestedWordsCount);
/// Print the returned words
for (int j = 0; j < suggestedWordsCount; ++j) {
#ifdef FLAG_DBG
- short unsigned int* w = mOutputChars + j * MAX_WORD_LENGTH;
+ short unsigned int* w = outWords + j * MAX_WORD_LENGTH;
char s[MAX_WORD_LENGTH];
for (int i = 0; i <= MAX_WORD_LENGTH; i++) s[i] = w[i];
- LOGI("%s %i", s, mFrequencies[j]);
+ LOGI("%s %i", s, frequencies[j]);
#endif
}
}
@@ -175,23 +180,19 @@ 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, const int flags) {
+ const int *xcoordinates, const int *ycoordinates, const int *codes,
+ const int inputLength, const int flags, Correction *correction,
+ WordsPriorityQueuePool *queuePool) {
PROF_OPEN;
PROF_START(0);
- initSuggestions(
- proximityInfo, xcoordinates, ycoordinates, codes, codesSize, outWords, frequencies);
- if (DEBUG_DICT) assert(codesSize == mInputLength);
-
- const int maxDepth = min(mInputLength * MAX_DEPTH_MULTIPLIER, MAX_WORD_LENGTH);
- mCorrection->initCorrection(mProximityInfo, mInputLength, maxDepth);
+ // Note: This line is intentionally left blank
PROF_END(0);
- const bool useFullEditDistance = USE_FULL_EDIT_DISTANCE & flags;
- // TODO: remove
PROF_START(1);
- getSuggestionCandidates(useFullEditDistance);
+ const bool useFullEditDistance = USE_FULL_EDIT_DISTANCE & flags;
+ getOneWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, useFullEditDistance,
+ inputLength, correction, queuePool);
PROF_END(1);
PROF_START(2);
@@ -209,12 +210,13 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
PROF_START(5);
// Suggestions with missing space
if (SUGGEST_WORDS_WITH_MISSING_SPACE_CHARACTER
- && mInputLength >= MIN_USER_TYPED_LENGTH_FOR_MISSING_SPACE_SUGGESTION) {
- for (int i = 1; i < codesSize; ++i) {
+ && inputLength >= MIN_USER_TYPED_LENGTH_FOR_MISSING_SPACE_SUGGESTION) {
+ for (int i = 1; i < inputLength; ++i) {
if (DEBUG_DICT) {
LOGI("--- Suggest missing space characters %d", i);
}
- getMissingSpaceWords(mInputLength, i, mCorrection, useFullEditDistance);
+ getMissingSpaceWords(proximityInfo, xcoordinates, ycoordinates, codes,
+ useFullEditDistance, inputLength, i, correction, queuePool);
}
}
PROF_END(5);
@@ -222,7 +224,7 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
PROF_START(6);
if (SUGGEST_WORDS_WITH_SPACE_PROXIMITY && proximityInfo) {
// The first and last "mistyped spaces" are taken care of by excessive character handling
- for (int i = 1; i < codesSize - 1; ++i) {
+ for (int i = 1; i < inputLength - 1; ++i) {
if (DEBUG_DICT) {
LOGI("--- Suggest words with proximity space %d", i);
}
@@ -233,7 +235,8 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
i, x, y, proximityInfo->hasSpaceProximity(x, y));
}
if (proximityInfo->hasSpaceProximity(x, y)) {
- getMistypedSpaceWords(mInputLength, i, mCorrection, useFullEditDistance);
+ getMistypedSpaceWords(proximityInfo, xcoordinates, ycoordinates, codes,
+ useFullEditDistance, inputLength, i, correction, queuePool);
}
}
}
@@ -241,153 +244,118 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
}
void UnigramDictionary::initSuggestions(ProximityInfo *proximityInfo, const int *xCoordinates,
- const int *yCoordinates, const int *codes, const int codesSize,
- unsigned short *outWords, int *frequencies) {
+ const int *yCoordinates, const int *codes, const int inputLength,
+ WordsPriorityQueue *queue, Correction *correction) {
if (DEBUG_DICT) {
LOGI("initSuggest");
}
- mFrequencies = frequencies;
- mOutputChars = outWords;
- mInputLength = codesSize;
- proximityInfo->setInputParams(codes, codesSize, xCoordinates, yCoordinates);
- mProximityInfo = proximityInfo;
-}
-
-static inline void registerNextLetter(unsigned short c, int *nextLetters, int nextLettersSize) {
- if (c < nextLettersSize) {
- nextLetters[c]++;
- }
-}
-
-// TODO: We need to optimize addWord by using STL or something
-// TODO: This needs to take an const unsigned short* and not tinker with its contents
-bool UnigramDictionary::addWord(unsigned short *word, int length, int frequency) {
- word[length] = 0;
- if (DEBUG_DICT && DEBUG_SHOW_FOUND_WORD) {
-#ifdef FLAG_DBG
- char s[length + 1];
- for (int i = 0; i <= length; i++) s[i] = word[i];
- LOGI("Found word = %s, freq = %d", s, frequency);
-#endif
- }
- if (length > MAX_WORD_LENGTH) {
- if (DEBUG_DICT) {
- LOGI("Exceeded max word length.");
- }
- return false;
- }
-
- // Find the right insertion point
- int insertAt = 0;
- while (insertAt < MAX_WORDS) {
- // TODO: How should we sort words with the same frequency?
- if (frequency > mFrequencies[insertAt]) {
- break;
- }
- insertAt++;
- }
- if (insertAt < MAX_WORDS) {
- if (DEBUG_DICT) {
-#ifdef FLAG_DBG
- char s[length + 1];
- for (int i = 0; i <= length; i++) s[i] = word[i];
- LOGI("Added word = %s, freq = %d, %d", s, frequency, S_INT_MAX);
-#endif
- }
- memmove((char*) mFrequencies + (insertAt + 1) * sizeof(mFrequencies[0]),
- (char*) mFrequencies + insertAt * sizeof(mFrequencies[0]),
- (MAX_WORDS - insertAt - 1) * sizeof(mFrequencies[0]));
- mFrequencies[insertAt] = frequency;
- memmove((char*) mOutputChars + (insertAt + 1) * MAX_WORD_LENGTH * sizeof(short),
- (char*) mOutputChars + insertAt * MAX_WORD_LENGTH * sizeof(short),
- (MAX_WORDS - insertAt - 1) * sizeof(short) * MAX_WORD_LENGTH);
- unsigned short *dest = mOutputChars + insertAt * MAX_WORD_LENGTH;
- while (length--) {
- *dest++ = *word++;
- }
- *dest = 0; // NULL terminate
- if (DEBUG_DICT) {
- LOGI("Added word at %d", insertAt);
- }
- return true;
+ proximityInfo->setInputParams(codes, inputLength, xCoordinates, yCoordinates);
+ if (queue) {
+ queue->clear();
}
- return false;
+ const int maxDepth = min(inputLength * MAX_DEPTH_MULTIPLIER, MAX_WORD_LENGTH);
+ correction->initCorrection(proximityInfo, inputLength, maxDepth);
}
static const char QUOTE = '\'';
static const char SPACE = ' ';
-void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance) {
+void UnigramDictionary::getOneWordSuggestions(ProximityInfo *proximityInfo,
+ const int *xcoordinates, const int *ycoordinates, const int *codes,
+ const bool useFullEditDistance, const int inputLength, Correction *correction,
+ WordsPriorityQueuePool *queuePool) {
+ WordsPriorityQueue *masterQueue = queuePool->getMasterQueue();
+ initSuggestions(
+ proximityInfo, xcoordinates, ycoordinates, codes, inputLength, masterQueue, correction);
+ getSuggestionCandidates(useFullEditDistance, inputLength, correction, masterQueue,
+ true /* doAutoCompletion */, DEFAULT_MAX_ERRORS);
+}
+
+void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance,
+ const int inputLength, Correction *correction, WordsPriorityQueue *queue,
+ const bool doAutoCompletion, const int maxErrors) {
// TODO: Remove setCorrectionParams
- mCorrection->setCorrectionParams(0, 0, 0,
- -1 /* spaceProximityPos */, -1 /* missingSpacePos */, useFullEditDistance);
+ correction->setCorrectionParams(0, 0, 0,
+ -1 /* spaceProximityPos */, -1 /* missingSpacePos */, useFullEditDistance,
+ doAutoCompletion, maxErrors);
int rootPosition = ROOT_POS;
// Get the number of children of root, then increment the position
int childCount = Dictionary::getCount(DICT_ROOT, &rootPosition);
int outputIndex = 0;
- mCorrection->initCorrectionState(rootPosition, childCount, (mInputLength <= 0));
+ correction->initCorrectionState(rootPosition, childCount, (inputLength <= 0));
// Depth first search
while (outputIndex >= 0) {
- if (mCorrection->initProcessState(outputIndex)) {
- int siblingPos = mCorrection->getTreeSiblingPos(outputIndex);
+ if (correction->initProcessState(outputIndex)) {
+ int siblingPos = correction->getTreeSiblingPos(outputIndex);
int firstChildPos;
const bool needsToTraverseChildrenNodes = processCurrentNode(siblingPos,
- mCorrection, &childCount, &firstChildPos, &siblingPos);
+ correction, &childCount, &firstChildPos, &siblingPos, queue);
// Update next sibling pos
- mCorrection->setTreeSiblingPos(outputIndex, siblingPos);
+ correction->setTreeSiblingPos(outputIndex, siblingPos);
if (needsToTraverseChildrenNodes) {
// Goes to child node
- outputIndex = mCorrection->goDownTree(outputIndex, childCount, firstChildPos);
+ outputIndex = correction->goDownTree(outputIndex, childCount, firstChildPos);
}
} else {
// Goes to parent sibling node
- outputIndex = mCorrection->getTreeParentIndex(outputIndex);
+ outputIndex = correction->getTreeParentIndex(outputIndex);
}
}
}
-void UnigramDictionary::getMissingSpaceWords(
+void UnigramDictionary::getMissingSpaceWords(ProximityInfo *proximityInfo, const int *xcoordinates,
+ const int *ycoordinates, const int *codes, const bool useFullEditDistance,
const int inputLength, const int missingSpacePos, Correction *correction,
- const bool useFullEditDistance) {
- correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
- -1 /* transposedPos */, -1 /* spaceProximityPos */, missingSpacePos,
- useFullEditDistance);
- getSplitTwoWordsSuggestion(inputLength, correction);
+ WordsPriorityQueuePool* queuePool) {
+ getSplitTwoWordsSuggestions(proximityInfo, xcoordinates, ycoordinates, codes,
+ useFullEditDistance, inputLength, missingSpacePos, -1/* spaceProximityPos */,
+ correction, queuePool);
}
-void UnigramDictionary::getMistypedSpaceWords(
+void UnigramDictionary::getMistypedSpaceWords(ProximityInfo *proximityInfo, const int *xcoordinates,
+ const int *ycoordinates, const int *codes, const bool useFullEditDistance,
const int inputLength, const int spaceProximityPos, Correction *correction,
- const bool useFullEditDistance) {
- correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
- -1 /* transposedPos */, spaceProximityPos, -1 /* missingSpacePos */,
- useFullEditDistance);
- getSplitTwoWordsSuggestion(inputLength, correction);
-}
-
-inline bool UnigramDictionary::needsToSkipCurrentNode(const unsigned short c,
- const int inputIndex, const int skipPos, const int depth) {
- const unsigned short userTypedChar = mProximityInfo->getPrimaryCharAt(inputIndex);
- // Skip the ' or other letter and continue deeper
- return (c == QUOTE && userTypedChar != QUOTE) || skipPos == depth;
+ WordsPriorityQueuePool* queuePool) {
+ getSplitTwoWordsSuggestions(proximityInfo, xcoordinates, ycoordinates, codes,
+ useFullEditDistance, inputLength, -1 /* missingSpacePos */, spaceProximityPos,
+ correction, queuePool);
}
-inline void UnigramDictionary::onTerminal(const int freq, Correction *correction) {
+inline void UnigramDictionary::onTerminal(const int freq,
+ const TerminalAttributes& terminalAttributes, Correction *correction,
+ WordsPriorityQueue *queue) {
int wordLength;
unsigned short* wordPointer;
const int finalFreq = correction->getFinalFreq(freq, &wordPointer, &wordLength);
if (finalFreq >= 0) {
- addWord(wordPointer, wordLength, finalFreq);
+ if (!terminalAttributes.isShortcutOnly()) {
+ addWord(wordPointer, wordLength, finalFreq, queue);
+ }
+ TerminalAttributes::ShortcutIterator iterator = terminalAttributes.getShortcutIterator();
+ while (iterator.hasNextShortcutTarget()) {
+ // TODO: addWord only supports weak ordering, meaning we have no means to control the
+ // order of the shortcuts relative to one another or to the word. We need to either
+ // modulate the frequency of each shortcut according to its own shortcut frequency or
+ // to make the queue so that the insert order is protected inside the queue for words
+ // with the same score.
+ uint16_t shortcutTarget[MAX_WORD_LENGTH_INTERNAL];
+ const int shortcutTargetStringLength = iterator.getNextShortcutTarget(
+ MAX_WORD_LENGTH_INTERNAL, shortcutTarget);
+ addWord(shortcutTarget, shortcutTargetStringLength, finalFreq, queue);
+ }
}
}
-void UnigramDictionary::getSplitTwoWordsSuggestion(
- const int inputLength, Correction* correction) {
- const int spaceProximityPos = correction->getSpaceProximityPos();
- const int missingSpacePos = correction->getMissingSpacePos();
+void UnigramDictionary::getSplitTwoWordsSuggestions(ProximityInfo *proximityInfo,
+ const int *xcoordinates, const int *ycoordinates, const int *codes,
+ const bool useFullEditDistance, const int inputLength, const int missingSpacePos,
+ const int spaceProximityPos, Correction *correction, WordsPriorityQueuePool* queuePool) {
+ WordsPriorityQueue *masterQueue = queuePool->getMasterQueue();
+
if (DEBUG_DICT) {
int inputCount = 0;
if (spaceProximityPos >= 0) ++inputCount;
@@ -408,9 +376,19 @@ void UnigramDictionary::getSplitTwoWordsSuggestion(
return;
const int newWordLength = firstWordLength + secondWordLength + 1;
+
+
+ // Space proximity preparation
+ //WordsPriorityQueue *subQueue = queuePool->getSubQueue1();
+ //initSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, firstWordLength, subQueue,
+ //correction);
+ //getSuggestionCandidates(useFullEditDistance, firstWordLength, correction, subQueue, false,
+ //MAX_ERRORS_FOR_TWO_WORDS);
+
// Allocating variable length array on stack
unsigned short word[newWordLength];
- const int firstFreq = getMostFrequentWordLike(firstWordStartPos, firstWordLength, mWord);
+ const int firstFreq = getMostFrequentWordLike(
+ firstWordStartPos, firstWordLength, proximityInfo, mWord);
if (DEBUG_DICT) {
LOGI("First freq: %d", firstFreq);
}
@@ -420,7 +398,8 @@ void UnigramDictionary::getSplitTwoWordsSuggestion(
word[i] = mWord[i];
}
- const int secondFreq = getMostFrequentWordLike(secondWordStartPos, secondWordLength, mWord);
+ const int secondFreq = getMostFrequentWordLike(
+ secondWordStartPos, secondWordLength, proximityInfo, mWord);
if (DEBUG_DICT) {
LOGI("Second freq: %d", secondFreq);
}
@@ -431,22 +410,29 @@ void UnigramDictionary::getSplitTwoWordsSuggestion(
word[i] = mWord[i - firstWordLength - 1];
}
- const int pairFreq = mCorrection->getFreqForSplitTwoWords(firstFreq, secondFreq, word);
+ // TODO: Remove initSuggestions and correction->setCorrectionParams
+ initSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, inputLength,
+ 0 /* do not clear queue */, correction);
+
+ correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
+ -1 /* transposedPos */, spaceProximityPos, missingSpacePos,
+ useFullEditDistance, false /* doAutoCompletion */, MAX_ERRORS_FOR_TWO_WORDS);
+ const int pairFreq = correction->getFreqForSplitTwoWords(firstFreq, secondFreq, word);
if (DEBUG_DICT) {
LOGI("Split two words: %d, %d, %d, %d", firstFreq, secondFreq, pairFreq, inputLength);
}
- addWord(word, newWordLength, pairFreq);
+ addWord(word, newWordLength, pairFreq, masterQueue);
return;
}
// Wrapper for getMostFrequentWordLikeInner, which matches it to the previous
// interface.
inline int UnigramDictionary::getMostFrequentWordLike(const int startInputIndex,
- const int inputLength, unsigned short *word) {
+ const int inputLength, ProximityInfo *proximityInfo, unsigned short *word) {
uint16_t inWord[inputLength];
for (int i = 0; i < inputLength; ++i) {
- inWord[i] = (uint16_t)mProximityInfo->getPrimaryCharAt(startInputIndex + i);
+ inWord[i] = (uint16_t)proximityInfo->getPrimaryCharAt(startInputIndex + i);
}
return getMostFrequentWordLikeInner(inWord, inputLength, word);
}
@@ -470,8 +456,8 @@ static inline bool testCharGroupForContinuedLikeness(const uint8_t flags,
const bool hasMultipleChars = (0 != (UnigramDictionary::FLAG_HAS_MULTIPLE_CHARS & flags));
int pos = startPos;
int32_t character = BinaryFormat::getCharCodeAndForwardPointer(root, &pos);
- int32_t baseChar = Dictionary::toBaseLowerCase(character);
- const uint16_t wChar = Dictionary::toBaseLowerCase(inWord[startInputIndex]);
+ int32_t baseChar = toBaseLowerCase(character);
+ const uint16_t wChar = toBaseLowerCase(inWord[startInputIndex]);
if (baseChar != wChar) {
*outPos = hasMultipleChars ? BinaryFormat::skipOtherCharacters(root, pos) : pos;
@@ -483,8 +469,8 @@ static inline bool testCharGroupForContinuedLikeness(const uint8_t flags,
if (hasMultipleChars) {
character = BinaryFormat::getCharCodeAndForwardPointer(root, &pos);
while (NOT_A_CHARACTER != character) {
- baseChar = Dictionary::toBaseLowerCase(character);
- if (Dictionary::toBaseLowerCase(inWord[++inputIndex]) != baseChar) {
+ baseChar = toBaseLowerCase(character);
+ if (toBaseLowerCase(inWord[++inputIndex]) != baseChar) {
*outPos = BinaryFormat::skipOtherCharacters(root, pos);
*outInputIndex = startInputIndex;
return false;
@@ -597,7 +583,7 @@ int UnigramDictionary::getBigramPosition(int pos, unsigned short *word, int offs
// given level, as output into newCount when traversing this level's parent.
inline bool UnigramDictionary::processCurrentNode(const int initialPos,
Correction *correction, int *newCount,
- int *newChildrenPosition, int *nextSiblingPosition) {
+ int *newChildrenPosition, int *nextSiblingPosition, WordsPriorityQueue *queue) {
if (DEBUG_DICT) {
correction->checkState();
}
@@ -648,7 +634,7 @@ inline bool UnigramDictionary::processCurrentNode(const int initialPos,
if (stateType == Correction::TRAVERSE_ALL_ON_TERMINAL
|| stateType == Correction::ON_TERMINAL) {
needsToInvokeOnTerminal = true;
- } else if (stateType == Correction::UNRELATED) {
+ } else if (stateType == Correction::UNRELATED || correction->needsToPrune()) {
// We found that this is an unrelated character, so we should give up traversing
// this node and its children entirely.
// However we may not be on the last virtual node yet so we skip the remaining
@@ -676,7 +662,9 @@ inline bool UnigramDictionary::processCurrentNode(const int initialPos,
// The frequency should be here, because we come here only if this is actually
// a terminal node, and we are on its last char.
const int freq = BinaryFormat::readFrequencyWithoutMovingPointer(DICT_ROOT, pos);
- onTerminal(freq, mCorrection);
+ TerminalAttributes terminalAttributes(DICT_ROOT, flags,
+ BinaryFormat::skipFrequency(flags, pos));
+ onTerminal(freq, terminalAttributes, correction, queue);
}
// If there are more chars in this node, then this virtual node has children.
generated by cgit v1.2.3-83-g751a (git 2.47.1) at 2025-05-31 19:59:37 +0000