diff --git a/native/jni/src/dictionary.h b/native/jni/src/dictionary.h
index dea5763d0496ec78c8adbca632f6c232433d285c..5b9ddb3e9ccfcf9d84e05cf0e16c122f4c04b95d 100644
--- a/native/jni/src/dictionary.h
+++ b/native/jni/src/dictionary.h
@@ -35,9 +35,13 @@ class Dictionary {
int getSuggestions(ProximityInfo *proximityInfo, int *xcoordinates, int *ycoordinates,
int *codes, int codesSize, bool useFullEditDistance, unsigned short *outWords,
int *frequencies) {
+ // bigramListPosition is, as an int, the offset of the bigram list in the file.
+ // If none, it's zero.
+ // TODO: get this from the bigram dictionary instance
+ const int bigramListPosition = 0;
return mUnigramDictionary->getSuggestions(proximityInfo, mWordsPriorityQueuePool,
- mCorrection, xcoordinates, ycoordinates, codes,
- codesSize, useFullEditDistance, outWords, frequencies);
+ mCorrection, xcoordinates, ycoordinates, codes, codesSize, bigramListPosition,
+ useFullEditDistance, outWords, frequencies);
}
int getBigrams(const int32_t *word, int length, int *codes, int codesSize,
diff --git a/native/jni/src/unigram_dictionary.cpp b/native/jni/src/unigram_dictionary.cpp
index a3eda006192a4e9148afe19f2923692910d85f00..0c759d438aa5970540710b0714142134663516ba 100644
--- a/native/jni/src/unigram_dictionary.cpp
+++ b/native/jni/src/unigram_dictionary.cpp
@@ -98,7 +98,8 @@ int UnigramDictionary::getDigraphReplacement(const int *codes, const int i, cons
void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
const int *xcoordinates, const int *ycoordinates, const int *codesBuffer,
int *xCoordinatesBuffer, int *yCoordinatesBuffer,
- const int codesBufferSize, const bool useFullEditDistance, const int *codesSrc,
+ const int codesBufferSize, const int bigramListPosition,
+ const bool useFullEditDistance, const int *codesSrc,
const int codesRemain, const int currentDepth, int *codesDest, Correction *correction,
WordsPriorityQueuePool *queuePool,
const digraph_t* const digraphs, const unsigned int digraphsSize) {
@@ -127,8 +128,8 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
replacementCodePoint;
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, xCoordinatesBuffer, yCoordinatesBuffer, codesBufferSize,
- useFullEditDistance, codesSrc + i + 1, codesRemain - i - 1,
- currentDepth + 1, codesDest + i, correction,
+ bigramListPosition, useFullEditDistance, codesSrc + i + 1,
+ codesRemain - i - 1, currentDepth + 1, codesDest + i, correction,
queuePool, digraphs, digraphsSize);
// Copy the second char of the digraph in place, then continue processing on
@@ -137,9 +138,9 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
memcpy(codesDest + i, codesSrc + i, BYTES_IN_ONE_CHAR);
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
codesBuffer, xCoordinatesBuffer, yCoordinatesBuffer, codesBufferSize,
- useFullEditDistance, codesSrc + i, codesRemain - i, currentDepth + 1,
- codesDest + i, correction, queuePool,
- digraphs, digraphsSize);
+ bigramListPosition, useFullEditDistance, codesSrc + i, codesRemain - i,
+ currentDepth + 1, codesDest + i, correction, queuePool, digraphs,
+ digraphsSize);
return;
}
}
@@ -160,14 +161,16 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
}
getWordSuggestions(proximityInfo, xCoordinatesBuffer, yCoordinatesBuffer, codesBuffer,
- startIndex + codesRemain, useFullEditDistance, correction,
+ startIndex + codesRemain, bigramListPosition, useFullEditDistance, correction,
queuePool);
}
+// bigramListPosition is the offset in the file to the list of bigrams for the previous word.
int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
WordsPriorityQueuePool *queuePool, Correction *correction, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize,
- const bool useFullEditDistance, unsigned short *outWords, int *frequencies) {
+ const int bigramListPosition, const bool useFullEditDistance, unsigned short *outWords,
+ int *frequencies) {
queuePool->clearAll();
Correction* masterCorrection = correction;
@@ -177,8 +180,8 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
int xCoordinatesBuffer[codesSize];
int yCoordinatesBuffer[codesSize];
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
- xCoordinatesBuffer, yCoordinatesBuffer,
- codesSize, useFullEditDistance, codes, codesSize, 0, codesBuffer, masterCorrection,
+ xCoordinatesBuffer, yCoordinatesBuffer, codesSize, bigramListPosition,
+ useFullEditDistance, codes, codesSize, 0, codesBuffer, masterCorrection,
queuePool, GERMAN_UMLAUT_DIGRAPHS,
sizeof(GERMAN_UMLAUT_DIGRAPHS) / sizeof(GERMAN_UMLAUT_DIGRAPHS[0]));
} else if (BinaryFormat::REQUIRES_FRENCH_LIGATURES_PROCESSING & FLAGS) {
@@ -186,13 +189,13 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
int xCoordinatesBuffer[codesSize];
int yCoordinatesBuffer[codesSize];
getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
- xCoordinatesBuffer, yCoordinatesBuffer,
- codesSize, useFullEditDistance, codes, codesSize, 0, codesBuffer, masterCorrection,
+ xCoordinatesBuffer, yCoordinatesBuffer, codesSize, bigramListPosition,
+ useFullEditDistance, codes, codesSize, 0, codesBuffer, masterCorrection,
queuePool, FRENCH_LIGATURES_DIGRAPHS,
sizeof(FRENCH_LIGATURES_DIGRAPHS) / sizeof(FRENCH_LIGATURES_DIGRAPHS[0]));
} else { // Normal processing
getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize,
- useFullEditDistance, masterCorrection, queuePool);
+ bigramListPosition, useFullEditDistance, masterCorrection, queuePool);
}
PROF_START(20);
@@ -225,16 +228,16 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo,
void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
const int *xcoordinates, const int *ycoordinates, const int *codes,
- const int inputLength, const bool useFullEditDistance, Correction *correction,
- WordsPriorityQueuePool *queuePool) {
+ const int inputLength, const int bigramListPosition, const bool useFullEditDistance,
+ Correction *correction, WordsPriorityQueuePool *queuePool) {
PROF_OPEN;
PROF_START(0);
PROF_END(0);
PROF_START(1);
- getOneWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, useFullEditDistance,
- inputLength, correction, queuePool);
+ getOneWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, bigramListPosition,
+ useFullEditDistance, inputLength, correction, queuePool);
PROF_END(1);
PROF_START(2);
@@ -305,15 +308,16 @@ static const char SPACE = ' ';
void UnigramDictionary::getOneWordSuggestions(ProximityInfo *proximityInfo,
const int *xcoordinates, const int *ycoordinates, const int *codes,
- const bool useFullEditDistance, const int inputLength, Correction *correction,
- WordsPriorityQueuePool *queuePool) {
+ const int bigramListPosition, const bool useFullEditDistance, const int inputLength,
+ Correction *correction, WordsPriorityQueuePool *queuePool) {
initSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, inputLength, correction);
- getSuggestionCandidates(useFullEditDistance, inputLength, correction, queuePool,
- true /* doAutoCompletion */, DEFAULT_MAX_ERRORS, FIRST_WORD_INDEX);
+ getSuggestionCandidates(useFullEditDistance, inputLength, bigramListPosition, correction,
+ queuePool, true /* doAutoCompletion */, DEFAULT_MAX_ERRORS, FIRST_WORD_INDEX);
}
void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance,
- const int inputLength, Correction *correction, WordsPriorityQueuePool *queuePool,
+ const int inputLength, const int bigramListPosition,
+ Correction *correction, WordsPriorityQueuePool *queuePool,
const bool doAutoCompletion, const int maxErrors, const int currentWordIndex) {
// TODO: Remove setCorrectionParams
correction->setCorrectionParams(0, 0, 0,
@@ -333,8 +337,8 @@ void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance,
int firstChildPos;
const bool needsToTraverseChildrenNodes = processCurrentNode(siblingPos,
- correction, &childCount, &firstChildPos, &siblingPos, queuePool,
- currentWordIndex);
+ bigramListPosition, correction, &childCount, &firstChildPos, &siblingPos,
+ queuePool, currentWordIndex);
// Update next sibling pos
correction->setTreeSiblingPos(outputIndex, siblingPos);
@@ -426,8 +430,10 @@ bool UnigramDictionary::getSubStringSuggestion(
initSuggestions(proximityInfo, &xcoordinates[offset], &ycoordinates[offset],
codes + offset, inputWordLength, correction);
queuePool->clearSubQueue(currentWordIndex);
- getSuggestionCandidates(useFullEditDistance, inputWordLength, correction,
- queuePool, false, MAX_ERRORS_FOR_TWO_WORDS, currentWordIndex);
+ // TODO: pass the bigram list for substring suggestion
+ getSuggestionCandidates(useFullEditDistance, inputWordLength,
+ 0 /* bigramListPosition */, correction, queuePool, false /* doAutoCompletion */,
+ MAX_ERRORS_FOR_TWO_WORDS, currentWordIndex);
if (DEBUG_DICT) {
if (currentWordIndex < MULTIPLE_WORDS_SUGGESTION_MAX_WORDS) {
AKLOGI("Dump word candidates(%d) %d", currentWordIndex, inputWordLength);
@@ -757,15 +763,13 @@ int UnigramDictionary::getBigramPosition(int pos, unsigned short *word, int offs
// the current node in nextSiblingPosition. Thus, the caller must keep count of the nodes at any
// given level, as output into newCount when traversing this level's parent.
inline bool UnigramDictionary::processCurrentNode(const int initialPos,
- Correction *correction, int *newCount,
+ const int bigramListPosition, Correction *correction, int *newCount,
int *newChildrenPosition, int *nextSiblingPosition, WordsPriorityQueuePool *queuePool,
const int currentWordIndex) {
if (DEBUG_DICT) {
correction->checkState();
}
int pos = initialPos;
- // TODO: get this as an argument
- const int bigramListPosition = 0;
// Flags contain the following information:
// - Address type (MASK_GROUP_ADDRESS_TYPE) on two bits:
@@ -842,6 +846,8 @@ inline bool UnigramDictionary::processCurrentNode(const int initialPos,
const int childrenAddressPos = BinaryFormat::skipFrequency(flags, pos);
const int attributesPos = BinaryFormat::skipChildrenPosition(flags, childrenAddressPos);
TerminalAttributes terminalAttributes(DICT_ROOT, flags, attributesPos);
+ // The bigramListPosition is the offset in the file of the bigrams for the previous word,
+ // or zero if we don't know of any bigrams for it.
const int probability = BinaryFormat::getProbability(bigramListPosition, unigramFreq);
onTerminal(probability, terminalAttributes, correction, queuePool, needsToInvokeOnTerminal,
currentWordIndex);
diff --git a/native/jni/src/unigram_dictionary.h b/native/jni/src/unigram_dictionary.h
index b09c0006d403cc88c577a589afeb0b2c3b2371e9..0cc59bac87e27959c18b5fc8fb915ca9114ec4f7 100644
--- a/native/jni/src/unigram_dictionary.h
+++ b/native/jni/src/unigram_dictionary.h
@@ -74,34 +74,35 @@ class UnigramDictionary {
bool isValidWord(const int32_t* const inWord, const int length) const;
int getBigramPosition(int pos, unsigned short *word, int offset, int length) const;
int getSuggestions(ProximityInfo *proximityInfo, WordsPriorityQueuePool *queuePool,
- Correction *correction, const int *xcoordinates,
- const int *ycoordinates, const int *codes, const int codesSize,
+ Correction *correction, const int *xcoordinates, const int *ycoordinates,
+ const int *codes, const int codesSize, const int bigramListPosition,
const bool useFullEditDistance, unsigned short *outWords, int *frequencies);
virtual ~UnigramDictionary();
private:
void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int inputLength,
- const bool useFullEditDistance, Correction *correction,
+ const int bigramListPosition, const bool useFullEditDistance, Correction *correction,
WordsPriorityQueuePool *queuePool);
int getDigraphReplacement(const int *codes, const int i, const int codesSize,
const digraph_t* const digraphs, const unsigned int digraphsSize) const;
void getWordWithDigraphSuggestionsRec(ProximityInfo *proximityInfo,
const int *xcoordinates, const int* ycoordinates, const int *codesBuffer,
- int *xCoordinatesBuffer, int *yCoordinatesBuffer,
- const int codesBufferSize, const bool useFullEditDistance, const int* codesSrc,
+ int *xCoordinatesBuffer, int *yCoordinatesBuffer, const int codesBufferSize,
+ const int bigramListPosition, const bool useFullEditDistance, const int* codesSrc,
const int codesRemain, const int currentDepth, int* codesDest, Correction *correction,
WordsPriorityQueuePool* queuePool, const digraph_t* const digraphs,
const unsigned int digraphsSize);
void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
const int *ycoordinates, const int *codes, const int codesSize, Correction *correction);
void getOneWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
- const int *ycoordinates, const int *codes, const bool useFullEditDistance,
- const int inputLength, Correction *correction, WordsPriorityQueuePool* queuePool);
- void getSuggestionCandidates(
+ const int *ycoordinates, const int *codes, const int bigramListPosition,
const bool useFullEditDistance, const int inputLength, Correction *correction,
- WordsPriorityQueuePool* queuePool, const bool doAutoCompletion, const int maxErrors,
- const int currentWordIndex);
+ WordsPriorityQueuePool* queuePool);
+ void getSuggestionCandidates(
+ const bool useFullEditDistance, const int inputLength, const int bigramListPosition,
+ Correction *correction, WordsPriorityQueuePool* queuePool, const bool doAutoCompletion,
+ const int maxErrors, const int currentWordIndex);
void getSplitMultipleWordsSuggestions(ProximityInfo *proximityInfo,
const int *xcoordinates, const int *ycoordinates, const int *codes,
const bool useFullEditDistance, const int inputLength,
@@ -113,9 +114,9 @@ class UnigramDictionary {
bool needsToSkipCurrentNode(const unsigned short c,
const int inputIndex, const int skipPos, const int depth);
// Process a node by considering proximity, missing and excessive character
- bool processCurrentNode(const int initialPos, Correction *correction, int *newCount,
- int *newChildPosition, int *nextSiblingPosition, WordsPriorityQueuePool *queuePool,
- const int currentWordIndex);
+ bool processCurrentNode(const int initialPos, const int bigramListPosition,
+ Correction *correction, int *newCount, int *newChildPosition, int *nextSiblingPosition,
+ WordsPriorityQueuePool *queuePool, const int currentWordIndex);
int getMostFrequentWordLike(const int startInputIndex, const int inputLength,
ProximityInfo *proximityInfo, unsigned short *word);
int getMostFrequentWordLikeInner(const uint16_t* const inWord, const int length,