Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit a7e5a5a6 authored by satok's avatar satok
Browse files

Add words priority queue pool

Change-Id: I152df7b876a1756b69ded2ca4fb3ee26b38c971f
parent 4d355989
Loading
Loading
Loading
Loading
+4 −0
Original line number Diff line number Diff line
@@ -202,6 +202,10 @@ static void dumpWord(const unsigned short* word, const int length) {
// This is only used for the size of array. Not to be used in c functions.
#define MAX_WORD_LENGTH_INTERNAL 48

// Word limit for sub queues used in WordsPriorityQueuePool.  Sub queues are temporary queues used
// for better performance.
#define SUB_QUEUE_MAX_WORDS 5

#define MAX_DEPTH_MULTIPLIER 3

// TODO: Reduce this constant if possible; check the maximum number of umlauts in the same German
+3 −2
Original line number Diff line number Diff line
@@ -39,7 +39,8 @@ Dictionary::Dictionary(void *dict, int dictSize, int mmapFd, int dictBufAdjust,
        }
    }
    mCorrection = new Correction(typedLetterMultiplier, fullWordMultiplier);
    mWordsPriorityQueue = new WordsPriorityQueue(maxWords, maxWordLength);
    mWordsPriorityQueuePool = new WordsPriorityQueuePool(
            maxWords, SUB_QUEUE_MAX_WORDS, maxWordLength);
    mUnigramDictionary = new UnigramDictionary(mDict, typedLetterMultiplier, fullWordMultiplier,
            maxWordLength, maxWords, maxAlternatives, IS_LATEST_DICT_VERSION);
    mBigramDictionary = new BigramDictionary(mDict, maxWordLength, maxAlternatives,
@@ -48,7 +49,7 @@ Dictionary::Dictionary(void *dict, int dictSize, int mmapFd, int dictBufAdjust,

Dictionary::~Dictionary() {
    delete mCorrection;
    delete mWordsPriorityQueue;
    delete mWordsPriorityQueuePool;
    delete mUnigramDictionary;
    delete mBigramDictionary;
}
+3 −3
Original line number Diff line number Diff line
@@ -23,7 +23,7 @@
#include "defines.h"
#include "proximity_info.h"
#include "unigram_dictionary.h"
#include "words_priority_queue.h"
#include "words_priority_queue_pool.h"

namespace latinime {

@@ -34,7 +34,7 @@ public:

    int getSuggestions(ProximityInfo *proximityInfo, int *xcoordinates, int *ycoordinates,
            int *codes, int codesSize, int flags, unsigned short *outWords, int *frequencies) {
        return mUnigramDictionary->getSuggestions(proximityInfo, mWordsPriorityQueue,
        return mUnigramDictionary->getSuggestions(proximityInfo, mWordsPriorityQueuePool,
                mCorrection, xcoordinates, ycoordinates, codes,
                codesSize, flags, outWords, frequencies);
    }
@@ -81,7 +81,7 @@ private:
    const bool IS_LATEST_DICT_VERSION;
    UnigramDictionary *mUnigramDictionary;
    BigramDictionary *mBigramDictionary;
    WordsPriorityQueue *mWordsPriorityQueue;
    WordsPriorityQueuePool *mWordsPriorityQueuePool;
    Correction *mCorrection;
};

+29 −23
Original line number Diff line number Diff line
@@ -93,7 +93,7 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
        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,
        WordsPriorityQueue *queue) {
        WordsPriorityQueuePool *queuePool) {

    if (currentDepth < MAX_UMLAUT_SEARCH_DEPTH) {
        for (int i = 0; i < codesRemain; ++i) {
@@ -110,7 +110,8 @@ 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, correction, queue);
                        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.
@@ -120,7 +121,7 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit
                getWordWithDigraphSuggestionsRec(proximityInfo, xcoordinates, ycoordinates,
                        codesBuffer, codesBufferSize, flags,
                        codesSrc + i * MAX_PROXIMITY_CHARS, codesRemain - i, currentDepth + 1,
                        codesDest + i * MAX_PROXIMITY_CHARS, correction, queue);
                        codesDest + i * MAX_PROXIMITY_CHARS, correction, queuePool);
                return;
            }
        }
@@ -137,27 +138,28 @@ void UnigramDictionary::getWordWithDigraphSuggestionsRec(ProximityInfo *proximit

    getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codesBuffer,
            (codesDest - codesBuffer) / MAX_PROXIMITY_CHARS + codesRemain, flags, correction,
            queue);
            queuePool);
}

int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, WordsPriorityQueue *queue,
        Correction *correction, const int *xcoordinates, const int *ycoordinates, const int *codes,
        const int codesSize, const int flags, unsigned short *outWords, int *frequencies) {
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) {

    WordsPriorityQueue* masterQueue = queue;
    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, masterCorrection, masterQueue);
                codesSize, flags, codes, codesSize, 0, codesBuffer, masterCorrection, queuePool);
    } else { // Normal processing
        getWordSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, codesSize, flags,
                masterCorrection, masterQueue);
                masterCorrection, queuePool);
    }

    PROF_START(20);
    const int suggestedWordsCount = masterQueue->outputSuggestions(frequencies, outWords);
    const int suggestedWordsCount =
            queuePool->getMasterQueue()->outputSuggestions(frequencies, outWords);

    if (DEBUG_DICT) {
        LOGI("Returning %d words", suggestedWordsCount);
@@ -178,11 +180,13 @@ int UnigramDictionary::getSuggestions(ProximityInfo *proximityInfo, WordsPriorit

void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
        const int *xcoordinates, const int *ycoordinates, const int *codes,
        const int inputLength, const int flags, Correction *correction, WordsPriorityQueue *queue) {
        const int inputLength, const int flags, Correction *correction,
        WordsPriorityQueuePool *queuePool) {
    WordsPriorityQueue *masterQueue = queuePool->getMasterQueue();

    PROF_OPEN;
    PROF_START(0);
    initSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, inputLength, queue);
    initSuggestions(proximityInfo, xcoordinates, ycoordinates, codes, inputLength, masterQueue);
    if (DEBUG_DICT) assert(codesSize == inputLength);

    const int maxDepth = min(inputLength * MAX_DEPTH_MULTIPLIER, MAX_WORD_LENGTH);
@@ -192,7 +196,7 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
    const bool useFullEditDistance = USE_FULL_EDIT_DISTANCE & flags;
    // TODO: remove
    PROF_START(1);
    getSuggestionCandidates(useFullEditDistance, inputLength, correction, queue);
    getSuggestionCandidates(useFullEditDistance, inputLength, correction, masterQueue);
    PROF_END(1);

    PROF_START(2);
@@ -216,7 +220,7 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
                LOGI("--- Suggest missing space characters %d", i);
            }
            getMissingSpaceWords(
                    inputLength, i, proximityInfo, correction, useFullEditDistance, queue);
                    inputLength, i, proximityInfo, correction, useFullEditDistance, queuePool);
        }
    }
    PROF_END(5);
@@ -235,8 +239,8 @@ void UnigramDictionary::getWordSuggestions(ProximityInfo *proximityInfo,
                        i, x, y, proximityInfo->hasSpaceProximity(x, y));
            }
            if (proximityInfo->hasSpaceProximity(x, y)) {
                getMistypedSpaceWords(
                        inputLength, i, proximityInfo, correction, useFullEditDistance, queue);
                getMistypedSpaceWords(inputLength, i, proximityInfo, correction,
                        useFullEditDistance, queuePool);
            }
        }
    }
@@ -293,20 +297,20 @@ void UnigramDictionary::getSuggestionCandidates(const bool useFullEditDistance,

void UnigramDictionary::getMissingSpaceWords(
        const int inputLength, const int missingSpacePos, ProximityInfo *proximityInfo,
        Correction *correction, const bool useFullEditDistance, WordsPriorityQueue *queue) {
        Correction *correction, const bool useFullEditDistance, WordsPriorityQueuePool *queuePool) {
    correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
            -1 /* transposedPos */, -1 /* spaceProximityPos */, missingSpacePos,
            useFullEditDistance, false /* doAutoCompletion */, MAX_ERRORS_FOR_TWO_WORDS);
    getSplitTwoWordsSuggestion(inputLength, proximityInfo, correction, queue);
    getSplitTwoWordsSuggestion(inputLength, proximityInfo, correction, queuePool);
}

void UnigramDictionary::getMistypedSpaceWords(
        const int inputLength, const int spaceProximityPos, ProximityInfo *proximityInfo,
        Correction *correction, const bool useFullEditDistance, WordsPriorityQueue *queue) {
        Correction *correction, const bool useFullEditDistance, WordsPriorityQueuePool *queuePool) {
    correction->setCorrectionParams(-1 /* skipPos */, -1 /* excessivePos */,
            -1 /* transposedPos */, spaceProximityPos, -1 /* missingSpacePos */,
            useFullEditDistance, false /* doAutoCompletion */, MAX_ERRORS_FOR_TWO_WORDS);
    getSplitTwoWordsSuggestion(inputLength, proximityInfo, correction, queue);
    getSplitTwoWordsSuggestion(inputLength, proximityInfo, correction, queuePool);
}

inline void UnigramDictionary::onTerminal(
@@ -321,7 +325,9 @@ inline void UnigramDictionary::onTerminal(

void UnigramDictionary::getSplitTwoWordsSuggestion(
        const int inputLength, ProximityInfo *proximityInfo, Correction *correction,
        WordsPriorityQueue *queue) {
        WordsPriorityQueuePool *queuePool) {
    WordsPriorityQueue *masterQueue = queuePool->getMasterQueue();

    const int spaceProximityPos = correction->getSpaceProximityPos();
    const int missingSpacePos = correction->getMissingSpacePos();
    if (DEBUG_DICT) {
@@ -373,7 +379,7 @@ void UnigramDictionary::getSplitTwoWordsSuggestion(
    if (DEBUG_DICT) {
        LOGI("Split two words:  %d, %d, %d, %d", firstFreq, secondFreq, pairFreq, inputLength);
    }
    addWord(word, newWordLength, pairFreq, queue);
    addWord(word, newWordLength, pairFreq, masterQueue);
    return;
}

+7 −6
Original line number Diff line number Diff line
@@ -23,6 +23,7 @@
#include "defines.h"
#include "proximity_info.h"
#include "words_priority_queue.h"
#include "words_priority_queue_pool.h"

namespace latinime {

@@ -70,7 +71,7 @@ public:
            const bool isLatestDictVersion);
    bool isValidWord(const uint16_t* const inWord, const int length) const;
    int getBigramPosition(int pos, unsigned short *word, int offset, int length) const;
    int getSuggestions(ProximityInfo *proximityInfo, WordsPriorityQueue *queue,
    int 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);
@@ -80,13 +81,13 @@ private:

    void getWordSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
            const int *ycoordinates, const int *codes, const int inputLength,
            const int flags, Correction *correction, WordsPriorityQueue *queue);
            const int flags, Correction *correction, WordsPriorityQueuePool *queuePool);
    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,
        const int codesBufferSize, const int flags, const int* codesSrc,
        const int codesRemain, const int currentDepth, int* codesDest, Correction *correction,
        WordsPriorityQueue* queue);
        WordsPriorityQueuePool* queuePool);
    void initSuggestions(ProximityInfo *proximityInfo, const int *xcoordinates,
            const int *ycoordinates, const int *codes, const int codesSize,
            WordsPriorityQueue *queue);
@@ -94,13 +95,13 @@ private:
            const bool useFullEditDistance, const int inputLength, Correction *correction,
            WordsPriorityQueue* queue);
    void getSplitTwoWordsSuggestion(const int inputLength, ProximityInfo *proximityInfo,
            Correction *correction, WordsPriorityQueue *queue);
            Correction *correction, WordsPriorityQueuePool *queuePool);
    void getMissingSpaceWords(const int inputLength, const int missingSpacePos,
            ProximityInfo *proximityInfo, Correction *correction,
            const bool useFullEditDistance, WordsPriorityQueue *queue);
            const bool useFullEditDistance, WordsPriorityQueuePool *queuePool);
    void getMistypedSpaceWords(const int inputLength, const int spaceProximityPos,
            ProximityInfo *proximityInfo, Correction *correction,
            const bool useFullEditDistance, WordsPriorityQueue *queue);
            const bool useFullEditDistance, WordsPriorityQueuePool *queuePool);
    void onTerminal(const int freq, Correction *correction, WordsPriorityQueue *queue);
    bool needsToSkipCurrentNode(const unsigned short c,
            const int inputIndex, const int skipPos, const int depth);
Loading