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

Commit cc3e5c7c authored by Amith Yamasani's avatar Amith Yamasani Committed by The Android Open Source Project
Browse files

AI 143659: am: CL 143472 Reduce dictionary size. 

  Changed the tree structure to have variable length nodes to save an average of 21% on the dictionary size.
  Created a shortened English dictionary for Dream - 50K words.
  Added a shortened Spanish dictionary for Dream - 32K words.
  Original author: yamasani
  Merged from: //branches/cupcake/...

Automated import of CL 143659
parent 152d8a79

dictionary/src/dictionary.cpp

+13 −6
Original line number Diff line number Diff line
@@ -85,10 +85,14 @@ int 
Dictionary::getAddress(int *pos)

{

    int address = 0;

    address += (mDict[*pos] & 0x7F) << 16;

    if ((mDict[*pos] & FLAG_ADDRESS_MASK) == 0) {

        *pos += 1;

    } else {

        address += (mDict[*pos] & (ADDRESS_MASK >> 16)) << 16;

        address += (mDict[*pos + 1] & 0xFF) << 8;

        address += (mDict[*pos + 2] & 0xFF);

        *pos += 3;

    }

    return address;

}
@@ -193,7 +197,8 @@ Dictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int s 
        unsigned short lowerC = toLowerCase(c, depth);

        bool terminal = getTerminal(&pos);

        int childrenAddress = getAddress(&pos);

        int freq = getFreq(&pos);

        int freq = 1;

        if (terminal) freq = getFreq(&pos);

        // If we are only doing completions, no need to look at the typed characters.

        if (completion) {

            mWord[depth] = c;
@@ -266,7 +271,9 @@ Dictionary::isValidWordRec(int pos, unsigned short *word, int offset, int length 
                }

            }

        }

        if (terminal) {

            getFreq(&pos);

        }

        // There could be two instances of each alphabet - upper and lower case. So continue

        // looking ...

    }

dictionary/src/dictionary.h

+16 −7
Original line number Diff line number Diff line
@@ -19,6 +19,15 @@ 


namespace latinime {



// 22-bit address = ~4MB dictionary size limit, which on average would be about 200k-300k words

#define ADDRESS_MASK 0x3FFFFF



// The bit that decides if an address follows in the next 22 bits

#define FLAG_ADDRESS_MASK 0x40

// The bit that decides if this is a terminal node for a word. The node could still have children,

// if the word has other endings.

#define FLAG_TERMINAL_MASK 0x80



class Dictionary {

public:

    Dictionary(void *dict, int typedLetterMultipler, int fullWordMultiplier);
@@ -32,7 +41,7 @@ public: 
private:



    int getAddress(int *pos);

    bool getTerminal(int *pos) { return (mDict[*pos] & 0x80) > 0; }

    bool getTerminal(int *pos) { return (mDict[*pos] & FLAG_TERMINAL_MASK) > 0; }

    int getFreq(int *pos) { return mDict[(*pos)++] & 0xFF; }

    int getCount(int *pos) { return mDict[(*pos)++] & 0xFF; }

    unsigned short getChar(int *pos);