mirror of
https://gitlab.futo.org/keyboard/latinime.git
synced 2024-09-28 14:54:30 +01:00
Detach bigram functionarities from unigram_dictionary
Change-Id: Ie35164a5f293e5370885a1ba13d6ed7caf6000ec
This commit is contained in:
parent
9fee221248
commit
18c28f431e
@ -15,19 +15,240 @@
|
|||||||
** limitations under the License.
|
** limitations under the License.
|
||||||
*/
|
*/
|
||||||
|
|
||||||
|
#include <string.h>
|
||||||
|
|
||||||
#define LOG_TAG "LatinIME: bigram_dictionary.cpp"
|
#define LOG_TAG "LatinIME: bigram_dictionary.cpp"
|
||||||
|
|
||||||
#include "bigram_dictionary.h"
|
#include "bigram_dictionary.h"
|
||||||
|
#include "dictionary.h"
|
||||||
|
|
||||||
namespace latinime {
|
namespace latinime {
|
||||||
|
|
||||||
BigramDictionary::BigramDictionary(void *dict, int typedLetterMultiplier, int fullWordMultiplier,
|
BigramDictionary::BigramDictionary(const unsigned char *dict, int maxWordLength,
|
||||||
int maxWordLength, int maxWords, int maxAlternatives, Dictionary *parentDictionary)
|
int maxAlternatives, const bool isLatestDictVersion, const bool hasBigram,
|
||||||
{
|
Dictionary *parentDictionary)
|
||||||
|
: DICT(dict), MAX_WORD_LENGTH(maxWordLength),
|
||||||
|
MAX_ALTERNATIVES(maxAlternatives), IS_LATEST_DICT_VERSION(isLatestDictVersion),
|
||||||
|
HAS_BIGRAM(hasBigram), mParentDictionary(parentDictionary) {
|
||||||
|
LOGI("BigramDictionary - constructor");
|
||||||
|
LOGI("Has Bigram : %d \n", hasBigram);
|
||||||
}
|
}
|
||||||
|
|
||||||
BigramDictionary::~BigramDictionary()
|
BigramDictionary::~BigramDictionary() {
|
||||||
{
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
bool BigramDictionary::addWordBigram(unsigned short *word, int length, int frequency) {
|
||||||
|
word[length] = 0;
|
||||||
|
if (DEBUG_DICT) {
|
||||||
|
char s[length + 1];
|
||||||
|
for (int i = 0; i <= length; i++) s[i] = word[i];
|
||||||
|
LOGI("Bigram: Found word = %s, freq = %d : \n", s, frequency);
|
||||||
|
}
|
||||||
|
|
||||||
|
// Find the right insertion point
|
||||||
|
int insertAt = 0;
|
||||||
|
while (insertAt < mMaxBigrams) {
|
||||||
|
if (frequency > mBigramFreq[insertAt] || (mBigramFreq[insertAt] == frequency
|
||||||
|
&& length < Dictionary::wideStrLen(mBigramChars + insertAt * MAX_WORD_LENGTH))) {
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
insertAt++;
|
||||||
|
}
|
||||||
|
LOGI("Bigram: InsertAt -> %d maxBigrams: %d\n", insertAt, mMaxBigrams);
|
||||||
|
if (insertAt < mMaxBigrams) {
|
||||||
|
memmove((char*) mBigramFreq + (insertAt + 1) * sizeof(mBigramFreq[0]),
|
||||||
|
(char*) mBigramFreq + insertAt * sizeof(mBigramFreq[0]),
|
||||||
|
(mMaxBigrams - insertAt - 1) * sizeof(mBigramFreq[0]));
|
||||||
|
mBigramFreq[insertAt] = frequency;
|
||||||
|
memmove((char*) mBigramChars + (insertAt + 1) * MAX_WORD_LENGTH * sizeof(short),
|
||||||
|
(char*) mBigramChars + (insertAt ) * MAX_WORD_LENGTH * sizeof(short),
|
||||||
|
(mMaxBigrams - insertAt - 1) * sizeof(short) * MAX_WORD_LENGTH);
|
||||||
|
unsigned short *dest = mBigramChars + (insertAt ) * MAX_WORD_LENGTH;
|
||||||
|
while (length--) {
|
||||||
|
*dest++ = *word++;
|
||||||
|
}
|
||||||
|
*dest = 0; // NULL terminate
|
||||||
|
if (DEBUG_DICT) LOGI("Bigram: Added word at %d\n", insertAt);
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
|
||||||
|
int BigramDictionary::getBigramAddress(int *pos, bool advance) {
|
||||||
|
int address = 0;
|
||||||
|
|
||||||
|
address += (DICT[*pos] & 0x3F) << 16;
|
||||||
|
address += (DICT[*pos + 1] & 0xFF) << 8;
|
||||||
|
address += (DICT[*pos + 2] & 0xFF);
|
||||||
|
|
||||||
|
if (advance) {
|
||||||
|
*pos += 3;
|
||||||
|
}
|
||||||
|
|
||||||
|
return address;
|
||||||
|
}
|
||||||
|
|
||||||
|
int BigramDictionary::getBigramFreq(int *pos) {
|
||||||
|
int freq = DICT[(*pos)++] & FLAG_BIGRAM_FREQ;
|
||||||
|
|
||||||
|
return freq;
|
||||||
|
}
|
||||||
|
|
||||||
|
|
||||||
|
int BigramDictionary::getBigrams(unsigned short *prevWord, int prevWordLength, int *codes,
|
||||||
|
int codesSize, unsigned short *bigramChars, int *bigramFreq, int maxWordLength,
|
||||||
|
int maxBigrams, int maxAlternatives) {
|
||||||
|
mBigramFreq = bigramFreq;
|
||||||
|
mBigramChars = bigramChars;
|
||||||
|
mInputCodes = codes;
|
||||||
|
mInputLength = codesSize;
|
||||||
|
mMaxBigrams = maxBigrams;
|
||||||
|
|
||||||
|
if (HAS_BIGRAM && IS_LATEST_DICT_VERSION) {
|
||||||
|
int pos = mParentDictionary->isValidWordRec(
|
||||||
|
DICTIONARY_HEADER_SIZE, prevWord, 0, prevWordLength);
|
||||||
|
LOGI("Pos -> %d\n", pos);
|
||||||
|
if (pos < 0) {
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
|
|
||||||
|
int bigramCount = 0;
|
||||||
|
int bigramExist = (DICT[pos] & FLAG_BIGRAM_READ);
|
||||||
|
if (bigramExist > 0) {
|
||||||
|
int nextBigramExist = 1;
|
||||||
|
while (nextBigramExist > 0 && bigramCount < maxBigrams) {
|
||||||
|
int bigramAddress = getBigramAddress(&pos, true);
|
||||||
|
int frequency = (FLAG_BIGRAM_FREQ & DICT[pos]);
|
||||||
|
// search for all bigrams and store them
|
||||||
|
searchForTerminalNode(bigramAddress, frequency);
|
||||||
|
nextBigramExist = (DICT[pos++] & FLAG_BIGRAM_CONTINUED);
|
||||||
|
bigramCount++;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
return bigramCount;
|
||||||
|
}
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
|
|
||||||
|
void BigramDictionary::searchForTerminalNode(int addressLookingFor, int frequency) {
|
||||||
|
// track word with such address and store it in an array
|
||||||
|
unsigned short word[MAX_WORD_LENGTH];
|
||||||
|
|
||||||
|
int pos;
|
||||||
|
int followDownBranchAddress = DICTIONARY_HEADER_SIZE;
|
||||||
|
bool found = false;
|
||||||
|
char followingChar = ' ';
|
||||||
|
int depth = -1;
|
||||||
|
|
||||||
|
while(!found) {
|
||||||
|
bool followDownAddressSearchStop = false;
|
||||||
|
bool firstAddress = true;
|
||||||
|
bool haveToSearchAll = true;
|
||||||
|
|
||||||
|
if (depth >= 0) {
|
||||||
|
word[depth] = (unsigned short) followingChar;
|
||||||
|
}
|
||||||
|
pos = followDownBranchAddress; // pos start at count
|
||||||
|
int count = DICT[pos] & 0xFF;
|
||||||
|
LOGI("count - %d\n",count);
|
||||||
|
pos++;
|
||||||
|
for (int i = 0; i < count; i++) {
|
||||||
|
// pos at data
|
||||||
|
pos++;
|
||||||
|
// pos now at flag
|
||||||
|
if (!getFirstBitOfByte(&pos)) { // non-terminal
|
||||||
|
if (!followDownAddressSearchStop) {
|
||||||
|
int addr = getBigramAddress(&pos, false);
|
||||||
|
if (addr > addressLookingFor) {
|
||||||
|
followDownAddressSearchStop = true;
|
||||||
|
if (firstAddress) {
|
||||||
|
firstAddress = false;
|
||||||
|
haveToSearchAll = true;
|
||||||
|
} else if (!haveToSearchAll) {
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
followDownBranchAddress = addr;
|
||||||
|
followingChar = (char)(0xFF & DICT[pos-1]);
|
||||||
|
if (firstAddress) {
|
||||||
|
firstAddress = false;
|
||||||
|
haveToSearchAll = false;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
pos += 3;
|
||||||
|
} else if (getFirstBitOfByte(&pos)) { // terminal
|
||||||
|
if (addressLookingFor == (pos-1)) { // found !!
|
||||||
|
depth++;
|
||||||
|
word[depth] = (0xFF & DICT[pos-1]);
|
||||||
|
found = true;
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
if (getSecondBitOfByte(&pos)) { // address + freq (4 byte)
|
||||||
|
if (!followDownAddressSearchStop) {
|
||||||
|
int addr = getBigramAddress(&pos, false);
|
||||||
|
if (addr > addressLookingFor) {
|
||||||
|
followDownAddressSearchStop = true;
|
||||||
|
if (firstAddress) {
|
||||||
|
firstAddress = false;
|
||||||
|
haveToSearchAll = true;
|
||||||
|
} else if (!haveToSearchAll) {
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
followDownBranchAddress = addr;
|
||||||
|
followingChar = (char)(0xFF & DICT[pos-1]);
|
||||||
|
if (firstAddress) {
|
||||||
|
firstAddress = false;
|
||||||
|
haveToSearchAll = true;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
pos += 4;
|
||||||
|
} else { // freq only (2 byte)
|
||||||
|
pos += 2;
|
||||||
|
}
|
||||||
|
|
||||||
|
// skipping bigram
|
||||||
|
int bigramExist = (DICT[pos] & FLAG_BIGRAM_READ);
|
||||||
|
if (bigramExist > 0) {
|
||||||
|
int nextBigramExist = 1;
|
||||||
|
while (nextBigramExist > 0) {
|
||||||
|
pos += 3;
|
||||||
|
nextBigramExist = (DICT[pos++] & FLAG_BIGRAM_CONTINUED);
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
pos++;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
}
|
||||||
|
depth++;
|
||||||
|
if (followDownBranchAddress == 0) {
|
||||||
|
LOGI("ERROR!!! Cannot find bigram!!");
|
||||||
|
break;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
if (checkFirstCharacter(word)) {
|
||||||
|
addWordBigram(word, depth, frequency);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
bool BigramDictionary::checkFirstCharacter(unsigned short *word) {
|
||||||
|
// Checks whether this word starts with same character or neighboring characters of
|
||||||
|
// what user typed.
|
||||||
|
|
||||||
|
int *inputCodes = mInputCodes;
|
||||||
|
int maxAlt = MAX_ALTERNATIVES;
|
||||||
|
while (maxAlt > 0) {
|
||||||
|
if ((unsigned int) *inputCodes == (unsigned int) *word) {
|
||||||
|
return true;
|
||||||
|
}
|
||||||
|
inputCodes++;
|
||||||
|
maxAlt--;
|
||||||
|
}
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
|
||||||
// TODO: Move functions related to bigram to here
|
// TODO: Move functions related to bigram to here
|
||||||
} // namespace latinime
|
} // namespace latinime
|
||||||
|
@ -22,10 +22,33 @@ namespace latinime {
|
|||||||
class Dictionary;
|
class Dictionary;
|
||||||
class BigramDictionary {
|
class BigramDictionary {
|
||||||
public:
|
public:
|
||||||
BigramDictionary(void *dict, int typedLetterMultipler, int fullWordMultiplier, int maxWordLength,
|
BigramDictionary(const unsigned char *dict, int maxWordLength, int maxAlternatives,
|
||||||
int maxWords, int maxAlternatives, Dictionary *parentDictionary);
|
const bool isLatestDictVersion, const bool hasBigram, Dictionary *parentDictionary);
|
||||||
|
int getBigrams(unsigned short *word, int length, int *codes, int codesSize,
|
||||||
|
unsigned short *outWords, int *frequencies, int maxWordLength, int maxBigrams,
|
||||||
|
int maxAlternatives);
|
||||||
~BigramDictionary();
|
~BigramDictionary();
|
||||||
private:
|
private:
|
||||||
|
bool addWordBigram(unsigned short *word, int length, int frequency);
|
||||||
|
int getBigramAddress(int *pos, bool advance);
|
||||||
|
int getBigramFreq(int *pos);
|
||||||
|
void searchForTerminalNode(int addressLookingFor, int frequency);
|
||||||
|
bool getFirstBitOfByte(int *pos) { return (DICT[*pos] & 0x80) > 0; }
|
||||||
|
bool getSecondBitOfByte(int *pos) { return (DICT[*pos] & 0x40) > 0; }
|
||||||
|
bool checkFirstCharacter(unsigned short *word);
|
||||||
|
|
||||||
|
const unsigned char *DICT;
|
||||||
|
const int MAX_WORD_LENGTH;
|
||||||
|
const int MAX_ALTERNATIVES;
|
||||||
|
const bool IS_LATEST_DICT_VERSION;
|
||||||
|
const bool HAS_BIGRAM;
|
||||||
|
|
||||||
|
Dictionary *mParentDictionary;
|
||||||
|
int *mBigramFreq;
|
||||||
|
int mMaxBigrams;
|
||||||
|
unsigned short *mBigramChars;
|
||||||
|
int *mInputCodes;
|
||||||
|
int mInputLength;
|
||||||
};
|
};
|
||||||
// ----------------------------------------------------------------------------
|
// ----------------------------------------------------------------------------
|
||||||
}; // namespace latinime
|
}; // namespace latinime
|
||||||
|
@ -31,10 +31,9 @@ Dictionary::Dictionary(void *dict, int typedLetterMultiplier, int fullWordMultip
|
|||||||
{
|
{
|
||||||
LOGI("IN NATIVE SUGGEST Version: %d \n", (DICT[0] & 0xFF));
|
LOGI("IN NATIVE SUGGEST Version: %d \n", (DICT[0] & 0xFF));
|
||||||
mUnigramDictionary = new UnigramDictionary(DICT, typedLetterMultiplier, fullWordMultiplier,
|
mUnigramDictionary = new UnigramDictionary(DICT, typedLetterMultiplier, fullWordMultiplier,
|
||||||
maxWordLength, maxWords, maxAlternatives, IS_LATEST_DICT_VERSION,
|
maxWordLength, maxWords, maxAlternatives, IS_LATEST_DICT_VERSION);
|
||||||
hasBigram(), this);
|
mBigramDictionary = new BigramDictionary(DICT, maxWordLength, maxAlternatives,
|
||||||
mBigramDictionary = new BigramDictionary(dict, typedLetterMultiplier, fullWordMultiplier,
|
IS_LATEST_DICT_VERSION, hasBigram(), this);
|
||||||
maxWordLength, maxWords, maxAlternatives, this);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
Dictionary::~Dictionary()
|
Dictionary::~Dictionary()
|
||||||
|
@ -37,7 +37,7 @@ public:
|
|||||||
int getBigrams(unsigned short *word, int length, int *codes, int codesSize,
|
int getBigrams(unsigned short *word, int length, int *codes, int codesSize,
|
||||||
unsigned short *outWords, int *frequencies, int maxWordLength, int maxBigrams,
|
unsigned short *outWords, int *frequencies, int maxWordLength, int maxBigrams,
|
||||||
int maxAlternatives) {
|
int maxAlternatives) {
|
||||||
return mUnigramDictionary->getBigrams(word, length, codes, codesSize, outWords, frequencies,
|
return mBigramDictionary->getBigrams(word, length, codes, codesSize, outWords, frequencies,
|
||||||
maxWordLength, maxBigrams, maxAlternatives);
|
maxWordLength, maxBigrams, maxAlternatives);
|
||||||
}
|
}
|
||||||
bool isValidWord(unsigned short *word, int length);
|
bool isValidWord(unsigned short *word, int length);
|
||||||
@ -53,6 +53,7 @@ public:
|
|||||||
static bool getTerminal(const unsigned char *dict, int *pos);
|
static bool getTerminal(const unsigned char *dict, int *pos);
|
||||||
static int getAddress(const unsigned char *dict, int *pos);
|
static int getAddress(const unsigned char *dict, int *pos);
|
||||||
static int getFreq(const unsigned char *dict, const bool isLatestDictVersion, int *pos);
|
static int getFreq(const unsigned char *dict, const bool isLatestDictVersion, int *pos);
|
||||||
|
static int wideStrLen(unsigned short *str);
|
||||||
|
|
||||||
private:
|
private:
|
||||||
bool hasBigram();
|
bool hasBigram();
|
||||||
@ -117,5 +118,14 @@ inline int Dictionary::getFreq(const unsigned char *dict,
|
|||||||
return freq;
|
return freq;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
||||||
|
inline int Dictionary::wideStrLen(unsigned short *str) {
|
||||||
|
if (!str) return 0;
|
||||||
|
unsigned short *end = str;
|
||||||
|
while (*end)
|
||||||
|
end++;
|
||||||
|
return end - str;
|
||||||
|
}
|
||||||
|
|
||||||
}; // namespace latinime
|
}; // namespace latinime
|
||||||
#endif // LATINIME_DICTIONARY_H
|
#endif // LATINIME_DICTIONARY_H
|
||||||
|
@ -31,23 +31,17 @@ namespace latinime {
|
|||||||
|
|
||||||
UnigramDictionary::UnigramDictionary(const unsigned char *dict, int typedLetterMultiplier,
|
UnigramDictionary::UnigramDictionary(const unsigned char *dict, int typedLetterMultiplier,
|
||||||
int fullWordMultiplier, int maxWordLength, int maxWords, int maxAlternatives,
|
int fullWordMultiplier, int maxWordLength, int maxWords, int maxAlternatives,
|
||||||
const bool isLatestDictVersion, const bool hasBigram, Dictionary *parentDictionary)
|
const bool isLatestDictVersion)
|
||||||
: DICT(dict), MAX_WORD_LENGTH(maxWordLength),MAX_WORDS(maxWords),
|
: DICT(dict), MAX_WORD_LENGTH(maxWordLength),MAX_WORDS(maxWords),
|
||||||
MAX_ALTERNATIVES(maxAlternatives), IS_LATEST_DICT_VERSION(isLatestDictVersion),
|
MAX_ALTERNATIVES(maxAlternatives), IS_LATEST_DICT_VERSION(isLatestDictVersion),
|
||||||
HAS_BIGRAM(hasBigram), mParentDictionary(parentDictionary)
|
TYPED_LETTER_MULTIPLIER(typedLetterMultiplier), FULL_WORD_MULTIPLIER(fullWordMultiplier) {
|
||||||
{
|
|
||||||
LOGI("UnigramDictionary - constructor");
|
LOGI("UnigramDictionary - constructor");
|
||||||
LOGI("Has Bigram : %d \n", hasBigram);
|
|
||||||
mTypedLetterMultiplier = typedLetterMultiplier;
|
|
||||||
mFullWordMultiplier = fullWordMultiplier;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
UnigramDictionary::~UnigramDictionary()
|
UnigramDictionary::~UnigramDictionary() {}
|
||||||
{
|
|
||||||
}
|
|
||||||
|
|
||||||
int UnigramDictionary::getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
|
int UnigramDictionary::getSuggestions(int *codes, int codesSize, unsigned short *outWords,
|
||||||
int *nextLetters, int nextLettersSize)
|
int *frequencies, int *nextLetters, int nextLettersSize)
|
||||||
{
|
{
|
||||||
|
|
||||||
initSuggestions(codes, codesSize, outWords, frequencies);
|
initSuggestions(codes, codesSize, outWords, frequencies);
|
||||||
@ -114,16 +108,6 @@ void UnigramDictionary::registerNextLetter(unsigned short c, int *nextLetters, i
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
int
|
|
||||||
UnigramDictionary::wideStrLen(unsigned short *str)
|
|
||||||
{
|
|
||||||
if (!str) return 0;
|
|
||||||
unsigned short *end = str;
|
|
||||||
while (*end)
|
|
||||||
end++;
|
|
||||||
return end - str;
|
|
||||||
}
|
|
||||||
|
|
||||||
bool
|
bool
|
||||||
UnigramDictionary::addWord(unsigned short *word, int length, int frequency)
|
UnigramDictionary::addWord(unsigned short *word, int length, int frequency)
|
||||||
{
|
{
|
||||||
@ -139,7 +123,7 @@ UnigramDictionary::addWord(unsigned short *word, int length, int frequency)
|
|||||||
while (insertAt < MAX_WORDS) {
|
while (insertAt < MAX_WORDS) {
|
||||||
if (frequency > mFrequencies[insertAt]
|
if (frequency > mFrequencies[insertAt]
|
||||||
|| (mFrequencies[insertAt] == frequency
|
|| (mFrequencies[insertAt] == frequency
|
||||||
&& length < wideStrLen(mOutputChars + insertAt * MAX_WORD_LENGTH))) {
|
&& length < Dictionary::wideStrLen(mOutputChars + insertAt * MAX_WORD_LENGTH))) {
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
insertAt++;
|
insertAt++;
|
||||||
@ -163,46 +147,6 @@ UnigramDictionary::addWord(unsigned short *word, int length, int frequency)
|
|||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
bool
|
|
||||||
UnigramDictionary::addWordBigram(unsigned short *word, int length, int frequency)
|
|
||||||
{
|
|
||||||
word[length] = 0;
|
|
||||||
if (DEBUG_DICT) {
|
|
||||||
char s[length + 1];
|
|
||||||
for (int i = 0; i <= length; i++) s[i] = word[i];
|
|
||||||
LOGI("Bigram: Found word = %s, freq = %d : \n", s, frequency);
|
|
||||||
}
|
|
||||||
|
|
||||||
// Find the right insertion point
|
|
||||||
int insertAt = 0;
|
|
||||||
while (insertAt < mMaxBigrams) {
|
|
||||||
if (frequency > mBigramFreq[insertAt]
|
|
||||||
|| (mBigramFreq[insertAt] == frequency
|
|
||||||
&& length < wideStrLen(mBigramChars + insertAt * MAX_WORD_LENGTH))) {
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
insertAt++;
|
|
||||||
}
|
|
||||||
LOGI("Bigram: InsertAt -> %d maxBigrams: %d\n", insertAt, mMaxBigrams);
|
|
||||||
if (insertAt < mMaxBigrams) {
|
|
||||||
memmove((char*) mBigramFreq + (insertAt + 1) * sizeof(mBigramFreq[0]),
|
|
||||||
(char*) mBigramFreq + insertAt * sizeof(mBigramFreq[0]),
|
|
||||||
(mMaxBigrams - insertAt - 1) * sizeof(mBigramFreq[0]));
|
|
||||||
mBigramFreq[insertAt] = frequency;
|
|
||||||
memmove((char*) mBigramChars + (insertAt + 1) * MAX_WORD_LENGTH * sizeof(short),
|
|
||||||
(char*) mBigramChars + (insertAt ) * MAX_WORD_LENGTH * sizeof(short),
|
|
||||||
(mMaxBigrams - insertAt - 1) * sizeof(short) * MAX_WORD_LENGTH);
|
|
||||||
unsigned short *dest = mBigramChars + (insertAt ) * MAX_WORD_LENGTH;
|
|
||||||
while (length--) {
|
|
||||||
*dest++ = *word++;
|
|
||||||
}
|
|
||||||
*dest = 0; // NULL terminate
|
|
||||||
if (DEBUG_DICT) LOGI("Bigram: Added word at %d\n", insertAt);
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
return false;
|
|
||||||
}
|
|
||||||
|
|
||||||
unsigned short
|
unsigned short
|
||||||
UnigramDictionary::toLowerCase(unsigned short c) {
|
UnigramDictionary::toLowerCase(unsigned short c) {
|
||||||
if (c < sizeof(BASE_CHARS) / sizeof(BASE_CHARS[0])) {
|
if (c < sizeof(BASE_CHARS) / sizeof(BASE_CHARS[0])) {
|
||||||
@ -236,8 +180,8 @@ UnigramDictionary::sameAsTyped(unsigned short *word, int length)
|
|||||||
static char QUOTE = '\'';
|
static char QUOTE = '\'';
|
||||||
|
|
||||||
void
|
void
|
||||||
UnigramDictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int snr, int inputIndex,
|
UnigramDictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion, int snr,
|
||||||
int diffs, int skipPos, int *nextLetters, int nextLettersSize)
|
int inputIndex, int diffs, int skipPos, int *nextLetters, int nextLettersSize)
|
||||||
{
|
{
|
||||||
// Optimization: Prune out words that are too long compared to how much was typed.
|
// Optimization: Prune out words that are too long compared to how much was typed.
|
||||||
if (depth > maxDepth) {
|
if (depth > maxDepth) {
|
||||||
@ -290,14 +234,14 @@ UnigramDictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion
|
|||||||
int j = 0;
|
int j = 0;
|
||||||
while (currentChars[j] > 0) {
|
while (currentChars[j] > 0) {
|
||||||
if (currentChars[j] == lowerC || currentChars[j] == c) {
|
if (currentChars[j] == lowerC || currentChars[j] == c) {
|
||||||
int addedWeight = j == 0 ? mTypedLetterMultiplier : 1;
|
int addedWeight = j == 0 ? TYPED_LETTER_MULTIPLIER : 1;
|
||||||
mWord[depth] = c;
|
mWord[depth] = c;
|
||||||
if (mInputLength == inputIndex + 1) {
|
if (mInputLength == inputIndex + 1) {
|
||||||
if (terminal) {
|
if (terminal) {
|
||||||
if (//INCLUDE_TYPED_WORD_IF_VALID ||
|
if (//INCLUDE_TYPED_WORD_IF_VALID ||
|
||||||
!sameAsTyped(mWord, depth + 1)) {
|
!sameAsTyped(mWord, depth + 1)) {
|
||||||
int finalFreq = freq * snr * addedWeight;
|
int finalFreq = freq * snr * addedWeight;
|
||||||
if (skipPos < 0) finalFreq *= mFullWordMultiplier;
|
if (skipPos < 0) finalFreq *= FULL_WORD_MULTIPLIER;
|
||||||
addWord(mWord, depth + 1, finalFreq);
|
addWord(mWord, depth + 1, finalFreq);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
@ -319,190 +263,4 @@ UnigramDictionary::getWordsRec(int pos, int depth, int maxDepth, bool completion
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
int
|
|
||||||
UnigramDictionary::getBigramAddress(int *pos, bool advance)
|
|
||||||
{
|
|
||||||
int address = 0;
|
|
||||||
|
|
||||||
address += (DICT[*pos] & 0x3F) << 16;
|
|
||||||
address += (DICT[*pos + 1] & 0xFF) << 8;
|
|
||||||
address += (DICT[*pos + 2] & 0xFF);
|
|
||||||
|
|
||||||
if (advance) {
|
|
||||||
*pos += 3;
|
|
||||||
}
|
|
||||||
|
|
||||||
return address;
|
|
||||||
}
|
|
||||||
|
|
||||||
int
|
|
||||||
UnigramDictionary::getBigramFreq(int *pos)
|
|
||||||
{
|
|
||||||
int freq = DICT[(*pos)++] & FLAG_BIGRAM_FREQ;
|
|
||||||
|
|
||||||
return freq;
|
|
||||||
}
|
|
||||||
|
|
||||||
|
|
||||||
int
|
|
||||||
UnigramDictionary::getBigrams(unsigned short *prevWord, int prevWordLength, int *codes, int codesSize,
|
|
||||||
unsigned short *bigramChars, int *bigramFreq, int maxWordLength, int maxBigrams,
|
|
||||||
int maxAlternatives)
|
|
||||||
{
|
|
||||||
mBigramFreq = bigramFreq;
|
|
||||||
mBigramChars = bigramChars;
|
|
||||||
mInputCodes = codes;
|
|
||||||
mInputLength = codesSize;
|
|
||||||
mMaxBigrams = maxBigrams;
|
|
||||||
|
|
||||||
if (HAS_BIGRAM && IS_LATEST_DICT_VERSION) {
|
|
||||||
int pos = mParentDictionary->isValidWordRec(
|
|
||||||
DICTIONARY_HEADER_SIZE, prevWord, 0, prevWordLength);
|
|
||||||
LOGI("Pos -> %d\n", pos);
|
|
||||||
if (pos < 0) {
|
|
||||||
return 0;
|
|
||||||
}
|
|
||||||
|
|
||||||
int bigramCount = 0;
|
|
||||||
int bigramExist = (DICT[pos] & FLAG_BIGRAM_READ);
|
|
||||||
if (bigramExist > 0) {
|
|
||||||
int nextBigramExist = 1;
|
|
||||||
while (nextBigramExist > 0 && bigramCount < maxBigrams) {
|
|
||||||
int bigramAddress = getBigramAddress(&pos, true);
|
|
||||||
int frequency = (FLAG_BIGRAM_FREQ & DICT[pos]);
|
|
||||||
// search for all bigrams and store them
|
|
||||||
searchForTerminalNode(bigramAddress, frequency);
|
|
||||||
nextBigramExist = (DICT[pos++] & FLAG_BIGRAM_CONTINUED);
|
|
||||||
bigramCount++;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
return bigramCount;
|
|
||||||
}
|
|
||||||
return 0;
|
|
||||||
}
|
|
||||||
|
|
||||||
void
|
|
||||||
UnigramDictionary::searchForTerminalNode(int addressLookingFor, int frequency)
|
|
||||||
{
|
|
||||||
// track word with such address and store it in an array
|
|
||||||
unsigned short word[MAX_WORD_LENGTH];
|
|
||||||
|
|
||||||
int pos;
|
|
||||||
int followDownBranchAddress = DICTIONARY_HEADER_SIZE;
|
|
||||||
bool found = false;
|
|
||||||
char followingChar = ' ';
|
|
||||||
int depth = -1;
|
|
||||||
|
|
||||||
while(!found) {
|
|
||||||
bool followDownAddressSearchStop = false;
|
|
||||||
bool firstAddress = true;
|
|
||||||
bool haveToSearchAll = true;
|
|
||||||
|
|
||||||
if (depth >= 0) {
|
|
||||||
word[depth] = (unsigned short) followingChar;
|
|
||||||
}
|
|
||||||
pos = followDownBranchAddress; // pos start at count
|
|
||||||
int count = DICT[pos] & 0xFF;
|
|
||||||
LOGI("count - %d\n",count);
|
|
||||||
pos++;
|
|
||||||
for (int i = 0; i < count; i++) {
|
|
||||||
// pos at data
|
|
||||||
pos++;
|
|
||||||
// pos now at flag
|
|
||||||
if (!getFirstBitOfByte(&pos)) { // non-terminal
|
|
||||||
if (!followDownAddressSearchStop) {
|
|
||||||
int addr = getBigramAddress(&pos, false);
|
|
||||||
if (addr > addressLookingFor) {
|
|
||||||
followDownAddressSearchStop = true;
|
|
||||||
if (firstAddress) {
|
|
||||||
firstAddress = false;
|
|
||||||
haveToSearchAll = true;
|
|
||||||
} else if (!haveToSearchAll) {
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
} else {
|
|
||||||
followDownBranchAddress = addr;
|
|
||||||
followingChar = (char)(0xFF & DICT[pos-1]);
|
|
||||||
if (firstAddress) {
|
|
||||||
firstAddress = false;
|
|
||||||
haveToSearchAll = false;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
pos += 3;
|
|
||||||
} else if (getFirstBitOfByte(&pos)) { // terminal
|
|
||||||
if (addressLookingFor == (pos-1)) { // found !!
|
|
||||||
depth++;
|
|
||||||
word[depth] = (0xFF & DICT[pos-1]);
|
|
||||||
found = true;
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
if (getSecondBitOfByte(&pos)) { // address + freq (4 byte)
|
|
||||||
if (!followDownAddressSearchStop) {
|
|
||||||
int addr = getBigramAddress(&pos, false);
|
|
||||||
if (addr > addressLookingFor) {
|
|
||||||
followDownAddressSearchStop = true;
|
|
||||||
if (firstAddress) {
|
|
||||||
firstAddress = false;
|
|
||||||
haveToSearchAll = true;
|
|
||||||
} else if (!haveToSearchAll) {
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
} else {
|
|
||||||
followDownBranchAddress = addr;
|
|
||||||
followingChar = (char)(0xFF & DICT[pos-1]);
|
|
||||||
if (firstAddress) {
|
|
||||||
firstAddress = false;
|
|
||||||
haveToSearchAll = true;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
pos += 4;
|
|
||||||
} else { // freq only (2 byte)
|
|
||||||
pos += 2;
|
|
||||||
}
|
|
||||||
|
|
||||||
// skipping bigram
|
|
||||||
int bigramExist = (DICT[pos] & FLAG_BIGRAM_READ);
|
|
||||||
if (bigramExist > 0) {
|
|
||||||
int nextBigramExist = 1;
|
|
||||||
while (nextBigramExist > 0) {
|
|
||||||
pos += 3;
|
|
||||||
nextBigramExist = (DICT[pos++] & FLAG_BIGRAM_CONTINUED);
|
|
||||||
}
|
|
||||||
} else {
|
|
||||||
pos++;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
depth++;
|
|
||||||
if (followDownBranchAddress == 0) {
|
|
||||||
LOGI("ERROR!!! Cannot find bigram!!");
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
if (checkFirstCharacter(word)) {
|
|
||||||
addWordBigram(word, depth, frequency);
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
bool
|
|
||||||
UnigramDictionary::checkFirstCharacter(unsigned short *word)
|
|
||||||
{
|
|
||||||
// Checks whether this word starts with same character or neighboring characters of
|
|
||||||
// what user typed.
|
|
||||||
|
|
||||||
int *inputCodes = mInputCodes;
|
|
||||||
int maxAlt = MAX_ALTERNATIVES;
|
|
||||||
while (maxAlt > 0) {
|
|
||||||
if ((unsigned int) *inputCodes == (unsigned int) *word) {
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
inputCodes++;
|
|
||||||
maxAlt--;
|
|
||||||
}
|
|
||||||
return false;
|
|
||||||
}
|
|
||||||
|
|
||||||
} // namespace latinime
|
} // namespace latinime
|
||||||
|
@ -24,63 +24,42 @@ namespace latinime {
|
|||||||
class UnigramDictionary {
|
class UnigramDictionary {
|
||||||
public:
|
public:
|
||||||
UnigramDictionary(const unsigned char *dict, int typedLetterMultipler, int fullWordMultiplier,
|
UnigramDictionary(const unsigned char *dict, int typedLetterMultipler, int fullWordMultiplier,
|
||||||
int maxWordLength, int maxWords, int maxAlternatives, const bool isLatestDictVersion,
|
int maxWordLength, int maxWords, int maxAlternatives, const bool isLatestDictVersion);
|
||||||
const bool hasBigram, Dictionary *parentDictionary);
|
|
||||||
int getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
|
int getSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies,
|
||||||
int *nextLetters, int nextLettersSize);
|
int *nextLetters, int nextLettersSize);
|
||||||
int getBigrams(unsigned short *word, int length, int *codes, int codesSize,
|
|
||||||
unsigned short *outWords, int *frequencies, int maxWordLength, int maxBigrams,
|
|
||||||
int maxAlternatives);
|
|
||||||
~UnigramDictionary();
|
~UnigramDictionary();
|
||||||
|
|
||||||
private:
|
private:
|
||||||
void initSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies);
|
void initSuggestions(int *codes, int codesSize, unsigned short *outWords, int *frequencies);
|
||||||
int getSuggestionCandidates(int inputLength, int skipPos, int *nextLetters, int nextLettersSize);
|
int getSuggestionCandidates(int inputLength, int skipPos, int *nextLetters,
|
||||||
|
int nextLettersSize);
|
||||||
void getVersionNumber();
|
void getVersionNumber();
|
||||||
bool checkIfDictVersionIsLatest();
|
bool checkIfDictVersionIsLatest();
|
||||||
int getAddress(int *pos);
|
int getAddress(int *pos);
|
||||||
int getBigramAddress(int *pos, bool advance);
|
|
||||||
int getFreq(int *pos);
|
int getFreq(int *pos);
|
||||||
int getBigramFreq(int *pos);
|
|
||||||
void searchForTerminalNode(int address, int frequency);
|
|
||||||
|
|
||||||
bool getFirstBitOfByte(int *pos) { return (DICT[*pos] & 0x80) > 0; }
|
|
||||||
bool getSecondBitOfByte(int *pos) { return (DICT[*pos] & 0x40) > 0; }
|
|
||||||
bool getTerminal(int *pos) { return (DICT[*pos] & FLAG_TERMINAL_MASK) > 0; }
|
|
||||||
int wideStrLen(unsigned short *str);
|
int wideStrLen(unsigned short *str);
|
||||||
|
|
||||||
bool sameAsTyped(unsigned short *word, int length);
|
bool sameAsTyped(unsigned short *word, int length);
|
||||||
bool checkFirstCharacter(unsigned short *word);
|
|
||||||
bool addWord(unsigned short *word, int length, int frequency);
|
bool addWord(unsigned short *word, int length, int frequency);
|
||||||
bool addWordBigram(unsigned short *word, int length, int frequency);
|
|
||||||
unsigned short toLowerCase(unsigned short c);
|
unsigned short toLowerCase(unsigned short c);
|
||||||
void getWordsRec(int pos, int depth, int maxDepth, bool completion, int frequency,
|
void getWordsRec(int pos, int depth, int maxDepth, bool completion, int frequency,
|
||||||
int inputIndex, int diffs, int skipPos, int *nextLetters, int nextLettersSize);
|
int inputIndex, int diffs, int skipPos, int *nextLetters, int nextLettersSize);
|
||||||
void registerNextLetter(unsigned short c, int *nextLetters, int nextLettersSize);
|
void registerNextLetter(unsigned short c, int *nextLetters, int nextLettersSize);
|
||||||
|
|
||||||
const unsigned char *DICT;
|
const unsigned char *DICT;
|
||||||
|
|
||||||
const int MAX_WORDS;
|
const int MAX_WORDS;
|
||||||
const int MAX_WORD_LENGTH;
|
const int MAX_WORD_LENGTH;
|
||||||
const int MAX_ALTERNATIVES;
|
const int MAX_ALTERNATIVES;
|
||||||
const bool IS_LATEST_DICT_VERSION;
|
const bool IS_LATEST_DICT_VERSION;
|
||||||
const bool HAS_BIGRAM;
|
const int TYPED_LETTER_MULTIPLIER;
|
||||||
|
const int FULL_WORD_MULTIPLIER;
|
||||||
|
|
||||||
Dictionary *mParentDictionary;
|
|
||||||
int *mFrequencies;
|
int *mFrequencies;
|
||||||
int *mBigramFreq;
|
|
||||||
int mMaxBigrams;
|
|
||||||
unsigned short *mOutputChars;
|
unsigned short *mOutputChars;
|
||||||
unsigned short *mBigramChars;
|
|
||||||
int *mInputCodes;
|
int *mInputCodes;
|
||||||
int mInputLength;
|
int mInputLength;
|
||||||
unsigned short mWord[128];
|
unsigned short mWord[128];
|
||||||
int mMaxEditDistance;
|
int mMaxEditDistance;
|
||||||
|
|
||||||
int mFullWordMultiplier;
|
|
||||||
int mTypedLetterMultiplier;
|
|
||||||
int mVersion;
|
|
||||||
int mBigram;
|
|
||||||
};
|
};
|
||||||
|
|
||||||
// ----------------------------------------------------------------------------
|
// ----------------------------------------------------------------------------
|
||||||
|
Loading…
x
Reference in New Issue
Block a user