/* find the maximum frequency of the same phrase */
static int LoadMaxFreq( const uint16 phoneSeq[], int len )
{
int pho_id;
Phrase *phrase = ALC( Phrase, 1 );
int maxFreq = FREQ_INIT_VALUE;
UserPhraseData *uphrase;
pho_id = TreeFindPhrase( 0, len - 1, phoneSeq );
if ( pho_id != -1 ) {
GetPhraseFirst( phrase, pho_id );
do {
if ( phrase->freq > maxFreq )
maxFreq = phrase->freq;
} while( GetPhraseNext( phrase ) );
}
free( phrase );
uphrase = UserGetPhraseFirst( phoneSeq );
while ( uphrase ) {
if ( uphrase->userfreq > maxFreq )
maxFreq = uphrase->userfreq;
uphrase = UserGetPhraseNext( phoneSeq );
}
return maxFreq;
}
/* find the maximum frequency of the same phrase */
static int LoadMaxFreq( ChewingData *pgdata, const uint16_t phoneSeq[], int len )
{
const TreeType *tree_pos;
Phrase *phrase = ALC( Phrase, 1 );
int maxFreq = FREQ_INIT_VALUE;
UserPhraseData *uphrase;
tree_pos = TreeFindPhrase( pgdata, 0, len - 1, phoneSeq );
if ( tree_pos ) {
GetPhraseFirst( pgdata, phrase, tree_pos );
do {
if ( phrase->freq > maxFreq )
maxFreq = phrase->freq;
} while( GetVocabNext( pgdata, phrase ) );
}
free( phrase );
uphrase = UserGetPhraseFirst( pgdata, phoneSeq );
while ( uphrase ) {
if ( uphrase->userfreq > maxFreq )
maxFreq = uphrase->userfreq;
uphrase = UserGetPhraseNext( pgdata, phoneSeq );
}
return maxFreq;
}
/* find the maximum frequency of the same phrase */
static int LoadMaxFreq(ChewingData *pgdata, const uint16_t phoneSeq[], int len)
{
const TreeType *tree_pos;
Phrase *phrase = ALC(Phrase, 1);
int maxFreq = FREQ_INIT_VALUE;
int max_userphrase_freq;
int ret;
tree_pos = TreeFindPhrase(pgdata, 0, len - 1, phoneSeq);
if (tree_pos) {
GetPhraseFirst(pgdata, phrase, tree_pos);
do {
if (phrase->freq > maxFreq)
maxFreq = phrase->freq;
} while (GetVocabNext(pgdata, phrase));
}
free(phrase);
assert(pgdata->static_data.stmt_userphrase[STMT_USERPHRASE_GET_MAX_FREQ]);
ret = UserBindPhone(pgdata, STMT_USERPHRASE_GET_MAX_FREQ, phoneSeq, len);
if (ret != SQLITE_OK) {
LOG_ERROR("UserBindPhone returns %d", ret);
return maxFreq;
}
ret = sqlite3_step(pgdata->static_data.stmt_userphrase[STMT_USERPHRASE_GET_MAX_FREQ]);
if (ret != SQLITE_ROW)
return maxFreq;
ret = sqlite3_reset(pgdata->static_data.stmt_userphrase[STMT_USERPHRASE_GET_MAX_FREQ]);
if (ret != SQLITE_OK) {
LOG_ERROR("sqlite3_reset returns %d", ret);
return maxFreq;
}
max_userphrase_freq = sqlite3_column_int(pgdata->static_data.stmt_userphrase[STMT_USERPHRASE_GET_MAX_FREQ],
SQL_STMT_USERPHRASE[STMT_USERPHRASE_GET_MAX_FREQ].column
[COLUMN_USERPHRASE_USER_FREQ]);
if (max_userphrase_freq > maxFreq)
maxFreq = max_userphrase_freq;
return maxFreq;
}
/** @brief Loading all possible phrases after the cursor from long to short into AvailInfo structure.*/
static void SetAvailInfo(
AvailInfo *pai, const uint16 phoneSeq[],
int nPhoneSeq, int begin, const int bSymbolArrBrkpt[] )
{
int end, pho_id;
int diff;
#ifndef FAFT_CHEWING
uint16 userPhoneSeq[ MAX_PHONE_SEQ_LEN ];
#endif
pai->nAvail = 0;
for ( end = begin; end < nPhoneSeq; end++ ) {
diff = end - begin;
if ( diff > 0 && bSymbolArrBrkpt[ end ] )
break;
pho_id = TreeFindPhrase( begin, end, phoneSeq );
if ( pho_id != -1 ) {
/* save it! */
pai->avail[ pai->nAvail ].len = diff + 1;
pai->avail[ pai->nAvail ].id = pho_id;
pai->nAvail++;
}
#ifndef FAFT_CHEWING
else {
memcpy(
userPhoneSeq,
&phoneSeq[ begin ],
sizeof( uint16 ) * ( diff + 1 ) ) ;
userPhoneSeq[ diff + 1 ] = 0;
if ( UserGetPhraseFirst( userPhoneSeq ) ) {
/* save it! */
pai->avail[ pai->nAvail ].len = diff + 1;
pai->avail[ pai->nAvail ].id = -1;
pai->nAvail++;
} else {
pai->avail[ pai->nAvail ].len = 0;
pai->avail[ pai->nAvail ].id = -1;
}
}
#endif
}
}
/* load the orginal frequency from the static dict */
static int LoadOriginalFreq(ChewingData *pgdata, const uint16_t phoneSeq[], const char wordSeq[], int len)
{
const TreeType *tree_pos;
int retval;
Phrase *phrase = ALC(Phrase, 1);
tree_pos = TreeFindPhrase(pgdata, 0, len - 1, phoneSeq);
if (tree_pos) {
GetPhraseFirst(pgdata, phrase, tree_pos);
do {
/* find the same phrase */
if (!strcmp(phrase->phrase, wordSeq)) {
retval = phrase->freq;
free(phrase);
return retval;
}
} while (GetVocabNext(pgdata, phrase));
}
free(phrase);
return FREQ_INIT_VALUE;
}
/* load the orginal frequency from the static dict */
static int LoadOriginalFreq( const uint16 phoneSeq[], const char wordSeq[], int len )
{
int pho_id;
int retval;
Phrase *phrase = ALC( Phrase, 1 );
pho_id = TreeFindPhrase( 0, len - 1, phoneSeq );
if ( pho_id != -1 ) {
GetPhraseFirst( phrase, pho_id );
do {
/* find the same phrase */
if ( ! strcmp(
phrase->phrase,
wordSeq ) ) {
retval = phrase->freq;
free( phrase );
return retval;
}
} while ( GetPhraseNext( phrase ) );
}
free( phrase );
return FREQ_INIT_VALUE;
}
static void FindInterval( ChewingData *pgdata, TreeDataType *ptd )
{
int end, begin, pho_id;
Phrase *p_phrase, *puserphrase, *pdictphrase;
UsedPhraseMode i_used_phrase;
uint16_t new_phoneSeq[ MAX_PHONE_SEQ_LEN ];
for ( begin = 0; begin < pgdata->nPhoneSeq; begin++ ) {
for ( end = begin; end < pgdata->nPhoneSeq; end++ ) {
if ( ! CheckBreakpoint( begin, end + 1, pgdata->bArrBrkpt ) )
continue;
/* set new_phoneSeq */
memcpy(
new_phoneSeq,
&pgdata->phoneSeq[ begin ],
sizeof( uint16_t ) * ( end - begin + 1 ) );
new_phoneSeq[ end - begin + 1 ] = 0;
puserphrase = pdictphrase = NULL;
i_used_phrase = USED_PHRASE_NONE;
/* check user phrase */
if ( UserGetPhraseFirst( pgdata, new_phoneSeq ) &&
CheckUserChoose( pgdata, new_phoneSeq, begin, end + 1,
&p_phrase, pgdata->selectStr, pgdata->selectInterval, pgdata->nSelect ) ) {
puserphrase = p_phrase;
}
/* check dict phrase */
pho_id = TreeFindPhrase( pgdata, begin, end, pgdata->phoneSeq );
if (
( pho_id != -1 ) &&
CheckChoose(
pgdata,
pho_id, begin, end + 1,
&p_phrase, pgdata->selectStr,
pgdata->selectInterval, pgdata->nSelect ) ) {
pdictphrase = p_phrase;
}
/* add only one interval, which has the largest freqency
* but when the phrase is the same, the user phrase overrides
* static dict
*/
if ( puserphrase != NULL && pdictphrase == NULL ) {
i_used_phrase = USED_PHRASE_USER;
}
else if ( puserphrase == NULL && pdictphrase != NULL ) {
i_used_phrase = USED_PHRASE_DICT;
}
else if ( puserphrase != NULL && pdictphrase != NULL ) {
/* the same phrase, userphrase overrides */
if ( ! strcmp(
puserphrase->phrase,
pdictphrase->phrase ) ) {
i_used_phrase = USED_PHRASE_USER;
}
else {
if ( puserphrase->freq > pdictphrase->freq ) {
i_used_phrase = USED_PHRASE_USER;
}
else {
i_used_phrase = USED_PHRASE_DICT;
}
}
}
switch ( i_used_phrase ) {
case USED_PHRASE_USER:
AddInterval( ptd, begin, end, -1, puserphrase,
IS_USER_PHRASE );
break;
case USED_PHRASE_DICT:
AddInterval( ptd, begin, end, pho_id, pdictphrase,
IS_DICT_PHRASE );
break;
case USED_PHRASE_NONE:
default:
break;
}
internal_release_Phrase(
i_used_phrase,
puserphrase,
pdictphrase );
}
}
}
/** @brief Loading all possible phrases after the cursor from long to short into AvailInfo structure.*/
static void SetAvailInfo(ChewingData *pgdata, int begin, int end)
{
AvailInfo *pai = &(pgdata->availInfo);
const uint16_t *phoneSeq = pgdata->phoneSeq;
int nPhoneSeq = pgdata->nPhoneSeq;
const int *bSymbolArrBrkpt = pgdata->bSymbolArrBrkpt;
int symbolArrBrkpt[ARRAY_SIZE(pgdata->bSymbolArrBrkpt)] = { 0 };
const TreeType *tree_pos;
int diff;
uint16_t userPhoneSeq[MAX_PHONE_SEQ_LEN];
int i, head, head_tmp;
int tail, tail_tmp;
int pos;
head = tail = 0;
pai->nAvail = 0;
/*
* XXX: The phoneSeq, nPhoneSeq skip any symbol in preedit buffer,
* while bSymbolArrBrkpt, does not skip any symbol in preedit
* buffer. So we need to do some translate here.
*/
for (i = 0; i < pgdata->chiSymbolBufLen; ++i) {
if (bSymbolArrBrkpt[i]) {
/*
* XXX: If preedit buffer starts with symbol, the pos
* will become negative. In this case, we just ignore
* this symbol because it does not create any break
* point.
*/
pos = i - CountSymbols(pgdata, i + 1);
if (pos >= 0)
symbolArrBrkpt[pos] = 1;
}
}
if (pgdata->config.bPhraseChoiceRearward) {
for (i = end; i >= begin; i--) {
if (symbolArrBrkpt[i])
break;
head = i;
}
head_tmp = end;
} else {
head_tmp = head = begin;
}
if (pgdata->config.bPhraseChoiceRearward) {
tail_tmp = tail = end;
} else {
for (i = begin; i < nPhoneSeq; i++) {
tail = i;
if (symbolArrBrkpt[i])
break;
}
tail_tmp = begin;
}
while (head <= head_tmp && tail_tmp <= tail) {
diff = tail_tmp - head_tmp;
tree_pos = TreeFindPhrase(pgdata, head_tmp, tail_tmp, phoneSeq);
if (tree_pos) {
/* save it! */
pai->avail[pai->nAvail].len = diff + 1;
pai->avail[pai->nAvail].id = tree_pos;
pai->nAvail++;
} else {
memcpy(userPhoneSeq, &phoneSeq[head_tmp], sizeof(uint16_t) * (diff + 1));
userPhoneSeq[diff + 1] = 0;
if (UserGetPhraseFirst(pgdata, userPhoneSeq)) {
/* save it! */
pai->avail[pai->nAvail].len = diff + 1;
pai->avail[pai->nAvail].id = NULL;
pai->nAvail++;
} else {
pai->avail[pai->nAvail].len = 0;
pai->avail[pai->nAvail].id = NULL;
}
UserGetPhraseEnd(pgdata, userPhoneSeq);
}
if (pgdata->config.bPhraseChoiceRearward) {
head_tmp--;
} else {
tail_tmp++;
}
}
}
/** @brief Loading all possible phrases after the cursor from long to short into AvailInfo structure.*/
static void SetAvailInfo( ChewingData *pgdata, int begin, int end)
{
AvailInfo *pai = &( pgdata->availInfo );
const uint16_t *phoneSeq = pgdata->phoneSeq;
int nPhoneSeq = pgdata->nPhoneSeq;
const int *bSymbolArrBrkpt = pgdata->bSymbolArrBrkpt;
int pho_id;
int diff;
uint16_t userPhoneSeq[ MAX_PHONE_SEQ_LEN ];
int i, head, head_tmp;
int tail, tail_tmp;
head = tail = 0;
pai->nAvail = 0;
if ( pgdata->config.bPhraseChoiceRearward ) {
for ( i = end; i >= begin; i--){
head = i;
if ( bSymbolArrBrkpt[ i ] )
break;
}
head_tmp = end;
} else {
head_tmp = head = begin;
}
if ( pgdata->config.bPhraseChoiceRearward ) {
tail_tmp = tail = end;
} else {
for ( i = begin; i < nPhoneSeq; i++ ) {
if ( bSymbolArrBrkpt[ i ] )
break;
tail = i;
}
tail_tmp = begin;
}
while ( head <= head_tmp && tail_tmp <= tail ) {
diff = tail_tmp - head_tmp;
pho_id = TreeFindPhrase( pgdata, head_tmp, tail_tmp, phoneSeq );
if ( pho_id != -1 ) {
/* save it! */
pai->avail[ pai->nAvail ].len = diff + 1;
pai->avail[ pai->nAvail ].id = pho_id;
pai->nAvail++;
}
else {
memcpy(
userPhoneSeq,
&phoneSeq[ head_tmp ],
sizeof( uint16_t ) * ( diff + 1 ) ) ;
userPhoneSeq[ diff + 1 ] = 0;
if ( UserGetPhraseFirst( pgdata, userPhoneSeq ) ) {
/* save it! */
pai->avail[ pai->nAvail ].len = diff + 1;
pai->avail[ pai->nAvail ].id = -1;
pai->nAvail++;
} else {
pai->avail[ pai->nAvail ].len = 0;
pai->avail[ pai->nAvail ].id = -1;
}
}
if ( pgdata->config.bPhraseChoiceRearward ) {
head_tmp--;
} else {
tail_tmp++;
}
}
}
