/* 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( 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;
}
/** @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
}
}
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 );
}
}
}
static int CheckUserChoose(
ChewingData *pgdata,
uint16_t *new_phoneSeq, int from , int to,
Phrase **pp_phr,
char selectStr[][ MAX_PHONE_SEQ_LEN * MAX_UTF8_SIZE + 1 ],
IntervalType selectInterval[], int nSelect )
{
IntervalType inte, c;
int chno, len;
int user_alloc;
UserPhraseData *pUserPhraseData;
Phrase *p_phr = ALC( Phrase, 1 );
assert( p_phr );
inte.from = from;
inte.to = to;
*pp_phr = NULL;
/* pass 1
* if these exist one selected interval which is not contained by inte
* but has intersection with inte, then inte is an unacceptable interval
*/
for ( chno = 0; chno < nSelect; chno++ ) {
c = selectInterval[ chno ];
if ( IsIntersect( inte, c ) && ! IsContain( inte, c ) ) {
free( p_phr );
return 0;
}
}
/* pass 2
* if there exist one phrase satisfied all selectStr then return 1, else return 0.
* also store the phrase with highest freq
*/
pUserPhraseData = UserGetPhraseFirst( pgdata, new_phoneSeq );
p_phr->freq = -1;
do {
for ( chno = 0; chno < nSelect; chno++ ) {
c = selectInterval[ chno ];
if ( IsContain( inte, c ) ) {
/*
* find a phrase of ph_id where the text contains
* 'selectStr[chno]' test if not ok then return 0,
* if ok then continue to test. */
len = c.to - c.from;
if ( memcmp(
ueStrSeek( pUserPhraseData->wordSeq, c.from - from ),
selectStr[ chno ],
ueStrNBytes( selectStr[ chno ], len ) ) )
break;
}
}
if ( chno == nSelect ) {
/* save phrase data to "pp_phr" */
if ( pUserPhraseData->userfreq > p_phr->freq ) {
if ( ( user_alloc = ( to - from ) ) > 0 ) {
ueStrNCpy( p_phr->phrase,
pUserPhraseData->wordSeq,
user_alloc, 1);
}
p_phr->freq = pUserPhraseData->userfreq;
*pp_phr = p_phr;
}
}
} while ( ( pUserPhraseData = UserGetPhraseNext( pgdata, new_phoneSeq ) ) != NULL );
if ( p_phr->freq != -1 )
return 1;
free( p_phr );
return 0;
}
/** @brief Loading all possible phrases of certain length.
*
* Loading all possible phrases of certain length into ChoiceInfo structure from static
* and dynamic dictionaries,\n
* including number of total pages and the number of current page.\n
*/
static void SetChoiceInfo(
ChoiceInfo *pci,AvailInfo *pai, uint16 *phoneSeq, int cursor,
int candPerPage )
{
Word tempWord;
Phrase tempPhrase;
int len;
UserPhraseData *pUserPhraseData;
uint16 userPhoneSeq[ MAX_PHONE_SEQ_LEN ];
/* Clears previous candidates. */
memset( pci->totalChoiceStr, '\0', sizeof(char) * MAX_CHOICE * MAX_PHRASE_LEN * MAX_UTF8_SIZE + 1);
pci->nTotalChoice = 0;
len = pai->avail[ pai->currentAvail ].len;
assert(len);
/* secondly, read tree phrase */
if ( len == 1 ) { /* single character */
GetCharFirst( &tempWord, phoneSeq[ cursor ] );
do {
if ( ChoiceTheSame( pci, tempWord.word, ueBytesFromChar( tempWord.word[0] ) * sizeof( char ) ) )
continue;
memcpy(
pci->totalChoiceStr[ pci->nTotalChoice ],
tempWord.word, ueBytesFromChar( tempWord.word[0] ) * sizeof( char ) );
assert(pci->nTotalChoice <= MAX_CHOICE);
pci->totalChoiceStr[ pci->nTotalChoice ][ ueBytesFromChar( tempWord.word[0] ) ] = '\0';
pci->nTotalChoice++;
} while( GetCharNext( &tempWord ) );
}
/* phrase */
else {
if ( pai->avail[ pai->currentAvail ].id != -1 ) {
GetPhraseFirst( &tempPhrase, pai->avail[ pai->currentAvail ].id );
do {
if ( ChoiceTheSame(
pci,
tempPhrase.phrase,
len * ueBytesFromChar( tempPhrase.phrase[0] ) * sizeof( char ) ) ) {
continue;
}
ueStrNCpy( pci->totalChoiceStr[ pci->nTotalChoice ],
tempPhrase.phrase, len, 1);
pci->nTotalChoice++;
} while( GetPhraseNext( &tempPhrase ) );
}
memcpy( userPhoneSeq, &phoneSeq[ cursor ], sizeof( uint16 ) * len );
userPhoneSeq[ len ] = 0;
pUserPhraseData = UserGetPhraseFirst( userPhoneSeq );
if ( pUserPhraseData ) {
do {
/* check if the phrase is already in the choice list */
if ( ChoiceTheSame(
pci,
pUserPhraseData->wordSeq,
len * ueBytesFromChar( pUserPhraseData->wordSeq[0] ) * sizeof( char ) ) )
continue;
/* otherwise store it */
ueStrNCpy(
pci->totalChoiceStr[ pci->nTotalChoice ],
pUserPhraseData->wordSeq,
len, 1);
pci->nTotalChoice++;
} while( ( pUserPhraseData =
UserGetPhraseNext( userPhoneSeq ) ) != NULL );
}
}
/* magic number */
pci->nChoicePerPage = candPerPage;
if ( pci->nChoicePerPage > MAX_SELKEY )
pci->nChoicePerPage = MAX_SELKEY;
pci->nPage = CEIL_DIV( pci->nTotalChoice, pci->nChoicePerPage );
pci->pageNo = 0;
}
/** @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 of certain length.
*
* Loading all possible phrases of certain length into ChoiceInfo structure
* from static and dynamic dictionaries, including number of total pages and
* the number of current page.
*/
static void SetChoiceInfo(ChewingData *pgdata)
{
Phrase tempPhrase;
int len;
UserPhraseData *pUserPhraseData;
uint16_t userPhoneSeq[MAX_PHONE_SEQ_LEN];
ChoiceInfo *pci = &(pgdata->choiceInfo);
AvailInfo *pai = &(pgdata->availInfo);
uint16_t *phoneSeq = pgdata->phoneSeq;
uint16_t *phoneSeqAlt = pgdata->phoneSeqAlt;
int cursor = PhoneSeqCursor(pgdata);
int candPerPage = pgdata->config.candPerPage;
/* Clears previous candidates. */
memset(pci->totalChoiceStr, '\0', MAX_CHOICE * MAX_PHRASE_LEN * MAX_UTF8_SIZE + 1);
pci->nTotalChoice = 0;
len = pai->avail[pai->currentAvail].len;
assert(len);
/* secondly, read tree phrase */
if (len == 1) { /* single character */
ChoiceInfoAppendChi(pgdata, pci, phoneSeq[cursor]);
if (phoneSeq[cursor] != phoneSeqAlt[cursor]) {
ChoiceInfoAppendChi(pgdata, pci, phoneSeqAlt[cursor]);
}
if (pgdata->bopomofoData.kbtype == KB_HSU || pgdata->bopomofoData.kbtype == KB_DVORAK_HSU) {
switch (phoneSeq[cursor]) {
case 0x2800: /* 'ㄘ' */
ChoiceInfoAppendChi(pgdata, pci, 0x30); /* 'ㄟ' */
break;
case 0x80: /* 'ㄧ' */
ChoiceInfoAppendChi(pgdata, pci, 0x20); /* 'ㄝ' */
break;
case 0x2A00: /* 'ㄙ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1); /* '˙' */
break;
case 0xA00: /* 'ㄉ' */
ChoiceInfoAppendChi(pgdata, pci, 0x2); /* 'ˊ' */
break;
case 0x800: /* 'ㄈ' */
ChoiceInfoAppendChi(pgdata, pci, 0x3); /* 'ˇ' */
break;
case 0x18: /* 'ㄜ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1200); /* 'ㄍ' */
break;
case 0x10: /* 'ㄛ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1600); /* 'ㄏ' */
break;
case 0x1E00: /* 'ㄓ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1800); /* 'ㄐ' */
ChoiceInfoAppendChi(pgdata, pci, 0x4); /* 'ˋ' */
break;
case 0x58: /* 'ㄤ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1400); /* 'ㄎ' */
break;
case 0x68: /* 'ㄦ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1000); /* 'ㄌ' */
ChoiceInfoAppendChi(pgdata, pci, 0x60); /* 'ㄥ' */
break;
case 0x2200: /* 'ㄕ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1C00); /* 'ㄒ' */
break;
case 0x2000: /* 'ㄔ' */
ChoiceInfoAppendChi(pgdata, pci, 0x1A00); /* 'ㄑ' */
break;
case 0x50: /* 'ㄣ' */
ChoiceInfoAppendChi(pgdata, pci, 0xE00); /* 'ㄋ' */
break;
case 0x48: /* 'ㄢ' */
ChoiceInfoAppendChi(pgdata, pci, 0x600); /* 'ㄇ' */
break;
default:
break;
}
}
}
/* phrase */
else {
if (pai->avail[pai->currentAvail].id) {
GetPhraseFirst(pgdata, &tempPhrase, pai->avail[pai->currentAvail].id);
do {
if (ChoiceTheSame(pci, tempPhrase.phrase, len * ueBytesFromChar(tempPhrase.phrase[0]))) {
continue;
}
ueStrNCpy(pci->totalChoiceStr[pci->nTotalChoice], tempPhrase.phrase, len, 1);
pci->nTotalChoice++;
} while (GetVocabNext(pgdata, &tempPhrase));
}
memcpy(userPhoneSeq, &phoneSeq[cursor], sizeof(uint16_t) * len);
userPhoneSeq[len] = 0;
pUserPhraseData = UserGetPhraseFirst(pgdata, userPhoneSeq);
if (pUserPhraseData) {
do {
/* check if the phrase is already in the choice list */
if (ChoiceTheSame(pci, pUserPhraseData->wordSeq, len * ueBytesFromChar(pUserPhraseData->wordSeq[0])))
continue;
/* otherwise store it */
ueStrNCpy(pci->totalChoiceStr[pci->nTotalChoice], pUserPhraseData->wordSeq, len, 1);
pci->nTotalChoice++;
} while ((pUserPhraseData = UserGetPhraseNext(pgdata, userPhoneSeq)) != NULL);
}
UserGetPhraseEnd(pgdata, userPhoneSeq);
}
/* magic number */
pci->nChoicePerPage = candPerPage;
assert(pci->nTotalChoice > 0);
pci->nPage = CEIL_DIV(pci->nTotalChoice, pci->nChoicePerPage);
pci->pageNo = 0;
pci->isSymbol = WORD_CHOICE;
}
/** @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++;
}
}
}
