bool convert_to_utf8(FacadePhraseIndex * phrase_index,
MatchResults match_results,
/* in */ const char * delimiter,
/* out */ char * & result_string){
//init variables
if ( NULL == delimiter )
delimiter = "";
result_string = NULL;
PhraseItem item;
for ( size_t i = 0; i < match_results->len; ++i ){
phrase_token_t * token = &g_array_index
(match_results, phrase_token_t, i);
if ( null_token == *token )
continue;
phrase_index->get_phrase_item(*token, item);
ucs4_t buffer[MAX_PHRASE_LENGTH];
item.get_phrase_string(buffer);
guint8 length = item.get_phrase_length();
gchar * phrase = g_ucs4_to_utf8(buffer, length, NULL, NULL, NULL);
char * tmp = result_string;
if ( NULL == result_string )
result_string = g_strdup(phrase);
else
result_string = g_strconcat(result_string, delimiter, phrase, NULL);
g_free(tmp); g_free(phrase);
}
return true;
}
bool gen_unigram(FILE * output, FacadePhraseIndex * phrase_index) {
fprintf(output, "\\1-gram\n");
for ( size_t i = 0; i < PHRASE_INDEX_LIBRARY_COUNT; i++) {
PhraseIndexRange range;
int result = phrase_index->get_range(i, range);
if (ERROR_OK != result )
continue;
PhraseItem item;
for (phrase_token_t token = range.m_range_begin;
token < range.m_range_end; token++) {
int result = phrase_index->get_phrase_item(token, item);
if ( result == ERROR_NO_ITEM )
continue;
assert( result == ERROR_OK);
size_t freq = item.get_unigram_frequency();
if ( 0 == freq )
continue;
char * phrase = taglib_token_to_string(phrase_index, token);
if ( phrase )
fprintf(output, "\\item %d %s count %ld\n", token, phrase, freq);
g_free(phrase);
}
}
return true;
}
parameter_t compute_interpolation(SingleGram * deleted_bigram,
FacadePhraseIndex * unigram,
SingleGram * bigram){
bool success;
parameter_t lambda = 0, next_lambda = 0.6;
parameter_t epsilon = 0.001;
while ( fabs(lambda - next_lambda) > epsilon){
lambda = next_lambda;
next_lambda = 0;
guint32 table_num = 0;
parameter_t numerator = 0;
parameter_t part_of_denominator = 0;
BigramPhraseWithCountArray array = g_array_new(FALSE, FALSE, sizeof(BigramPhraseItemWithCount));
deleted_bigram->retrieve_all(array);
for ( int i = 0; i < array->len; ++i){
BigramPhraseItemWithCount * item = &g_array_index(array, BigramPhraseItemWithCount, i);
//get the phrase token
phrase_token_t token = item->m_token;
guint32 deleted_count = item->m_count;
{
guint32 freq = 0;
parameter_t elem_poss = 0;
if (bigram && bigram->get_freq(token, freq)){
guint32 total_freq;
assert(bigram->get_total_freq(total_freq));
assert(0 != total_freq);
elem_poss = freq / (parameter_t) total_freq;
}
numerator = lambda * elem_poss;
}
{
guint32 freq = 0;
parameter_t elem_poss = 0;
PhraseItem item;
if (!unigram->get_phrase_item(token, item)){
guint32 freq = item.get_unigram_frequency();
guint32 total_freq = unigram->get_phrase_index_total_freq();
elem_poss = freq / (parameter_t)total_freq;
}
part_of_denominator = (1 - lambda) * elem_poss;
}
if (0 == (numerator + part_of_denominator))
continue;
next_lambda += deleted_count * (numerator / (numerator + part_of_denominator));
}
assert(deleted_bigram->get_total_freq(table_num));
next_lambda /= table_num;
g_array_free(array, TRUE);
}
lambda = next_lambda;
return lambda;
}
bool get_possible_pinyin(FacadePhraseIndex * phrase_index,
TokenVector tokens, ChewingKeyVector keys){
ChewingKey buffer[MAX_PHRASE_LENGTH];
size_t key_index; guint32 max_freq;
guint32 freq;
g_array_set_size(keys, 0);
for (size_t i = 0; i < tokens->len; ++i){
phrase_token_t * token = &g_array_index(tokens, phrase_token_t, i);
PhraseItem item;
phrase_index->get_phrase_item(*token, item);
key_index = 0; max_freq = 0;
for ( size_t m = 0; m < item.get_n_pronunciation(); ++m ) {
freq = 0;
assert(item.get_nth_pronunciation(m, buffer, freq));
if ( freq > max_freq ) {
key_index = m;
max_freq = freq;
}
}
assert(item.get_nth_pronunciation(key_index, buffer, freq));
assert(max_freq == freq);
guint8 len = item.get_phrase_length();
g_array_append_vals(keys, buffer, len);
}
return true;
}
int main( int argc, char * argv[]){
PinyinCustomSettings custom;
PinyinLargeTable largetable(&custom);
FacadePhraseIndex phrase_index;
FILE * gbfile = fopen("../../data/gb_char.table", "r");
if ( gbfile == NULL ) {
fprintf(stderr, "open gb_char.table failed!\n");
exit(ENOENT);
}
largetable.load_text(gbfile);
fseek(gbfile, 0L, SEEK_SET);
phrase_index.load_text(1, gbfile);
fclose(gbfile);
FILE * gbkfile = fopen("../../data/gbk_char.table","r");
if ( gbkfile == NULL ) {
fprintf(stderr, "open gb_char.table failed!\n");
exit(ENOENT);
}
largetable.load_text(gbkfile);
fseek(gbkfile, 0L, SEEK_SET);
phrase_index.load_text(2, gbkfile);
fclose(gbkfile);
MemoryChunk* new_chunk = new MemoryChunk;
largetable.store(new_chunk);
largetable.load(new_chunk);
char* linebuf = NULL;
size_t size = 0;
while( getline(&linebuf, &size, stdin) ){
linebuf[strlen(linebuf)-1] = '\0';
if ( strcmp ( linebuf, "quit" ) == 0)
break;
PinyinDefaultParser parser;
NullPinyinValidator validator;
PinyinKeyVector keys;
PinyinKeyPosVector poses;
keys = g_array_new(FALSE, FALSE, sizeof( PinyinKey));
poses = g_array_new(FALSE, FALSE, sizeof( PinyinKeyPos));
parser.parse(validator, keys, poses, linebuf);
guint32 start = record_time();
PhraseIndexRanges ranges;
for( size_t i = 0 ; i < PHRASE_INDEX_LIBRARY_COUNT ; ++i){
ranges[i] = g_array_new(FALSE, FALSE, sizeof (PhraseIndexRange));
}
for ( size_t i = 0 ; i < bench_times; ++i){
largetable.search(keys->len, (PinyinKey *)keys->data, ranges);
}
for( size_t i = 0 ; i < PHRASE_INDEX_LIBRARY_COUNT ; ++i){
GArray * range = ranges[i];
g_array_set_size( range, 0);
}
print_time(start, bench_times);
largetable.search(keys->len, (PinyinKey *)keys->data, ranges);
for( size_t i = 0 ; i < PHRASE_INDEX_LIBRARY_COUNT ; ++i){
GArray * range = ranges[i];
if ( range ){
for (size_t k = 0; k < range->len; ++k){
PhraseIndexRange* onerange = &g_array_index(range, PhraseIndexRange, k);
printf("start:%d\tend:%d\n", onerange->m_range_begin, onerange->m_range_end);
PhraseItem item;
for ( phrase_token_t token = onerange->m_range_begin; token != onerange->m_range_end; ++token){
phrase_index.get_phrase_item( token, item);
gunichar2 bufstr[1024];
item.get_phrase_string(bufstr);
char * string = g_utf16_to_utf8
( bufstr, item.get_phrase_length(),
NULL, NULL, NULL);
printf("%s\t", string);
g_free(string);
PinyinKey pinyin_buffer[1024];
size_t npron = item.get_n_pronunciation();
guint32 freq;
for ( size_t n = 0; n < npron; ++n){
item.get_nth_pronunciation(n, pinyin_buffer, freq);
for ( size_t o = 0; o < item.get_phrase_length(); ++o){
printf("%s'", pinyin_buffer[o].get_key_string());
}
printf("\b\t%d\t", freq);
}
printf("\n");
}
}
if ( range->len)
printf("range items number:%d\n", range->len);
}
g_array_set_size( range, 0);
}
g_array_free(keys, TRUE);
g_array_free(poses, TRUE);
}
if (linebuf)
free(linebuf);
return 0;
}
int main(int argc, char * argv[]) {
pinyin_option_t options = USE_TONE | PINYIN_INCOMPLETE;
ChewingLargeTable largetable(options);
FacadePhraseIndex phrase_index;
FILE * gbfile = fopen("../../data/gb_char.table", "r");
if (NULL == gbfile) {
fprintf(stderr, "open gb_char.table failed!\n");
exit(ENOENT);
}
largetable.load_text(gbfile);
fseek(gbfile, 0L, SEEK_SET);
phrase_index.load_text(1, gbfile);
fclose(gbfile);
FILE * gbkfile = fopen("../../data/gbk_char.table", "r");
if (NULL == gbkfile) {
fprintf(stderr, "open gbk_char.table failed!\n");
exit(ENOENT);
}
largetable.load_text(gbkfile);
fseek(gbkfile, 0L, SEEK_SET);
phrase_index.load_text(2, gbkfile);
fclose(gbkfile);
MemoryChunk * new_chunk = new MemoryChunk;
largetable.store(new_chunk);
largetable.load(new_chunk);
char* linebuf = NULL; size_t size = 0;
while( getline(&linebuf, &size, stdin) ){
linebuf[strlen(linebuf)-1] = '\0';
if ( strcmp ( linebuf, "quit" ) == 0)
break;
FullPinyinParser2 parser;
ChewingKeyVector keys = g_array_new(FALSE, FALSE, sizeof(ChewingKey));
ChewingKeyRestVector key_rests =
g_array_new(FALSE, FALSE, sizeof(ChewingKeyRest));
parser.parse(options, keys, key_rests, linebuf, strlen(linebuf));
if (0 == keys->len) {
fprintf(stderr, "Invalid input.\n");
continue;
}
guint32 start = record_time();
PhraseIndexRanges ranges;
memset(ranges, 0, sizeof(PhraseIndexRanges));
guint8 min_index, max_index;
phrase_index.get_sub_phrase_range(min_index, max_index);
for (size_t i = min_index; i < max_index; ++i) {
ranges[i] = g_array_new(FALSE, FALSE, sizeof(PhraseIndexRange));
}
for (size_t i = 0; i < bench_times; ++i) {
largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
}
for (size_t i = min_index; i < max_index; ++i) {
g_array_set_size(ranges[i], 0);
}
print_time(start, bench_times);
largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
for (size_t i = min_index; i < max_index; ++i) {
GArray * & range = ranges[i];
if (range) {
if (range->len)
printf("range items number:%d\n", range->len);
for (size_t k = 0; k < range->len; ++k) {
PhraseIndexRange * onerange =
&g_array_index(range, PhraseIndexRange, k);
printf("start:%d\tend:%d\n", onerange->m_range_begin,
onerange->m_range_end);
PhraseItem item;
for ( phrase_token_t token = onerange->m_range_begin;
token != onerange->m_range_end; ++token){
phrase_index.get_phrase_item( token, item);
/* get phrase string */
gunichar2 buffer[MAX_PHRASE_LENGTH + 1];
item.get_phrase_string(buffer);
char * string = g_utf16_to_utf8
( buffer, item.get_phrase_length(),
NULL, NULL, NULL);
printf("%s\t", string);
g_free(string);
ChewingKey chewing_buffer[MAX_PHRASE_LENGTH];
size_t npron = item.get_n_pronunciation();
guint32 freq;
for (size_t m = 0; m < npron; ++m){
item.get_nth_pronunciation(m, chewing_buffer, freq);
for (size_t n = 0; n < item.get_phrase_length();
++n){
printf("%s'",
chewing_buffer[n].get_pinyin_string());
}
printf("\b\t%d\t", freq);
}
}
printf("\n");
}
}
g_array_set_size(range, 0);
}
g_array_free(keys, TRUE);
g_array_free(key_rests, TRUE);
}
if (linebuf)
free(linebuf);
return 0;
}
int main(int argc, char * argv[]) {
SystemTableInfo system_table_info;
bool retval = system_table_info.load("../../data/table.conf");
if (!retval) {
fprintf(stderr, "load table.conf failed.\n");
exit(ENOENT);
}
pinyin_option_t options = USE_TONE | PINYIN_INCOMPLETE;
ChewingLargeTable largetable(options);
FacadePhraseIndex phrase_index;
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
if (!load_phrase_table(phrase_files, &largetable, NULL, &phrase_index))
exit(ENOENT);
MemoryChunk * new_chunk = new MemoryChunk;
largetable.store(new_chunk);
largetable.load(new_chunk);
char* linebuf = NULL; size_t size = 0; ssize_t read;
while ((read = getline(&linebuf, &size, stdin)) != -1) {
if ( '\n' == linebuf[strlen(linebuf) - 1] ) {
linebuf[strlen(linebuf) - 1] = '\0';
}
if ( strcmp ( linebuf, "quit" ) == 0)
break;
FullPinyinParser2 parser;
ChewingKeyVector keys = g_array_new(FALSE, FALSE, sizeof(ChewingKey));
ChewingKeyRestVector key_rests =
g_array_new(FALSE, FALSE, sizeof(ChewingKeyRest));
parser.parse(options, keys, key_rests, linebuf, strlen(linebuf));
if (0 == keys->len) {
fprintf(stderr, "Invalid input.\n");
continue;
}
guint32 start = record_time();
PhraseIndexRanges ranges;
memset(ranges, 0, sizeof(PhraseIndexRanges));
phrase_index.prepare_ranges(ranges);
for (size_t i = 0; i < bench_times; ++i) {
phrase_index.clear_ranges(ranges);
largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
}
print_time(start, bench_times);
phrase_index.clear_ranges(ranges);
largetable.search(keys->len, (ChewingKey *)keys->data, ranges);
for (size_t i = 0; i < PHRASE_INDEX_LIBRARY_COUNT; ++i) {
GArray * & range = ranges[i];
if (!range)
continue;
if (range->len)
printf("range items number:%d\n", range->len);
for (size_t k = 0; k < range->len; ++k) {
PhraseIndexRange * onerange =
&g_array_index(range, PhraseIndexRange, k);
printf("start:%d\tend:%d\n", onerange->m_range_begin,
onerange->m_range_end);
PhraseItem item;
for ( phrase_token_t token = onerange->m_range_begin;
token != onerange->m_range_end; ++token){
phrase_index.get_phrase_item( token, item);
/* get phrase string */
ucs4_t buffer[MAX_PHRASE_LENGTH + 1];
item.get_phrase_string(buffer);
char * string = g_ucs4_to_utf8
( buffer, item.get_phrase_length(),
NULL, NULL, NULL);
printf("%s\t", string);
g_free(string);
ChewingKey chewing_buffer[MAX_PHRASE_LENGTH];
size_t npron = item.get_n_pronunciation();
guint32 freq;
for (size_t m = 0; m < npron; ++m){
item.get_nth_pronunciation(m, chewing_buffer, freq);
for (size_t n = 0; n < item.get_phrase_length();
++n){
gchar * pinyins =
chewing_buffer[n].get_pinyin_string();
printf("%s'", pinyins);
g_free(pinyins);
}
printf("\b\t%d\t", freq);
}
}
printf("\n");
}
g_array_set_size(range, 0);
}
phrase_index.destroy_ranges(ranges);
g_array_free(keys, TRUE);
g_array_free(key_rests, TRUE);
}
if (linebuf)
free(linebuf);
/* mask out all index items. */
largetable.mask_out(0x0, 0x0);
return 0;
}
