int main(int argc, char * argv[]){
FILE * output = stdout;
const char * bigram_filename = "bigram.db";
FacadePhraseIndex phrase_index;
//gb_char binary file
MemoryChunk * chunk = new MemoryChunk;
chunk->load("gb_char.bin");
phrase_index.load(1, chunk);
//gbk_char binary file
chunk = new MemoryChunk;
chunk->load("gbk_char.bin");
phrase_index.load(2, chunk);
Bigram bigram;
bigram.attach(bigram_filename, ATTACH_READONLY);
begin_data(output);
gen_unigram(output, &phrase_index);
gen_bigram(output, &phrase_index, &bigram);
end_data(output);
return 0;
}
int main(int argc, char * argv[]){
FacadePhraseIndex phrase_index;
/* gb_char binary file */
MemoryChunk * chunk = new MemoryChunk;
bool retval = chunk->load("gb_char.bin");
if (!retval) {
fprintf(stderr, "open gb_char.bin failed!\n");
exit(ENOENT);
}
phrase_index.load(1, chunk);
/* gbk_char binary file */
chunk = new MemoryChunk;
retval = chunk->load("gbk_char.bin");
if (!retval) {
fprintf(stderr, "open gbk_char.bin failed!\n");
exit(ENOENT);
}
phrase_index.load(2, chunk);
/* Note: please increase the value when corpus size becomes larger.
* To avoid zero value when computing unigram frequency in float format.
*/
guint32 freq = 1; PhraseIndexRange range;
int result = phrase_index.get_range(1, range);
if ( result == ERROR_OK ) {
for ( size_t i = range.m_range_begin; i <= range.m_range_end; ++i ) {
phrase_index.add_unigram_frequency(i, freq);
}
}
#if 1
result = phrase_index.get_range(2, range);
if ( result == ERROR_OK ) {
for ( size_t i = range.m_range_begin; i <= range.m_range_end; ++i ) {
phrase_index.add_unigram_frequency(i, freq);
}
}
#endif
MemoryChunk * new_chunk = new MemoryChunk;
phrase_index.store(1, new_chunk);
new_chunk->save("gb_char.bin");
phrase_index.load(1, new_chunk);
new_chunk = new MemoryChunk;
phrase_index.store(2, new_chunk);
new_chunk->save("gbk_char.bin");
phrase_index.load(2, new_chunk);
return 0;
}
int main(int argc, char * argv[]){
FacadePhraseIndex phrase_index;
//gb_char binary file
MemoryChunk * chunk = new MemoryChunk;
chunk->load("gb_char.bin");
phrase_index.load(1, chunk);
//gbk_char binary file
chunk = new MemoryChunk;
chunk->load("gbk_char.bin");
phrase_index.load(2, chunk);
Bigram bigram;
bigram.attach("bigram.db", ATTACH_READONLY);
Bigram deleted_bigram;
deleted_bigram.attach("deleted_bigram.db", ATTACH_READONLY);
GArray * deleted_items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t));
deleted_bigram.get_all_items(deleted_items);
parameter_t lambda_sum = 0;
int lambda_count = 0;
for ( int i = 0; i < deleted_items->len; ++i ){
phrase_token_t * token = &g_array_index(deleted_items, phrase_token_t, i);
SingleGram * single_gram = NULL;
bigram.load(*token, single_gram);
SingleGram * deleted_single_gram = NULL;
deleted_bigram.load(*token, deleted_single_gram);
parameter_t lambda = compute_interpolation(deleted_single_gram, &phrase_index, single_gram);
printf("token:%d lambda:%f\n", *token, lambda);
lambda_sum += lambda;
lambda_count ++;
if (single_gram)
delete single_gram;
delete deleted_single_gram;
}
printf("average lambda:%f\n", (lambda_sum/lambda_count));
g_array_free(deleted_items, TRUE);
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
const char * k_mixture_model_filename = NULL;
FILE * output = stdout;
while ( i < argc ){
if ( strcmp ("--help", argv[i]) == 0 ){
print_help();
exit(0);
} else if ( strcmp ("--k-mixture-model-file", argv[i]) == 0 ){
if ( ++i > argc ){
print_help();
exit(EINVAL);
}
k_mixture_model_filename = argv[i];
} else {
print_help();
exit(EINVAL);
}
++i;
}
FacadePhraseIndex phrase_index;
//gb_char binary file
MemoryChunk * chunk = new MemoryChunk;
chunk->load("gb_char.bin");
phrase_index.load(1, chunk);
//gbk_char binary file
chunk = new MemoryChunk;
chunk->load("gbk_char.bin");
phrase_index.load(2, chunk);
KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
bigram.attach(k_mixture_model_filename, ATTACH_READONLY);
print_k_mixture_model_magic_header(output, &bigram);
print_k_mixture_model_array_headers(output, &bigram, &phrase_index);
print_k_mixture_model_array_items(output, &bigram, &phrase_index);
end_data(output);
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;
FacadeChewingTable largetable;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("../../data/pinyin_index.bin");
largetable.load(options, chunk, NULL);
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
FacadePhraseIndex phrase_index;
if (!load_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
Bigram system_bigram;
system_bigram.attach("../../data/bigram.db", ATTACH_READONLY);
Bigram user_bigram;
user_bigram.attach(NULL, ATTACH_CREATE|ATTACH_READWRITE);
gfloat lambda = system_table_info.get_lambda();
PinyinLookup2 pinyin_lookup(lambda, options,
&largetable, &phrase_index,
&system_bigram, &user_bigram);
/* prepare the prefixes for get_best_match. */
TokenVector prefixes = g_array_new
(FALSE, FALSE, sizeof(phrase_token_t));
g_array_append_val(prefixes, sentence_start);
CandidateConstraints constraints = g_array_new
(TRUE, FALSE, sizeof(lookup_constraint_t));
MatchResults results = g_array_new(FALSE, FALSE, sizeof(phrase_token_t));
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 ) /* invalid pinyin */
continue;
/* initialize constraints. */
g_array_set_size(constraints, keys->len);
for ( size_t i = 0; i < constraints->len; ++i){
lookup_constraint_t * constraint = &g_array_index(constraints, lookup_constraint_t, i);
constraint->m_type = NO_CONSTRAINT;
}
guint32 start_time = record_time();
for ( size_t i = 0; i < bench_times; ++i)
pinyin_lookup.get_best_match(prefixes, keys, constraints, results);
print_time(start_time, bench_times);
for ( size_t i = 0; i < results->len; ++i){
phrase_token_t * token = &g_array_index(results, phrase_token_t, i);
if ( null_token == *token)
continue;
printf("pos:%ld,token:%d\t", i, *token);
}
printf("\n");
char * sentence = NULL;
pinyin_lookup.convert_to_utf8(results, sentence);
printf("%s\n", sentence);
g_array_free(keys, TRUE);
g_array_free(key_rests, TRUE);
g_free(sentence);
}
g_array_free(prefixes, TRUE);
g_array_free(constraints, TRUE);
g_array_free(results, TRUE);
free(linebuf);
return 0;
}
int main(int argc, char * argv[]){
FILE * input = stdin;
const char * bigram_filename = "bigram.db";
PhraseLargeTable phrases;
MemoryChunk * chunk = new MemoryChunk;
bool retval = chunk->load("phrase_index.bin");
if (!retval) {
fprintf(stderr, "open phrase_index.bin failed!\n");
exit(ENOENT);
}
phrases.load(chunk);
FacadePhraseIndex phrase_index;
if (!load_phrase_index(&phrase_index))
exit(ENOENT);
Bigram bigram;
retval = bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE);
if (!retval) {
fprintf(stderr, "open %s failed!\n", bigram_filename);
exit(ENOENT);
}
taglib_init();
values = g_ptr_array_new();
required = g_hash_table_new(g_str_hash, g_str_equal);
//enter "\data" line
assert(taglib_add_tag(BEGIN_LINE, "\\data", 0, "model", ""));
ssize_t result = my_getline(input);
if ( result == -1 ) {
fprintf(stderr, "empty file input.\n");
exit(ENODATA);
}
//read "\data" line
if ( !taglib_read(linebuf, line_type, values, required) ) {
fprintf(stderr, "error: interpolation model expected.\n");
exit(ENODATA);
}
assert(line_type == BEGIN_LINE);
char * value = NULL;
assert(g_hash_table_lookup_extended(required, "model", NULL, (gpointer *)&value));
if ( !( strcmp("interpolation", value) == 0 ) ) {
fprintf(stderr, "error: interpolation model expected.\n");
exit(ENODATA);
}
result = my_getline(input);
if ( result != -1 )
parse_body(input, &phrases, &phrase_index, &bigram);
taglib_fini();
if (!save_phrase_index(&phrase_index))
exit(ENOENT);
return 0;
}
int main(int argc, char * argv[]){
MemoryChunk* chunk;
chunk = new MemoryChunk();
int i = 12;
chunk->set_content(0, &i, sizeof(int));
int * p = (int *)chunk->begin();
assert(chunk->size() == sizeof(int));
std::cout<<*p<<std::endl;
std::cout<<chunk->capacity()<<std::endl;
p = & i;
chunk->set_chunk(p, sizeof(int), NULL);
short t = 5;
chunk->set_content(sizeof(int), &t, sizeof(short));
assert( sizeof(int) + sizeof(short) == chunk->size());
std::cout<<chunk->capacity()<<std::endl;
p = (int *)chunk->begin();
short * p2 =(short *)(((char *) (chunk->begin())) + sizeof(int));
std::cout<<*p<<'\t'<<*p2<<std::endl;
chunk->set_content(sizeof(int) + sizeof(short), &t, sizeof(short));
assert( sizeof(int) + (sizeof(short) << 1) == chunk->size());
std::cout<<chunk->capacity()<<std::endl;
p = (int *)chunk->begin();
p2 =(short *)(((char *) (chunk->begin())) + sizeof(int));
std::cout<<*p<<'\t'<<*p2<<'\t'<<*(p2 + 1)<<std::endl;
chunk->set_size(sizeof(int) + sizeof(short) *3);
p = (int *)chunk->begin();
p2 =(short *)(((char *) (chunk->begin())) + sizeof(int));
chunk->set_content(0, &i, sizeof(int));
*(p2+2) = 3;
std::cout<<*p<<'\t'<<*p2<<'\t'<<*(p2 + 1)<<'\t'<<*(p2+2)<<std::endl;
int m = 10;
chunk->set_chunk(&m, sizeof(int), NULL);
int n = 12;
chunk->insert_content(sizeof(int), &n, sizeof(int));
n = 11;
chunk->insert_content(sizeof(int), &n, sizeof(int));
int * p3 = (int *)chunk->begin();
std::cout<<*p3<<'\t'<<*(p3+1)<<'\t'<<*(p3+2)<<std::endl;
chunk->remove_content(sizeof(int), sizeof(int));
std::cout<<*p3<<'\t'<<*(p3+1)<<std::endl;
int tmp;
assert(chunk->get_content(sizeof(int), &tmp, sizeof(int)));
std::cout<<tmp<<std::endl;
delete chunk;
const char * filename = "/tmp/version";
const char * version = "0.2.0";
chunk = new MemoryChunk;
bool retval = chunk->load(filename);
if ( !retval ){
std::cerr<<"can't find chunk"<<std::endl;
}else{
if ( memcmp(version, chunk->begin(), strlen(version) + 1) == 0){
std::cout<<"match"<<std::endl;
}
}
chunk->set_content(0, version, strlen(version) + 1);
chunk->save(filename);
retval = chunk->load(filename);
if ( !retval ){
std::cerr<<"can't find chunk"<<std::endl;
}
if ( memcmp(version, chunk->begin(), strlen(version) + 1) == 0){
std::cout<<"match"<<std::endl;
}
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
bool gen_extra_enter = false;
setlocale(LC_ALL, "");
//deal with options.
while ( i < argc ){
if ( strcmp ("--help", argv[i]) == 0 ){
print_help();
exit(0);
} else if ( strcmp("--generate-extra-enter", argv[i]) == 0 ){
gen_extra_enter = true;
} else {
print_help();
exit(EINVAL);
}
++i;
}
//init phrase table
PhraseLargeTable phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
phrase_table.load(chunk);
//init phrase index
FacadePhraseIndex phrase_index;
chunk = new MemoryChunk;
chunk->load("gb_char.bin");
phrase_index.load(1, chunk);
chunk = new MemoryChunk;
chunk->load("gbk_char.bin");
phrase_index.load(2, chunk);
//init bi-gram
Bigram system_bigram;
system_bigram.attach("bigram.db", ATTACH_READONLY);
Bigram user_bigram;
//init phrase lookup
g_phrase_lookup = new PhraseLookup(&phrase_table, &phrase_index,
&system_bigram, &user_bigram);
CONTEXT_STATE state, next_state;
GArray * current_utf16 = g_array_new(TRUE, TRUE, sizeof(utf16_t));
phrase_token_t token = null_token;
//split the sentence
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';
}
//check non-ucs2 characters
const glong num_of_chars = g_utf8_strlen(linebuf, -1);
glong len = 0;
utf16_t * sentence = g_utf8_to_utf16(linebuf, -1, NULL, &len, NULL);
if ( len != num_of_chars ) {
fprintf(stderr, "non-ucs2 characters encountered:%s.\n", linebuf);
printf("\n");
continue;
}
/* only new-line persists. */
if ( 0 == num_of_chars ) {
printf("\n");
continue;
}
state = CONTEXT_INIT;
bool result = phrase_table.search( 1, sentence, token);
g_array_append_val( current_utf16, sentence[0]);
if ( result & SEARCH_OK )
state = CONTEXT_SEGMENTABLE;
else
state = CONTEXT_UNKNOWN;
for ( int i = 1; i < num_of_chars; ++i) {
bool result = phrase_table.search( 1, sentence + i, token);
if ( result & SEARCH_OK )
next_state = CONTEXT_SEGMENTABLE;
else
next_state = CONTEXT_UNKNOWN;
if ( state == next_state ){
g_array_append_val(current_utf16, sentence[i]);
continue;
}
assert ( state != next_state );
if ( state == CONTEXT_SEGMENTABLE )
deal_with_segmentable(current_utf16);
if ( state == CONTEXT_UNKNOWN )
deal_with_unknown(current_utf16);
/* save the current character */
g_array_set_size(current_utf16, 0);
g_array_append_val(current_utf16, sentence[i]);
state = next_state;
}
if ( current_utf16->len ) {
/* this seems always true. */
if ( state == CONTEXT_SEGMENTABLE )
deal_with_segmentable(current_utf16);
if ( state == CONTEXT_UNKNOWN )
deal_with_unknown(current_utf16);
g_array_set_size(current_utf16, 0);
}
/* print extra enter */
if ( gen_extra_enter )
printf("\n");
}
delete g_phrase_lookup;
g_phrase_lookup = NULL;
/* print enter at file tail */
printf("\n");
g_array_free(current_utf16, TRUE);
free(linebuf);
return 0;
}
int main(int argc, char * argv[]){
const char * evals_text = "evals.text";
pinyin_option_t options = USE_TONE;
FacadeChewingTable largetable;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("pinyin_index.bin");
largetable.load(options, chunk, NULL);
FacadePhraseTable2 phrase_table;
chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
phrase_table.load(chunk, NULL);
FacadePhraseIndex phrase_index;
if (!load_phrase_index(&phrase_index))
exit(ENOENT);
Bigram system_bigram;
system_bigram.attach("bigram.db", ATTACH_READONLY);
Bigram user_bigram;
user_bigram.attach(NULL, ATTACH_CREATE|ATTACH_READWRITE);
PinyinLookup2 pinyin_lookup(options, &largetable, &phrase_index,
&system_bigram, &user_bigram);
/* open evals.text. */
FILE * evals_file = fopen(evals_text, "r");
if ( NULL == evals_file ) {
fprintf(stderr, "Can't open file:%s\n", evals_text);
exit(ENOENT);
}
PhraseTokens phrase_tokens;
memset(phrase_tokens, 0, sizeof(PhraseTokens));
phrase_index.prepare_tokens(phrase_tokens);
/* Evaluates the correction rate of test text documents. */
size_t tested_count = 0; size_t passed_count = 0;
char* linebuf = NULL; size_t size = 0;
TokenVector tokens = g_array_new(FALSE, TRUE, sizeof(phrase_token_t));
phrase_token_t token = null_token;
while( getline(&linebuf, &size, evals_file) ) {
if ( feof(evals_file) )
break;
if ( '\n' == linebuf[strlen(linebuf)-1] )
linebuf[strlen(linebuf)-1] = '\0';
glong phrase_len = 0;
ucs4_t * phrase = g_utf8_to_ucs4(linebuf, -1, NULL, &phrase_len, NULL);
token = null_token;
if ( 0 != phrase_len ) {
int result = phrase_table.search(phrase_len, phrase, phrase_tokens);
int num = get_first_token(phrase_tokens, token);
if ( !(result & SEARCH_OK) )
token = null_token;
g_free(phrase);
phrase = NULL;
}
if ( null_token == token ) {
if ( tokens->len ) { /* one test. */
if ( do_one_test(&pinyin_lookup, &phrase_index, tokens) ) {
tested_count ++; passed_count ++;
} else {
tested_count ++;
}
g_array_set_size(tokens, 0);
}
} else {
g_array_append_val(tokens, token);
}
}
if ( tokens->len ) { /* one test. */
if ( do_one_test(&pinyin_lookup, &phrase_index, tokens) ) {
tested_count ++; passed_count ++;
} else {
tested_count ++;
}
}
parameter_t rate = passed_count / (parameter_t) tested_count;
printf("correction rate:%f\n", rate);
g_array_free(tokens, TRUE);
fclose(evals_file);
free(linebuf);
phrase_index.destroy_tokens(phrase_tokens);
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
bool train_pi_gram = true;
const char * bigram_filename = "bigram.db";
setlocale(LC_ALL, "");
while ( i < argc ){
if ( strcmp("--help", argv[i]) == 0){
print_help();
exit(0);
}else if ( strcmp("--skip-pi-gram-training", argv[i]) == 0 ){
train_pi_gram = false;
}else if ( strcmp("--bigram-file", argv[i]) == 0){
if ( ++i >= argc ) {
print_help();
exit(EINVAL);
}
bigram_filename = argv[i];
}else{
print_help();
exit(EINVAL);
}
++i;
}
g_phrases = new PhraseLargeTable;
//init phrase lookup
MemoryChunk * chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
g_phrases->load(chunk);
FacadePhraseIndex phrase_index;
//gb_char binary file
chunk = new MemoryChunk;
chunk->load("gb_char.bin");
phrase_index.load(1, chunk);
//gbk_char binary file
chunk = new MemoryChunk;
chunk->load("gbk_char.bin");
phrase_index.load(2, chunk);
Bigram bigram;
bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE);
char* linebuf = NULL;
size_t size = 0;
phrase_token_t last_token, cur_token = last_token = 0;
while( getline(&linebuf, &size, stdin) ){
if ( feof(stdin) )
break;
linebuf[strlen(linebuf)-1] = '\0';
glong phrase_len = 0;
utf16_t * phrase = g_utf8_to_utf16(linebuf, -1, NULL, &phrase_len, NULL);
phrase_token_t token = 0;
if ( 0 != phrase_len ) {
int result = g_phrases->search( phrase_len, phrase, token);
if ( ! (result & SEARCH_OK) )
token = 0;
g_free(phrase);
phrase = NULL;
}
last_token = cur_token;
cur_token = token;
/* skip null_token in second word. */
if ( null_token == cur_token )
continue;
//training uni-gram
phrase_index.add_unigram_frequency(cur_token, 1);
/* skip pi-gram training. */
if ( null_token == last_token ){
if ( !train_pi_gram )
continue;
last_token = sentence_start;
}
//train bi-gram
SingleGram * single_gram = NULL;
bigram.load(last_token, single_gram);
if ( NULL == single_gram ){
single_gram = new SingleGram;
}
guint32 freq, total_freq;
//increase freq
if (single_gram->get_freq(cur_token, freq))
assert(single_gram->set_freq(cur_token, freq + 1));
else
assert(single_gram->insert_freq(cur_token, 1));
//increase total freq
single_gram->get_total_freq(total_freq);
single_gram->set_total_freq(total_freq + 1);
bigram.store(last_token, single_gram);
delete single_gram;
}
free(linebuf);
MemoryChunk * new_chunk = new MemoryChunk;
phrase_index.store(1, new_chunk);
new_chunk->save("gb_char.bin");
phrase_index.load(1, new_chunk);
new_chunk = new MemoryChunk;
phrase_index.store(2, new_chunk);
new_chunk->save("gbk_char.bin");
phrase_index.load(2, new_chunk);
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
bool gen_extra_enter = false;
setlocale(LC_ALL, "");
//deal with options.
while ( i < argc ){
if ( strcmp ("--help", argv[i]) == 0) {
print_help();
exit(0);
} else if (strcmp("--generate-extra-enter", argv[i]) == 0) {
gen_extra_enter = true;
} else {
print_help();
exit(EINVAL);
}
++i;
}
/* init phrase table */
FacadePhraseTable2 phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
phrase_table.load(chunk, NULL);
/* init phrase index */
FacadePhraseIndex phrase_index;
if (!load_phrase_index(&phrase_index))
exit(ENOENT);
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';
}
//check non-ucs4 characters
const glong num_of_chars = g_utf8_strlen(linebuf, -1);
glong len = 0;
ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL);
if ( len != num_of_chars ) {
fprintf(stderr, "non-ucs4 characters encountered:%s.\n", linebuf);
printf("\n");
continue;
}
//do segment stuff
GArray * strings = g_array_new(TRUE, TRUE, sizeof(SegmentStep));
segment(&phrase_table, &phrase_index, sentence, len, strings);
//print out the split phrase
for ( glong i = 0; i < strings->len; ++i ) {
SegmentStep * step = &g_array_index(strings, SegmentStep, i);
char * string = g_ucs4_to_utf8( step->m_phrase, step->m_phrase_len, NULL, NULL, NULL);
printf("%d %s\n", step->m_handle, string);
g_free(string);
}
/* print extra enter */
if ( gen_extra_enter )
printf("\n");
g_array_free(strings, TRUE);
g_free(sentence);
}
/* print enter at file tail */
printf("\n");
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
setlocale(LC_ALL, "");
//deal with options.
while ( i < argc ){
if ( strcmp ("--help", argv[i]) == 0 ){
print_help();
exit(0);
} else {
print_help();
exit(EINVAL);
}
++i;
}
//init phrase table
PhraseLargeTable phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("../../data/phrase_index.bin");
phrase_table.load(chunk);
//init phrase index
FacadePhraseIndex phrase_index;
chunk = new MemoryChunk;
chunk->load("../../data/gb_char.bin");
phrase_index.load(1, chunk);
chunk = new MemoryChunk;
chunk->load("../../data/gbk_char.bin");
phrase_index.load(2, chunk);
//init bi-gram
Bigram system_bigram;
system_bigram.attach("../../data/bigram.db", ATTACH_READONLY);
Bigram user_bigram;
//init phrase lookup
PhraseLookup phrase_lookup(&phrase_table, &phrase_index,
&system_bigram, &user_bigram);
//try one sentence
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;
//check non-ucs2 characters
const glong num_of_chars = g_utf8_strlen(linebuf, -1);
glong len = 0;
utf16_t * sentence = g_utf8_to_utf16(linebuf, -1, NULL, &len, NULL);
if ( len != num_of_chars ) {
fprintf(stderr, "non-ucs2 characters are not accepted.\n");
g_free(sentence);
continue;
}
try_phrase_lookup(&phrase_lookup, sentence, len);
g_free(sentence);
}
free(linebuf);
return 0;
}
int main(int argc, char * argv[]){
FILE * input = stdin;
FILE * output = stdout;
setlocale(LC_ALL, "");
GError * error = NULL;
GOptionContext * context;
context = g_option_context_new("- n-gram segment");
g_option_context_add_main_entries(context, entries, NULL);
if (!g_option_context_parse(context, &argc, &argv, &error)) {
g_print("option parsing failed:%s\n", error->message);
exit(EINVAL);
}
if (outputfile) {
output = fopen(outputfile, "w");
if (NULL == output) {
perror("open file failed");
exit(EINVAL);
}
}
if (argc > 2) {
fprintf(stderr, "too many arguments.\n");
exit(EINVAL);
}
if (2 == argc) {
input = fopen(argv[1], "r");
if (NULL == input) {
perror("open file failed");
exit(EINVAL);
}
}
SystemTableInfo system_table_info;
bool retval = system_table_info.load(SYSTEM_TABLE_INFO);
if (!retval) {
fprintf(stderr, "load table.conf failed.\n");
exit(ENOENT);
}
/* init phrase table */
FacadePhraseTable2 phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load(SYSTEM_PHRASE_INDEX);
phrase_table.load(chunk, NULL);
/* init phrase index */
FacadePhraseIndex phrase_index;
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
if (!load_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
/* init bi-gram */
Bigram system_bigram;
system_bigram.attach(SYSTEM_BIGRAM, ATTACH_READONLY);
Bigram user_bigram;
gfloat lambda = system_table_info.get_lambda();
/* init phrase lookup */
PhraseLookup phrase_lookup(lambda,
&phrase_table, &phrase_index,
&system_bigram, &user_bigram);
CONTEXT_STATE state, next_state;
GArray * current_ucs4 = g_array_new(TRUE, TRUE, sizeof(ucs4_t));
PhraseTokens tokens;
memset(tokens, 0, sizeof(PhraseTokens));
phrase_index.prepare_tokens(tokens);
/* split the sentence */
char * linebuf = NULL; size_t size = 0; ssize_t read;
while( (read = getline(&linebuf, &size, input)) != -1 ){
if ( '\n' == linebuf[strlen(linebuf) - 1] ) {
linebuf[strlen(linebuf) - 1] = '\0';
}
/* check non-ucs4 characters */
const glong num_of_chars = g_utf8_strlen(linebuf, -1);
glong len = 0;
ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL);
if ( len != num_of_chars ) {
fprintf(stderr, "non-ucs4 characters encountered:%s.\n", linebuf);
fprintf(output, "%d \n", null_token);
continue;
}
/* only new-line persists. */
if ( 0 == num_of_chars ) {
fprintf(output, "%d \n", null_token);
continue;
}
state = CONTEXT_INIT;
int result = phrase_table.search( 1, sentence, tokens);
g_array_append_val( current_ucs4, sentence[0]);
if ( result & SEARCH_OK )
state = CONTEXT_SEGMENTABLE;
else
state = CONTEXT_UNKNOWN;
for ( int i = 1; i < num_of_chars; ++i) {
int result = phrase_table.search( 1, sentence + i, tokens);
if ( result & SEARCH_OK )
next_state = CONTEXT_SEGMENTABLE;
else
next_state = CONTEXT_UNKNOWN;
if ( state == next_state ){
g_array_append_val(current_ucs4, sentence[i]);
continue;
}
assert ( state != next_state );
if ( state == CONTEXT_SEGMENTABLE )
deal_with_segmentable(&phrase_lookup, current_ucs4, output);
if ( state == CONTEXT_UNKNOWN )
deal_with_unknown(current_ucs4, output);
/* save the current character */
g_array_set_size(current_ucs4, 0);
g_array_append_val(current_ucs4, sentence[i]);
state = next_state;
}
if ( current_ucs4->len ) {
/* this seems always true. */
if ( state == CONTEXT_SEGMENTABLE )
deal_with_segmentable(&phrase_lookup, current_ucs4, output);
if ( state == CONTEXT_UNKNOWN )
deal_with_unknown(current_ucs4, output);
g_array_set_size(current_ucs4, 0);
}
/* print extra enter */
if ( gen_extra_enter )
fprintf(output, "%d \n", null_token);
g_free(sentence);
}
phrase_index.destroy_tokens(tokens);
/* print enter at file tail */
fprintf(output, "%d \n", null_token);
g_array_free(current_ucs4, TRUE);
free(linebuf);
fclose(input);
fclose(output);
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
setlocale(LC_ALL, "");
GError * error = NULL;
GOptionContext * context;
context = g_option_context_new("- generate k mixture model");
g_option_context_add_main_entries(context, entries, NULL);
if (!g_option_context_parse(context, &argc, &argv, &error)) {
g_print("option parsing failed:%s\n", error->message);
exit(EINVAL);
}
SystemTableInfo system_table_info;
bool retval = system_table_info.load(SYSTEM_TABLE_INFO);
if (!retval) {
fprintf(stderr, "load table.conf failed.\n");
exit(ENOENT);
}
PhraseLargeTable2 phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load(SYSTEM_PHRASE_INDEX);
phrase_table.load(chunk);
FacadePhraseIndex phrase_index;
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
if (!load_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
bigram.attach(g_k_mixture_model_filename, ATTACH_READWRITE|ATTACH_CREATE);
while ( i < argc ){
const char * filename = argv[i];
FILE * document = fopen(filename, "r");
if ( NULL == document ){
int err_saved = errno;
fprintf(stderr, "can't open file: %s.\n", filename);
fprintf(stderr, "error:%s.\n", strerror(err_saved));
exit(err_saved);
}
HashofDocument hash_of_document = g_hash_table_new
(g_direct_hash, g_direct_equal);
HashofUnigram hash_of_unigram = g_hash_table_new
(g_direct_hash, g_direct_equal);
assert(read_document(&phrase_table, &phrase_index, document,
hash_of_document, hash_of_unigram));
fclose(document);
document = NULL;
GHashTableIter iter;
gpointer key, value;
/* train the document, and convert it to k mixture model. */
g_hash_table_iter_init(&iter, hash_of_document);
while (g_hash_table_iter_next(&iter, &key, &value)) {
phrase_token_t token1 = GPOINTER_TO_UINT(key);
train_second_word(hash_of_unigram, &bigram,
hash_of_document, token1);
}
KMixtureModelMagicHeader magic_header;
assert(bigram.get_magic_header(magic_header));
magic_header.m_N ++;
assert(bigram.set_magic_header(magic_header));
post_processing_unigram(&bigram, hash_of_unigram);
/* free resources of g_hash_of_document */
g_hash_table_iter_init(&iter, hash_of_document);
while (g_hash_table_iter_next(&iter, &key, &value)) {
HashofSecondWord second_word = (HashofSecondWord) value;
g_hash_table_iter_steal(&iter);
g_hash_table_unref(second_word);
}
g_hash_table_unref(hash_of_document);
hash_of_document = NULL;
g_hash_table_unref(hash_of_unigram);
hash_of_unigram = NULL;
++i;
}
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
bool train_pi_gram = true;
const char * bigram_filename = "bigram.db";
setlocale(LC_ALL, "");
while ( i < argc ){
if ( strcmp("--help", argv[i]) == 0){
print_help();
exit(0);
}else if ( strcmp("--skip-pi-gram-training", argv[i]) == 0 ){
train_pi_gram = false;
}else if ( strcmp("--bigram-file", argv[i]) == 0){
if ( ++i >= argc ) {
print_help();
exit(EINVAL);
}
bigram_filename = argv[i];
}else{
print_help();
exit(EINVAL);
}
++i;
}
PhraseLargeTable2 phrase_table;
/* init phrase table */
MemoryChunk * chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
phrase_table.load(chunk);
FacadePhraseIndex phrase_index;
if (!load_phrase_index(&phrase_index))
exit(ENOENT);
Bigram bigram;
bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE);
PhraseTokens tokens;
memset(tokens, 0, sizeof(PhraseTokens));
phrase_index.prepare_tokens(tokens);
char* linebuf = NULL;
size_t size = 0;
phrase_token_t last_token, cur_token = last_token = 0;
while( getline(&linebuf, &size, stdin) ){
if ( feof(stdin) )
break;
linebuf[strlen(linebuf)-1] = '\0';
glong phrase_len = 0;
ucs4_t * phrase = g_utf8_to_ucs4(linebuf, -1, NULL, &phrase_len, NULL);
phrase_token_t token = null_token;
if ( 0 != phrase_len ) {
phrase_index.clear_tokens(tokens);
int result = phrase_table.search(phrase_len, phrase, tokens);
int num = get_first_token(tokens, token);
if ( !(result & SEARCH_OK) )
token = null_token;
g_free(phrase);
phrase = NULL;
}
last_token = cur_token;
cur_token = token;
/* skip null_token in second word. */
if ( null_token == cur_token )
continue;
/* training uni-gram */
phrase_index.add_unigram_frequency(cur_token, 1);
/* skip pi-gram training. */
if ( null_token == last_token ){
if ( !train_pi_gram )
continue;
last_token = sentence_start;
}
/* train bi-gram */
SingleGram * single_gram = NULL;
bigram.load(last_token, single_gram);
if ( NULL == single_gram ){
single_gram = new SingleGram;
}
guint32 freq, total_freq;
/* increase freq */
if (single_gram->get_freq(cur_token, freq))
assert(single_gram->set_freq(cur_token, freq + 1));
else
assert(single_gram->insert_freq(cur_token, 1));
/* increase total freq */
single_gram->get_total_freq(total_freq);
single_gram->set_total_freq(total_freq + 1);
bigram.store(last_token, single_gram);
delete single_gram;
}
phrase_index.destroy_tokens(tokens);
free(linebuf);
if (!save_phrase_index(&phrase_index))
exit(ENOENT);
return 0;
}
int main(int argc, char * argv[]){
int i = 1;
const char * k_mixture_model_filename = NULL;
setlocale(LC_ALL, "");
while ( i < argc ){
if ( strcmp("--help", argv[i]) == 0 ){
print_help();
exit(0);
} else if ( strcmp("--skip-pi-gram-training", argv[i]) == 0 ){
g_train_pi_gram = false;
} else if ( strcmp("--maximum-occurs-allowed", argv[i]) == 0 ){
if ( ++i >= argc ){
print_help();
exit(EINVAL);
}
g_maximum_occurs = atoi(argv[i]);
} else if ( strcmp("--maximum-increase-rates-allowed", argv[i]) == 0 ){
if ( ++i >= argc ){
print_help();
exit(EINVAL);
}
g_maximum_increase_rates = atof(argv[i]);
} else if ( strcmp("--k-mixture-model-file", argv[i]) == 0 ){
if ( ++i >= argc ){
print_help();
exit(EINVAL);
}
k_mixture_model_filename = argv[i];
} else {
break;
}
++i;
}
PhraseLargeTable2 phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("phrase_index.bin");
phrase_table.load(chunk);
FacadePhraseIndex phrase_index;
if (!load_phrase_index(&phrase_index))
exit(ENOENT);
KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
bigram.attach(k_mixture_model_filename, ATTACH_READWRITE|ATTACH_CREATE);
while ( i < argc ){
const char * filename = argv[i];
FILE * document = fopen(filename, "r");
if ( NULL == document ){
int err_saved = errno;
fprintf(stderr, "can't open file: %s.\n", filename);
fprintf(stderr, "error:%s.\n", strerror(err_saved));
exit(err_saved);
}
HashofDocument hash_of_document = g_hash_table_new
(g_direct_hash, g_direct_equal);
HashofUnigram hash_of_unigram = g_hash_table_new
(g_direct_hash, g_direct_equal);
assert(read_document(&phrase_table, &phrase_index, document,
hash_of_document, hash_of_unigram));
fclose(document);
document = NULL;
GHashTableIter iter;
gpointer key, value;
/* train the document, and convert it to k mixture model. */
g_hash_table_iter_init(&iter, hash_of_document);
while (g_hash_table_iter_next(&iter, &key, &value)) {
phrase_token_t token1 = GPOINTER_TO_UINT(key);
train_second_word(hash_of_unigram, &bigram,
hash_of_document, token1);
}
KMixtureModelMagicHeader magic_header;
assert(bigram.get_magic_header(magic_header));
magic_header.m_N ++;
assert(bigram.set_magic_header(magic_header));
post_processing_unigram(&bigram, hash_of_unigram);
/* free resources of g_hash_of_document */
g_hash_table_iter_init(&iter, hash_of_document);
while (g_hash_table_iter_next(&iter, &key, &value)) {
HashofSecondWord second_word = (HashofSecondWord) value;
g_hash_table_iter_steal(&iter);
g_hash_table_unref(second_word);
}
g_hash_table_unref(hash_of_document);
hash_of_document = NULL;
g_hash_table_unref(hash_of_unigram);
hash_of_unigram = NULL;
++i;
}
return 0;
}
int main(int argc, char * argv[]){
FILE * input = stdin;
setlocale(LC_ALL, "");
GError * error = NULL;
GOptionContext * context;
context = g_option_context_new("- generate n-gram");
g_option_context_add_main_entries(context, entries, NULL);
if (!g_option_context_parse(context, &argc, &argv, &error)) {
g_print("option parsing failed:%s\n", error->message);
exit(EINVAL);
}
SystemTableInfo system_table_info;
bool retval = system_table_info.load(SYSTEM_TABLE_INFO);
if (!retval) {
fprintf(stderr, "load table.conf failed.\n");
exit(ENOENT);
}
PhraseLargeTable2 phrase_table;
/* init phrase table */
MemoryChunk * chunk = new MemoryChunk;
chunk->load(SYSTEM_PHRASE_INDEX);
phrase_table.load(chunk);
FacadePhraseIndex phrase_index;
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
if (!load_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
Bigram bigram;
bigram.attach(bigram_filename, ATTACH_CREATE|ATTACH_READWRITE);
char* linebuf = NULL; size_t size = 0;
phrase_token_t last_token, cur_token = last_token = 0;
while( getline(&linebuf, &size, input) ){
if ( feof(input) )
break;
if ( '\n' == linebuf[strlen(linebuf) - 1] ) {
linebuf[strlen(linebuf) - 1] = '\0';
}
TAGLIB_PARSE_SEGMENTED_LINE(&phrase_index, token, linebuf);
last_token = cur_token;
cur_token = token;
/* skip null_token in second word. */
if ( null_token == cur_token )
continue;
/* training uni-gram */
phrase_index.add_unigram_frequency(cur_token, 1);
/* skip pi-gram training. */
if ( null_token == last_token ){
if ( !train_pi_gram )
continue;
last_token = sentence_start;
}
/* train bi-gram */
SingleGram * single_gram = NULL;
bigram.load(last_token, single_gram);
if ( NULL == single_gram ){
single_gram = new SingleGram;
}
guint32 freq, total_freq;
/* increase freq */
if (single_gram->get_freq(cur_token, freq))
assert(single_gram->set_freq(cur_token, freq + 1));
else
assert(single_gram->insert_freq(cur_token, 1));
/* increase total freq */
single_gram->get_total_freq(total_freq);
single_gram->set_total_freq(total_freq + 1);
bigram.store(last_token, single_gram);
delete single_gram;
}
free(linebuf);
if (!save_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
return 0;
}
int main(int argc, char * argv[]){
setlocale(LC_ALL, "");
SystemTableInfo system_table_info;
bool retval = system_table_info.load("../../data/table.conf");
if (!retval) {
fprintf(stderr, "load table.conf failed.\n");
exit(ENOENT);
}
/* init phrase table */
FacadePhraseTable2 phrase_table;
MemoryChunk * chunk = new MemoryChunk;
chunk->load("../../data/phrase_index.bin");
phrase_table.load(chunk, NULL);
const pinyin_table_info_t * phrase_files =
system_table_info.get_table_info();
/* init phrase index */
FacadePhraseIndex phrase_index;
if (!load_phrase_index(phrase_files, &phrase_index))
exit(ENOENT);
/* init bi-gram */
Bigram system_bigram;
system_bigram.attach("../../data/bigram.db", ATTACH_READONLY);
Bigram user_bigram;
gfloat lambda = system_table_info.get_lambda();
/* init phrase lookup */
PhraseLookup phrase_lookup(lambda,
&phrase_table, &phrase_index,
&system_bigram, &user_bigram);
/* try one sentence */
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;
/* check non-ucs4 characters */
const glong num_of_chars = g_utf8_strlen(linebuf, -1);
glong len = 0;
ucs4_t * sentence = g_utf8_to_ucs4(linebuf, -1, NULL, &len, NULL);
if ( len != num_of_chars ) {
fprintf(stderr, "non-ucs4 characters are not accepted.\n");
g_free(sentence);
continue;
}
try_phrase_lookup(&phrase_lookup, sentence, len);
g_free(sentence);
}
free(linebuf);
return 0;
}