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[]){
int i = 1;
const char * table_dir = ".";
setlocale(LC_ALL, "");
while ( i < argc ){
if ( strcmp("--help", argv[i]) == 0 ){
print_help();
exit(0);
} else if ( strcmp("--table-dir", argv[i]) == 0){
if ( ++i >= argc ){
print_help();
exit(EINVAL);
}
table_dir = argv[i];
}
++i;
}
/* generate pinyin index*/
pinyin_option_t options = USE_TONE;
ChewingLargeTable chewinglargetable(options);
PhraseLargeTable phraselargetable;
/* generate phrase index */
FacadePhraseIndex phrase_index;
for (size_t i = 0; i < PHRASE_INDEX_LIBRARY_COUNT; ++i) {
const pinyin_table_info_t * table_info = pinyin_phrase_files + i;
if (SYSTEM_FILE != table_info->m_file_type)
continue;
const char * tablename = table_info->m_table_filename;
gchar * filename = g_build_filename(table_dir, tablename, NULL);
FILE * tablefile = fopen(filename, "r");
if (NULL == tablefile) {
fprintf(stderr, "open %s failed!\n", tablename);
exit(ENOENT);
}
chewinglargetable.load_text(tablefile);
fseek(tablefile, 0L, SEEK_SET);
phraselargetable.load_text(tablefile);
fseek(tablefile, 0L, SEEK_SET);
phrase_index.load_text(i, tablefile);
fclose(tablefile);
g_free(filename);
}
MemoryChunk * new_chunk = new MemoryChunk;
chewinglargetable.store(new_chunk);
new_chunk->save("pinyin_index.bin");
chewinglargetable.load(new_chunk);
new_chunk = new MemoryChunk;
phraselargetable.store(new_chunk);
new_chunk->save("phrase_index.bin");
phraselargetable.load(new_chunk);
phrase_index.compact();
if (!save_phrase_index(&phrase_index))
exit(ENOENT);
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[]){
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;
}
