int main(int argc, char * argv[]){
PhraseLargeTable largetable;
FILE * gbfile = fopen("../../data/gb_char.table", "r");
if ( gbfile == NULL ) {
fprintf(stderr, "open gb_char.table failed!\n");
return 1;
}
largetable.load_text(gbfile);
fclose(gbfile);
FILE * gbkfile = fopen("../../data/gbk_char.table", "r");
if (gbkfile == NULL ) {
fprintf(stderr, "open gbk_char.table failed!\n");
return 1;
}
largetable.load_text(gbkfile);
fclose(gbkfile);
MemoryChunk * chunk = new MemoryChunk;
largetable.store(chunk);
largetable.load(chunk);
char * linebuf = NULL;
size_t size = 0;
while( getline(&linebuf, &size, stdin) ){
linebuf[strlen(linebuf) - 1] = '\0';
if ( strcmp ( linebuf, "quit" ) == 0)
break;
glong phrase_len = g_utf8_strlen(linebuf, -1);
utf16_t * new_phrase = g_utf8_to_utf16(linebuf, -1, NULL, NULL, NULL);
phrase_token_t token;
guint32 start = record_time();
for ( size_t i = 0; i < bench_times; ++i){
largetable.search(phrase_len, new_phrase, token);
}
print_time(start, bench_times);
int retval = largetable.search(phrase_len, new_phrase, token);
if ( retval & SEARCH_OK )
printf("%s:\t%d\n", linebuf, token);
else
printf("phrase %s not found.\n", linebuf);
g_free(new_phrase);
}
if ( linebuf )
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[]){
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[]) {
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*/
PinyinCustomSettings custom;
PinyinLargeTable pinyinlargetable(&custom);
PhraseLargeTable phraselargetable;
/* generate phrase index */
FacadePhraseIndex phrase_index;
gchar * filename = g_build_filename(table_dir, "gb_char.table", NULL);
FILE * gbfile = fopen(filename, "r");
g_free(filename);
if ( gbfile == NULL) {
fprintf(stderr, "open gb_char.table failed!");
exit(ENOENT);
}
pinyinlargetable.load_text(gbfile);
fseek(gbfile, 0L, SEEK_SET);
phraselargetable.load_text(gbfile);
fseek(gbfile, 0L, SEEK_SET);
phrase_index.load_text(1, gbfile);
fclose(gbfile);
filename = g_build_filename(table_dir, "gbk_char.table", NULL);
FILE * gbkfile = fopen(filename, "r");
g_free(filename);
if ( gbkfile == NULL) {
fprintf(stderr, "open gbk_char.table failed!");
exit(ENOENT);
}
pinyinlargetable.load_text(gbkfile);
fseek(gbkfile, 0L, SEEK_SET);
phraselargetable.load_text(gbkfile);
fseek(gbkfile, 0L, SEEK_SET);
phrase_index.load_text(2, gbkfile);
fclose(gbkfile);
MemoryChunk * new_chunk = new MemoryChunk;
pinyinlargetable.store(new_chunk);
new_chunk->save("pinyin_index.bin");
pinyinlargetable.load(new_chunk);
new_chunk = new MemoryChunk;
phraselargetable.store(new_chunk);
new_chunk->save("phrase_index.bin");
phraselargetable.load(new_chunk);
phrase_index.compat();
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;
}