void gen_phrase_file(const char * outfilename, int phrase_index){
FILE * outfile = fopen(outfilename, "w");
if (NULL == outfile ) {
fprintf(stderr, "Can't write file %s.\n", outfilename);
exit(ENOENT);
}
phrase_token_t token = 1;
char pinyin_buffer[4096];
//phrase length
for ( size_t i = 1; i < MAX_PHRASE_LENGTH + 1; ++i){
GArray * item_array = g_item_array[i];
//item array
for( size_t m = 0; m < item_array->len; ++m){
item* oneitem = & g_array_index(item_array, item, m);
phrase_item * phrase = oneitem->phrase;
GArray * pinyin_and_freqs = oneitem->pinyin_and_freq_array;
const char * phrase_buffer = g_ucs4_to_utf8(phrase->uniphrase,
phrase->length,
NULL, NULL, NULL);
//each pinyin
for( size_t n = 0 ; n < pinyin_and_freqs->len; ++n){
pinyin_and_freq_item * pinyin_and_freq = &g_array_index(pinyin_and_freqs, pinyin_and_freq_item, n);
GArray * pinyin = pinyin_and_freq->pinyin;
PinyinKey * key = &g_array_index(pinyin, PinyinKey, 0);
strcpy(pinyin_buffer,key->get_key_string());
for (size_t k = 1; k < pinyin->len; ++k){
strcat(pinyin_buffer, "'");
PinyinKey * key = &g_array_index(pinyin, PinyinKey, k);
strcat(pinyin_buffer, key->get_key_string ());
}
guint32 freq = pinyin_and_freq -> freq;
if ( freq < 3 )
freq = 3;
fprintf( outfile, "%s\t%s\t%d\t%d\n",
pinyin_buffer, phrase_buffer,
PHRASE_INDEX_MAKE_TOKEN(phrase_index, token),
freq);
}
token++;
}
}
fclose(outfile);
}
int main (int argc, char * argv [])
{
NullPinyinValidator validator;
PinyinKeyVector keys;
PinyinKeyPosVector poses;
PinyinCustomSettings custom;
PinyinParser *parser = 0;
//PinyinTable table;
const char *tablefile = "../data/pinyin-table.txt";
keys = g_array_new(FALSE, FALSE, sizeof( PinyinKey));
poses = g_array_new(FALSE, FALSE, sizeof( PinyinKeyPos));
int i = 0;
while (i<argc) {
if (++i >= argc) break;
if ( !strcmp("-h", argv [i]) || !strcmp ("--help", argv [i]) ) {
print_help ();
return 0;
}
if ( !strcmp("-i", argv [i]) ) {
custom.set_use_incomplete (true);
continue;
}
if ( !strcmp("-p", argv [i]) ) {
if (++i >= argc) {
fprintf(stderr, "No argument for option %s.\n", argv [i-1]);
return -1;
}
if (!strcmp (argv[i], "sp") || !strcmp (argv[i], "sp-default"))
parser = new PinyinShuangPinParser ();
else if (!strcmp (argv[i], "sp-stone"))
parser = new PinyinShuangPinParser (SHUANG_PIN_STONE);
else if (!strcmp (argv[i], "sp-zrm"))
parser = new PinyinShuangPinParser (SHUANG_PIN_ZRM);
else if (!strcmp (argv[i], "sp-ms"))
parser = new PinyinShuangPinParser (SHUANG_PIN_MS);
else if (!strcmp (argv[i], "sp-ziguang"))
parser = new PinyinShuangPinParser (SHUANG_PIN_ZIGUANG);
else if (!strcmp (argv[i], "sp-abc"))
parser = new PinyinShuangPinParser (SHUANG_PIN_ABC);
else if (!strcmp (argv[i], "sp-liushi"))
parser = new PinyinShuangPinParser (SHUANG_PIN_LIUSHI);
if (!strcmp (argv[i], "zy") || !strcmp (argv[i], "zy-standard") || !strcmp (argv[i], "zy-default"))
parser = new PinyinZhuYinParser ();
else if (!strcmp (argv[i], "zy-hsu"))
parser = new PinyinZhuYinParser (ZHUYIN_HSU);
else if (!strcmp (argv[i], "zy-ibm"))
parser = new PinyinZhuYinParser (ZHUYIN_IBM);
else if (!strcmp (argv[i], "zy-gin-yieh"))
parser = new PinyinZhuYinParser (ZHUYIN_GIN_YIEH);
else if (!strcmp (argv[i], "zy-et"))
parser = new PinyinZhuYinParser (ZHUYIN_ET);
else if (!strcmp (argv[i], "zy-et26"))
parser = new PinyinZhuYinParser (ZHUYIN_ET26);
else if (!strcmp (argv[i], "zy-zhuyin"))
parser = new PinyinZhuYinParser (ZHUYIN_ZHUYIN);
else {
fprintf(stderr, "Unknown Parser:%s.\n", argv[i]);
print_help();
exit(EINVAL);
}
continue;
}
if (!strcmp("-f", argv [i])) {
if (++i >= argc) {
fprintf(stderr, "No argument for option %s.\n", argv [i-1]);
return -1;
}
tablefile = argv [i];
continue;
}
fprintf(stderr, "Invalid option: %s.\n", argv [i]);
return -1;
};
if (!parser) parser = new PinyinDefaultParser ();
char * line = NULL;
size_t len = 0;
while (1) {
printf("Input:"); fflush(stdout);
getline(&line, &len, stdin);
if (!strncmp (line, "quit", 4)) break;
int len = parser->parse (validator, keys, poses,(const char *) line);
printf("Parsed %d chars, %d keys:\n", len, keys->len);
for (size_t i=0; i < keys->len; ++i){
PinyinKey * key = &g_array_index(keys, PinyinKey, i);
printf("%s ", key->get_key_string ());
}
printf("\n");
for ( size_t i=0; i < poses->len; ++i){
PinyinKeyPos * pos = &g_array_index(poses, PinyinKeyPos, i);
printf("%d %ld ", pos->get_pos(), pos->get_length());
}
printf("\n");
for (size_t i=0; i < keys->len; ++i){
PinyinKey * key = &g_array_index(keys, PinyinKey, i);
printf("%s ", key->get_key_zhuyin_string ());
}
printf("\n");
}
if (line)
free(line);
return 0;
}