};
/* For Phrase Index */
-enum PhraseIndexError{
+enum PhraseIndexResult{
ERROR_OK = 0, /* operate ok */
ERROR_NO_SUB_PHRASE_INDEX, /* sub phrase index is not loaded */
ERROR_NO_ITEM, /* item has a null slot */
ERROR_INTEGER_OVERFLOW /* integer is overflowed */
};
+/* For N-gram */
+enum ATTACH_FLAG{
+ ATTACH_READONLY = 1,
+ ATTACH_READWRITE = 0x1 << 1,
+ ATTACH_CREATE = 0x1 << 2,
+};
+
/*
* n-gram Definition
* no B parameter(there are duplicated items in uni-gram and bi-gram)
}
/* attach berkeley db on filesystem for training purpose. */
- bool attach(const char * dbfile){
+ bool attach(const char * dbfile, guint32 flags){
reset();
+ u_int32_t db_flags = 0;
+
+ if ( flags & ATTACH_READONLY )
+ db_flags |= DB_RDONLY;
+ if ( flags & ATTACH_READWRITE )
+ assert( !(flags & ATTACH_READONLY ) );
+
if ( !dbfile )
return false;
int ret = db_create(&m_db, NULL, 0);
if ( ret != 0 )
assert(false);
- ret = m_db->open(m_db, NULL, dbfile, NULL, DB_HASH, 0, 0644);
- if ( ret != 0 ) {
+ ret = m_db->open(m_db, NULL, dbfile, NULL, DB_HASH, db_flags, 0644);
+ if ( ret != 0 && (flags & ATTACH_CREATE) ) {
+ db_flags |= DB_CREATE;
/* Create database file here, and write the signature. */
- ret = m_db->open(m_db, NULL, dbfile, NULL, DB_HASH, DB_CREATE, 0644);
+ ret = m_db->open(m_db, NULL, dbfile, NULL, DB_HASH, db_flags, 0644);
if ( ret != 0 )
return false;
namespace pinyin{
-enum {
- ATTACH_READONLY = 1,
- ATTACH_READWRITE = 0x1 << 1,
- ATTACH_CREATE = 0x1 << 2,
-};
-
class Bigram;
/* Note:
assert(freq == total_freq);
FlexibleBigram<guint32, guint32, guint32> bigram("TEST");
- assert(bigram.attach("/tmp/training.db"));
+ assert(bigram.attach("/tmp/training.db", ATTACH_READWRITE|ATTACH_CREATE));
bigram.store(1, &single_gram);
assert(single_gram.insert_array_item(5, 8));
assert(single_gram.remove_array_item(1, freq));
/* TODO: magic header signature check here. */
KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
- bigram.attach(bigram_filename);
+ bigram.attach(bigram_filename, ATTACH_READONLY);
KMixtureModelBigram deleted_bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
- deleted_bigram.attach(deleted_bigram_filename);
+ deleted_bigram.attach(deleted_bigram_filename, ATTACH_READONLY);
GArray * deleted_items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t));
deleted_bigram.get_all_items(deleted_items);
/* TODO: magic header signature check here. */
KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
- bigram.attach(bigram_filename);
+ bigram.attach(bigram_filename, ATTACH_READWRITE);
KMixtureModelMagicHeader magic_header;
bigram.get_magic_header(magic_header);