void dictionary_delete(Dictionary dict)
{
if (!dict) return;
if (verbosity > 0) {
prt_error("Info: Freeing dictionary %s", dict->name);
}
#ifdef USE_CORPUS
lg_corpus_delete(dict->corpus);
#endif
if (dict->affix_table != NULL) {
affix_list_delete(dict->affix_table);
dictionary_delete(dict->affix_table);
}
spellcheck_destroy(dict->spell_checker);
connector_set_delete(dict->unlimited_connector_set);
if (dict->close) dict->close(dict);
pp_knowledge_close(dict->base_knowledge);
pp_knowledge_close(dict->hpsg_knowledge);
string_set_delete(dict->string_set);
free_regexs(dict->regex_root);
#ifdef USE_ANYSPLIT
free_anysplit(dict);
#endif
free_dictionary(dict);
xfree(dict, sizeof(struct Dictionary_s));
object_open(NULL, NULL, NULL); /* Free the directory path cache */
}
int
commit_read (state_t *ur, struct commit *commit, const unsigned char sha1[20])
{
int fd = -1;
char * buf = NULL;
int len;
struct tm tm;
char *cp;
commit->msg = NULL;
if (commit->alive) goto error;
if ((fd = object_open (ur, sha1)) < -1) goto error;
buf = readline (fd);
if (buf == NULL) goto error;
cp = strptime (buf, "%a %b %d %T %Y", &tm);
commit->ctime = mktime (&tm);
free (buf); buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
hex_to_sha1 (buf, commit->parent_sha1_1);
free (buf); buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
hex_to_sha1 (buf, commit->parent_sha1_2);
free (buf); buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
commit->object_type = atoi (buf);
free (buf); buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
hex_to_sha1 (buf, commit->object_sha1);
free (buf); buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
len = atoi (buf);
free (buf); buf = NULL;
commit->msg = (char *) malloc (len);
readn (fd, commit->msg, len);
close (fd);
commit->alive = true;
return 0;
error:
if (commit->msg != NULL) free (commit->msg);
return -1;
}
Dictionary dictionary_create_from_db(const char *lang)
{
char *dbname;
const char * t;
Dictionary dict;
Dict_node *dict_node;
dict = (Dictionary) xalloc(sizeof(struct Dictionary_s));
memset(dict, 0, sizeof(struct Dictionary_s));
dict->version = NULL;
dict->num_entries = 0;
dict->affix_table = NULL;
dict->regex_root = NULL;
/* Language and file-name stuff */
dict->string_set = string_set_create();
dict->lang = lang;
t = strrchr (lang, '/');
if (t) dict->lang = string_set_add(t+1, dict->string_set);
/* To disable spell-checking, just set the checker to NULL */
dict->spell_checker = spellcheck_create(dict->lang);
dict->base_knowledge = NULL;
dict->hpsg_knowledge = NULL;
dbname = join_path (lang, "dict.db");
dict->name = string_set_add(dbname, dict->string_set);
free(dbname);
/* Set up the database */
dict->db_handle = object_open(dict->name, db_open, NULL);
dict->lookup_list = db_lookup_list;
dict->free_lookup = db_free_llist;
dict->lookup = db_lookup;
dict->close = db_close;
/* Misc remaining common (generic) dict setup work */
dict->left_wall_defined = boolean_dictionary_lookup(dict, LEFT_WALL_WORD);
dict->right_wall_defined = boolean_dictionary_lookup(dict, RIGHT_WALL_WORD);
dict->empty_word_defined = boolean_dictionary_lookup(dict, EMPTY_WORD_MARK);
dict->unknown_word_defined = boolean_dictionary_lookup(dict, UNKNOWN_WORD);
dict->use_unknown_word = true;
dict_node = dictionary_lookup_list(dict, UNLIMITED_CONNECTORS_WORD);
if (dict_node != NULL) {
dict->unlimited_connector_set = connector_set_create(dict_node->exp);
} else {
dict->unlimited_connector_set = NULL;
}
free_lookup_list(dict, dict_node);
return dict;
}
Dictionary dictionary_create_from_db(const char *lang)
{
char *dbname;
const char * t;
Dictionary dict;
Dict_node *dict_node;
dict = (Dictionary) xalloc(sizeof(struct Dictionary_s));
memset(dict, 0, sizeof(struct Dictionary_s));
/* Language and file-name stuff */
dict->string_set = string_set_create();
t = strrchr (lang, '/');
t = (NULL == t) ? lang : t+1;
dict->lang = string_set_add(t, dict->string_set);
lgdebug(D_USER_FILES, "Debug: Language: %s\n", dict->lang);
/* To disable spell-checking, just set the checker to NULL */
dict->spell_checker = spellcheck_create(dict->lang);
#if defined HAVE_HUNSPELL || defined HAVE_ASPELL
if (NULL == dict->spell_checker)
prt_error("Info: Spell checker disabled.");
#endif
dict->base_knowledge = NULL;
dict->hpsg_knowledge = NULL;
dbname = join_path (lang, "dict.db");
dict->name = string_set_add(dbname, dict->string_set);
free(dbname);
/* Set up the database */
dict->db_handle = object_open(dict->name, db_open, NULL);
dict->lookup_list = db_lookup_list;
dict->free_lookup = db_free_llist;
dict->lookup = db_lookup;
dict->close = db_close;
/* Misc remaining common (generic) dict setup work */
dict->left_wall_defined = boolean_dictionary_lookup(dict, LEFT_WALL_WORD);
dict->right_wall_defined = boolean_dictionary_lookup(dict, RIGHT_WALL_WORD);
dict->empty_word_defined = boolean_dictionary_lookup(dict, EMPTY_WORD_MARK);
dict->unknown_word_defined = boolean_dictionary_lookup(dict, UNKNOWN_WORD);
dict->use_unknown_word = true;
dict_node = dictionary_lookup_list(dict, UNLIMITED_CONNECTORS_WORD);
if (dict_node != NULL)
dict->unlimited_connector_set = connector_set_create(dict_node->exp);
free_lookup_list(dict, dict_node);
return dict;
}
void add_one_object( const char *name )
{
OBJECT_FILE *file;
OBJECT_FILE **save;
int available;
file = object_open( name );
if (file)
{
save = array_alloc( &link_array, &available );
*save = file;
array_update( &link_array, 1 );
}
}
Dictionary dictionary_create_lang(const char * lang)
{
Dictionary dictionary = NULL;
object_open(NULL, NULL, NULL); /* Invalidate the directory path cache */
/* If an sql database exists, try to read that. */
if (check_db(lang))
{
dictionary = dictionary_create_from_db(lang);
}
/* Fallback to a plain-text dictionary */
if (NULL == dictionary)
{
dictionary = dictionary_create_from_file(lang);
}
return dictionary;
}
FILE *dictopen(const char *filename, const char *how)
{
return object_open(filename, dict_file_open, how);
}
int
tree_read (state_t *ur, struct tree *tree, unsigned char sha1[20])
{
int fd = -1, i;
int blob_cnt = 0;
int branch_cnt = 0;
char * buf = NULL;
char *name = NULL, *branch = NULL;
unsigned char commit[20];
struct blob_tree_entry *blob_entry = NULL;
struct branch_tree_entry *branch_entry = NULL;
tree_init (tree);
if ((fd = object_open (ur, sha1)) < -1)
goto error;
buf = readline (fd);
if (buf == NULL) goto error;
blob_cnt = atoi (buf);
free (buf); buf = NULL;
for (i = 0; i < blob_cnt; i ++)
{
buf = readline (fd);
if (buf == NULL) goto error;
name = buf;
buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
hex_to_sha1 (buf, commit);
free (buf); buf = NULL;
blob_entry = (struct blob_tree_entry *) malloc (sizeof (struct blob_tree_entry));
if (blob_entry == NULL)
goto error;
blob_entry->name = name;
memcpy (blob_entry->commit, commit, 20);
list_push_back (&tree->blob_entries, &blob_entry->elem);
name = NULL;
blob_entry = NULL;
}
buf = readline (fd);
if (buf == NULL) goto error;
branch_cnt = atoi (buf);
free (buf); buf = NULL;
for (i = 0; i < branch_cnt; i ++)
{
buf = readline (fd);
if (buf == NULL) goto error;
name = buf;
buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
branch = buf;
buf = NULL;
buf = readline (fd);
if (buf == NULL) goto error;
hex_to_sha1 (buf, commit);
free (buf); buf = NULL;
branch_entry = (struct branch_tree_entry *) malloc (sizeof (struct branch_tree_entry));
if (branch_entry == NULL)
goto error;
branch_entry->name = name;
branch_entry->branch = branch;
memcpy (branch_entry->commit, commit, 20);
list_push_back (&tree->branch_entries, &branch_entry->elem);
name = NULL;
branch = NULL;
branch_entry = NULL;
}
close (fd);
return 0;
error:
if (name != NULL) free (name);
if (branch != NULL) free (branch);
if (blob_entry != NULL) free (blob_entry);
if (branch_entry != NULL) free (branch_entry);
tree_destroy (tree);
return -1;
}
/**
* Initialize the corpus statistics subsystem.
*/
Corpus * lg_corpus_new(void)
{
int rc;
Corpus *c = (Corpus *) malloc(sizeof(Corpus));
c->rank_query = NULL;
c->sense_query = NULL;
c->errmsg = NULL;
c->dbname = NULL;
/* dbname = "/link-grammar/data/en/sql/disjuncts.db"; */
#ifdef _WIN32
#define DBNAME "sql\\disjuncts.db"
#else
#define DBNAME "sql/disjuncts.db"
#endif
c->dbconn = object_open(DBNAME, db_file_open, c);
if (NULL == c->dbconn)
{
if (SQLITE_CANTOPEN == c->rc)
{
prt_error("Warning: File not found: %s\n"
"\tWas looking for: " DBNAME,
c->errmsg);
}
else
{
prt_error("Warning: Can't open database: %s\n"
"\tWas looking for: " DBNAME,
c->errmsg);
}
return c;
}
/* Now prepare the statements we plan to use */
rc = sqlite3_prepare_v2(c->dbconn,
"SELECT log_cond_probability FROM Disjuncts "
"WHERE inflected_word = ? AND disjunct = ?;",
-1, &c->rank_query, NULL);
if (rc != SQLITE_OK)
{
prt_error("Error: Can't prepare the ranking statment: %s\n",
sqlite3_errmsg(c->dbconn));
}
/* Results are returned in sorted order .. would it be faster
* to sort locally? Don't know ... */
rc = sqlite3_prepare_v2(c->dbconn,
"SELECT word_sense, log_cond_probability FROM DisjunctSenses "
"WHERE inflected_word = ? AND disjunct = ? "
"ORDER BY log_cond_probability ASC;",
-1, &c->sense_query, NULL);
if (rc != SQLITE_OK)
{
prt_error("Error: Can't prepare the sense statment: %s\n",
sqlite3_errmsg(c->dbconn));
}
prt_error("Info: Corpus statistics database found at %s\n", c->dbname);
return c;
}
/**
* Open a file in the dictionary search path.
* Experimental API (may be unstable).
* @param filename Filename to be opened.
* @return FILE pointer, or NULL if the file was no found.
*/
FILE *linkgrammar_open_data_file(const char *filename)
{
object_open(NULL, NULL, NULL); /* Invalidate the directory path cache */
return object_open(filename, data_file_open, "r");
}
