tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[phrases] adding _from_index_get_prefix_char/_from_index_get_suffix_char methods

Browse Source
This commit is contained in:
Al
2015-08-09 03:38:28 -04:00
parent 2eb67ad850
commit 11a9881988
2 changed files with 30 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
src/trie_search.c
View File

@@ -505,18 +505,25 @@ phrase_t trie_search_suffixes_from_index(trie_t *self, char *word, size_t len, u
return (phrase_t) {phrase_start, phrase_len, value}; return (phrase_t) {phrase_start, phrase_len, value};
} }
inline phrase_t trie_search_suffixes(trie_t *self, char *word, size_t len) { inline phrase_t trie_search_suffixes_from_index_get_suffix_char(trie_t *self, char *word, size_t len, uint32_t start_node_id) {
trie_node_t root_node = trie_get_root(self); if (word == NULL || len == 0) return NULL_PHRASE;
uint32_t node_id = trie_get_transition_index(self, root_node, TRIE_SUFFIX_CHAR); trie_node_t node = trie_get_node(self, start_node_id);
trie_node_t node = trie_get_node(self, node_id); uint32_t node_id = trie_get_transition_index(self, node, TRIE_SUFFIX_CHAR);
node = trie_get_node(self, node_id);
if (node.check != ROOT_NODE_ID) { if (node.check != start_node_id) {
return (phrase_t){0, 0, 0}; return NULL_PHRASE;
} }
return trie_search_suffixes_from_index(self, word, len, node_id); return trie_search_suffixes_from_index(self, word, len, node_id);
} }
inline phrase_t trie_search_suffixes(trie_t *self, char *word, size_t len) {
if (word == NULL || len == 0) return NULL_PHRASE;
return trie_search_suffixes_from_index_get_suffix_char(self, word, len, ROOT_NODE_ID);
}
phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id) { phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id) {
log_debug("Call to trie_search_prefixes_from_index\n"); log_debug("Call to trie_search_prefixes_from_index\n");
uint32_t node_id = start_node_id, last_node_id = node_id; uint32_t node_id = start_node_id, last_node_id = node_id;
@@ -625,21 +632,26 @@ phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, u
ptr++; ptr++;
} }
if (phrase_len == 0) return (phrase_t){0, 0, 0}; if (phrase_len == 0) return NULL_PHRASE;
return (phrase_t) {phrase_start, phrase_len, value}; return (phrase_t) {phrase_start, phrase_len, value};
} }
inline phrase_t trie_search_prefixes(trie_t *self, char *word, size_t len) { inline phrase_t trie_search_prefixes_from_index_get_prefix_char(trie_t *self, char *word, size_t len, uint32_t start_node_id) {
trie_node_t root_node = trie_get_root(self); trie_node_t node = trie_get_node(self, start_node_id);
uint32_t node_id = trie_get_transition_index(self, root_node, TRIE_PREFIX_CHAR); uint32_t node_id = trie_get_transition_index(self, node, TRIE_PREFIX_CHAR);
trie_node_t node = trie_get_node(self, node_id); node = trie_get_node(self, node_id);
if (node.check != ROOT_NODE_ID) { if (node.check != start_node_id) {
return (phrase_t){0, 0, 0}; return NULL_PHRASE;
} }
return trie_search_prefixes_from_index(self, word, len, node_id); return trie_search_prefixes_from_index(self, word, len, node_id);
} }
inline phrase_t trie_search_prefixes(trie_t *self, char *word, size_t len) {
if (word == NULL || len == 0) return NULL_PHRASE;
return trie_search_prefixes_from_index_get_prefix_char(self, word, len, ROOT_NODE_ID);
}

6
src/trie_search.h
View File

@@ -25,15 +25,19 @@ typedef struct phrase {
VECTOR_INIT(phrase_array, phrase_t) VECTOR_INIT(phrase_array, phrase_t)
#define NULL_PHRASE (phrase_t){0, 0, 0};
phrase_array *trie_search(trie_t *self, char *text); phrase_array *trie_search(trie_t *self, char *text);
phrase_array *trie_search_from_index(trie_t *self, char *text, uint32_t start_node_id); phrase_array *trie_search_from_index(trie_t *self, char *text, uint32_t start_node_id);
phrase_array *trie_search_tokens(trie_t *self, char *str, token_array *tokens); phrase_array *trie_search_tokens(trie_t *self, char *str, token_array *tokens);
phrase_array *trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens, uint32_t start_node_id); phrase_array *trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens, uint32_t start_node_id);
phrase_t trie_search_suffixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id); phrase_t trie_search_suffixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id);
phrase_t trie_search_suffixes_from_index_get_suffix_char(trie_t *self, char *word, size_t len, uint32_t start_node_id);
phrase_t trie_search_suffixes(trie_t *self, char *word, size_t len); phrase_t trie_search_suffixes(trie_t *self, char *word, size_t len);
phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id); phrase_t trie_search_prefixes_from_index(trie_t *self, char *word, size_t len, uint32_t start_node_id);
phrase_t trie_search_prefixes_from_index_get_prefix_char(trie_t *self, char *word, size_t len, uint32_t start_node_id);
phrase_t trie_search_prefixes(trie_t *self, char *word, size_t len); phrase_t trie_search_prefixes(trie_t *self, char *word, size_t len);
#endif #endif