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

[phrases] trie_get_prefix_* methods return a struct including tail position

Browse Source
This commit is contained in:
Al
2015-05-21 05:38:18 -04:00
parent 395fbcb8b5 e84f3d93d2
commit 91ccdf6f7b
2 changed files with 20 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
src/trie.c
View File

@@ -695,10 +695,9 @@ inline trie_data_node_t trie_get_data_node(trie_t *self, trie_node_t node) {
return data_node; return data_node;
} }
uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_t i, size_t *tail_pos) { trie_prefix_result_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_t i) {
if (key == NULL) { if (key == NULL) {
*tail_pos = 0; return (trie_prefix_result_t){NULL_NODE_ID, 0};
return NULL_NODE_ID;
} }
unsigned char *ptr = (unsigned char *)key; unsigned char *ptr = (unsigned char *)key;
@@ -706,8 +705,7 @@ uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_
uint32_t node_id = i; uint32_t node_id = i;
trie_node_t node = trie_get_node(self, i); trie_node_t node = trie_get_node(self, i);
if (node.base == NULL_NODE_ID) { if (node.base == NULL_NODE_ID) {
*tail_pos = 0; return (trie_prefix_result_t){NULL_NODE_ID, 0};
return NULL_NODE_ID;
} }
uint32_t next_id = NULL_NODE_ID; uint32_t next_id = NULL_NODE_ID;
@@ -719,7 +717,7 @@ uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_
node = trie_get_node(self, next_id); node = trie_get_node(self, next_id);
if (node.check != node_id) { if (node.check != node_id) {
return NULL_NODE_ID; return (trie_prefix_result_t){NULL_NODE_ID, 0};
} }
if (node.base < 0) break; if (node.base < 0) break;
@@ -733,25 +731,24 @@ uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_
size_t query_len = strlen(query_tail); size_t query_len = strlen(query_tail);
if (data_node.tail != 0 && trie_compare_tail(self, query_tail, query_len, data_node.tail)) { if (data_node.tail != 0 && trie_compare_tail(self, query_tail, query_len, data_node.tail)) {
*tail_pos = query_len; return (trie_prefix_result_t){next_id, query_len};
return next_id;
} else { } else {
*tail_pos = 0; return (trie_prefix_result_t){NULL_NODE_ID, 0};
return NULL_NODE_ID;
} }
} }
return next_id; return (trie_prefix_result_t){next_id, 0};
} }
uint32_t trie_get_prefix_len(trie_t *self, char *key, size_t len, size_t *tail_pos) { trie_prefix_result_t trie_get_prefix_len(trie_t *self, char *key, size_t len) {
return trie_get_prefix_from_index(self, key, len, ROOT_NODE_ID, tail_pos); return trie_get_prefix_from_index(self, key, len, ROOT_NODE_ID);
} }
uint32_t trie_get_prefix(trie_t *self, char *key, size_t *tail_pos) { trie_prefix_result_t trie_get_prefix(trie_t *self, char *key) {
return trie_get_prefix_from_index(self, key, strlen(key), ROOT_NODE_ID, tail_pos); return trie_get_prefix_from_index(self, key, strlen(key), ROOT_NODE_ID);
} }
uint32_t trie_get_from_index(trie_t *self, char *word, size_t len, uint32_t i) { uint32_t trie_get_from_index(trie_t *self, char *word, size_t len, uint32_t i) {

11
src/trie.h
View File

@@ -105,9 +105,14 @@ uint32_t trie_get_from_index(trie_t *self, char *word, size_t len, uint32_t i);
uint32_t trie_get_len(trie_t *self, char *word, size_t len); uint32_t trie_get_len(trie_t *self, char *word, size_t len);
uint32_t trie_get(trie_t *self, char *word); uint32_t trie_get(trie_t *self, char *word);
uint32_t trie_get_prefix(trie_t *self, char *key, size_t *tail_pos); typedef struct trie_prefix_result {
uint32_t trie_get_prefix_len(trie_t *self, char *key, size_t len, size_t *tail_pos); uint32_t node_id;
uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_t i, size_t *tail_pos); size_t tail_pos;
} trie_prefix_result_t;
trie_prefix_result_t trie_get_prefix(trie_t *self, char *key);
trie_prefix_result_t trie_get_prefix_len(trie_t *self, char *key, size_t len);
trie_prefix_result_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_t i);
void trie_print(trie_t *self); void trie_print(trie_t *self);