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

[phrases] adding a public get_data_node method for tries

Browse Source
This commit is contained in:
Al
2015-05-19 18:02:29 -04:00
parent 9d309ca9d3
commit fdf988cb27
2 changed files with 27 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
src/trie.c
View File

@@ -34,7 +34,7 @@ uint8_t DEFAULT_ALPHABET[] = {
Constructors
*/
trie_t *trie_new_empty(uint8_t *alphabet, uint32_t alphabet_size) {
static trie_t *trie_new_empty(uint8_t *alphabet, uint32_t alphabet_size) {
trie_t *self = malloc(sizeof(trie_t));
if (!self)
goto exit_no_malloc;
@@ -679,6 +679,26 @@ bool trie_add_suffix(trie_t *self, char *key, uint32_t data) {
return success;
}
trie_data_node_t trie_get_data_node(trie_t *self, trie_node_t node, char *str) {
if (node.base >= 0) {
return NULL_DATA_NODE;
}
int32_t data_index = -1*node.base;
trie_data_node_t data_node = self->data->a[data_index];
unsigned char *current_tail = self->tail->a + data_node.tail;
size_t tail_len = strlen((char *)current_tail);
char *query_tail = *str ? str + 1 : str;
size_t query_tail_len = strlen(query_tail);
int tail_match = strncmp((char *)current_tail, query_tail, query_tail_len);
if (tail_match == 0) {
return data_node;
} else {
return NULL_DATA_NODE;
}
}
uint32_t trie_get_prefix_from_index(trie_t *self, char *key, size_t len, uint32_t i) {
if (key == NULL) return NULL_NODE_ID;
@@ -736,17 +756,9 @@ uint32_t trie_get_from_index(trie_t *self, char *word, size_t len, uint32_t i) {
}
if (node.check == node_id && node.base < 0) {
int32_t data_index = -1*node.base;
trie_data_node_t data_node = self->data->a[data_index];
unsigned char *current_tail = self->tail->a + data_node.tail;
trie_data_node_t data_node = trie_get_data_node(self, node, (char *)ptr);
size_t tail_len = strlen((char *)current_tail);
char *query_tail = (char *)(*ptr ? ptr + 1 : ptr);
size_t query_tail_len = strlen((char *)query_tail);
int tail_match = strncmp((char *)current_tail, query_tail, query_tail_len);
if (tail_match == 0) {
if (data_node.tail != 0) {
return next_id;
} else {
return NULL_NODE_ID;

4
src/trie.h
View File

@@ -55,6 +55,8 @@ typedef struct trie_data_node {
uint32_t data;
} trie_data_node_t;
#define NULL_DATA_NODE (trie_data_node_t){0, 0};
VECTOR_INIT(trie_node_array, trie_node_t)
VECTOR_INIT(trie_data_array, trie_data_node_t)
@@ -83,6 +85,8 @@ void trie_set_check(trie_t *self, uint32_t index, int32_t check);
trie_node_t trie_get_root(trie_t *self);
trie_node_t trie_get_free_list(trie_t *self);
trie_data_node_t trie_get_data_node(trie_t *self, trie_node_t node, char *str);
uint32_t trie_add_transition(trie_t *self, uint32_t node_id, unsigned char c);
void trie_make_room_for(trie_t *self, uint32_t next_id);