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

[fix] tokenized trie search when falling off the trie at the start of a valid phrase

Browse Source
This commit is contained in:
Al
2016-06-21 15:48:43 -04:00
parent 935a31df07
commit 3ff2f726d0
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/trie_search.c
View File

@@ -275,8 +275,8 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
if (node.check != last_node_id && last_node.base >= 0) { if (node.check != last_node_id && last_node.base >= 0) {
log_debug("Fell off trie. last_node_id=%d and node.check=%d\n", last_node_id, node.check); log_debug("Fell off trie. last_node_id=%d and node.check=%d\n", last_node_id, node.check);
node_id = start_node_id; node_id = last_node_id = start_node_id;
node = trie_get_node(self, node_id); node = last_node = trie_get_node(self, node_id);
break; break;
} else if (node.base < 0) { } else if (node.base < 0) {
log_debug("Searching tail at index %d\n", i); log_debug("Searching tail at index %d\n", i);
@@ -439,6 +439,7 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
if (*phrases == NULL) { if (*phrases == NULL) {
*phrases = phrase_array_new_size(1); *phrases = phrase_array_new_size(1);
} }
log_debug("adding phrase, last_match_index=%d\n", last_match_index);
phrase_array_push(*phrases, (phrase_t){phrase_start, last_match_index - phrase_start + 1, data}); phrase_array_push(*phrases, (phrase_t){phrase_start, last_match_index - phrase_start + 1, data});
} }