[phrases] Case where trie search finds a match, makes progress beyond the next token but has to fall back. Adding trie search test case

This commit is contained in:
Al
2016-02-08 01:07:56 -05:00
parent 3701d8380f
commit 9ac0379a65
5 changed files with 82 additions and 17 deletions

View File

@@ -84,7 +84,7 @@ static void add_normalized_strings_token(cstring_array *strings, char *str, toke
} }
} }
if (is_numeric_token(token.type) && options.split_alpha_from_numeric) { if (is_numeric_token(token.type) && options.split_alpha_from_numeric && numeric_starts_with_alpha(str, token)) {
normalize_token_options |= NORMALIZE_TOKEN_SPLIT_ALPHA_FROM_NUMERIC; normalize_token_options |= NORMALIZE_TOKEN_SPLIT_ALPHA_FROM_NUMERIC;
normalize_token(strings, str, token, normalize_token_options); normalize_token(strings, str, token, normalize_token_options);
normalize_token_options ^= NORMALIZE_TOKEN_SPLIT_ALPHA_FROM_NUMERIC; normalize_token_options ^= NORMALIZE_TOKEN_SPLIT_ALPHA_FROM_NUMERIC;
@@ -699,6 +699,7 @@ static void expand_alternative(cstring_array *strings, khash_t(str_set) *unique_
string_tree_t *token_tree = string_tree_new_size(len); string_tree_t *token_tree = string_tree_new_size(len);
add_normalized_strings_tokenized(token_tree, str, tokens, options); add_normalized_strings_tokenized(token_tree, str, tokens, options);
string_tree_iterator_t *tokenized_iter = string_tree_iterator_new(token_tree); string_tree_iterator_t *tokenized_iter = string_tree_iterator_new(token_tree);
string_tree_iterator_t *iter; string_tree_iterator_t *iter;
@@ -796,9 +797,6 @@ static void expand_alternative(cstring_array *strings, khash_t(str_set) *unique_
char_array_destroy(temp_string); char_array_destroy(temp_string);
} }
char **expand_address(char *input, normalize_options_t options, size_t *n) { char **expand_address(char *input, normalize_options_t options, size_t *n) {
options.address_components |= ADDRESS_ANY; options.address_components |= ADDRESS_ANY;

View File

@@ -304,6 +304,7 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
if (strncmp((char *)current_tail, ptr + 1, ptr_len) == 0) { if (strncmp((char *)current_tail, ptr + 1, ptr_len) == 0) {
log_debug("node tail matches first token\n"); log_debug("node tail matches first token\n");
int tail_search_result = trie_node_search_tail_tokens(self, node, str, tokens, ptr_len, i + 1); int tail_search_result = trie_node_search_tail_tokens(self, node, str, tokens, ptr_len, i + 1);
log_debug("tail_search_result=%d\n", tail_search_result);
node_id = start_node_id; node_id = start_node_id;
node = trie_get_node(self, node_id); node = trie_get_node(self, node_id);
check_continuation = false; check_continuation = false;
@@ -331,17 +332,18 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
if (node.check <= 0 || node_id == start_node_id) { if (node.check <= 0 || node_id == start_node_id) {
log_debug("state = SEARCH_STATE_NO_MATCH\n");
state = SEARCH_STATE_NO_MATCH; state = SEARCH_STATE_NO_MATCH;
// check // check
if (last_match_index != -1) { if (last_match_index != -1) {
log_debug("last_match not NULL and state==SEARCH_STATE_NO_MATCH, data=%d", data); log_debug("last_match not NULL and state==SEARCH_STATE_NO_MATCH, data=%d\n", data);
if (*phrases == NULL) { if (*phrases == NULL) {
*phrases = phrase_array_new_size(1); *phrases = phrase_array_new_size(1);
} }
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});
i = last_match_index; i = last_match_index;
last_match_index = -1; last_match_index = -1;
phrase_start = 0; phrase_start = phrase_len = 0;
node_id = last_node_id = start_node_id; node_id = last_node_id = start_node_id;
node = last_node = trie_get_node(self, start_node_id); node = last_node = trie_get_node(self, start_node_id);
continue; continue;
@@ -360,7 +362,7 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
state = SEARCH_STATE_PARTIAL_MATCH; state = SEARCH_STATE_PARTIAL_MATCH;
if (!(node.base < 0) && (last_state == SEARCH_STATE_NO_MATCH || last_state == SEARCH_STATE_BEGIN)) { if (!(node.base < 0) && (last_state == SEARCH_STATE_NO_MATCH || last_state == SEARCH_STATE_BEGIN)) {
log_debug("phrase_start=%d\n", i); log_debug("phrase_start=%d, node.base = %d, last_state=%d\n", i, node.base, last_state);
phrase_start = i; phrase_start = i;
} }
@@ -373,6 +375,8 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
data = data_node.data; data = data_node.data;
log_debug("data = %d\n", data); log_debug("data = %d\n", data);
log_debug("phrase_start = %d\n", phrase_start);
last_match_index = i; last_match_index = i;
log_debug("last_match_index = %d\n", i); log_debug("last_match_index = %d\n", i);
} }
@@ -388,7 +392,7 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
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});
i = last_match_index; i = last_match_index;
last_match_index = -1; last_match_index = -1;
phrase_start = 0; phrase_start = phrase_len = 0;
node_id = last_node_id = start_node_id; node_id = last_node_id = start_node_id;
node = last_node = trie_get_node(self, start_node_id); node = last_node = trie_get_node(self, start_node_id);
state = SEARCH_STATE_NO_MATCH; state = SEARCH_STATE_NO_MATCH;
@@ -407,22 +411,24 @@ bool trie_search_tokens_from_index(trie_t *self, char *str, token_array *tokens,
log_debug("Ideographic character\n"); log_debug("Ideographic character\n");
last_node_id = node_id; last_node_id = node_id;
last_node = node; last_node = node;
} else if (continuation.check != node_id && last_match_index != i) { } else if (continuation.check != node_id && last_match_index != -1) {
log_debug("No continuation for phrase with start=%d, yielding tokens\n", phrase_start);
state = SEARCH_STATE_NO_MATCH;
phrase_start = 0;
node_id = last_node_id = start_node_id;
node = last_node = trie_get_node(self, start_node_id);
} else if (continuation.check != node_id && last_match_index == i) {
log_debug("node->match no continuation\n"); log_debug("node->match no continuation\n");
if (*phrases == NULL) { if (*phrases == NULL) {
*phrases = phrase_array_new_size(1); *phrases = phrase_array_new_size(1);
} }
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});
i = last_match_index;
last_match_index = -1; last_match_index = -1;
phrase_start = phrase_len = 0;
node_id = last_node_id = start_node_id; node_id = last_node_id = start_node_id;
node = last_node = trie_get_node(self, start_node_id); node = last_node = trie_get_node(self, start_node_id);
state = SEARCH_STATE_BEGIN; state = SEARCH_STATE_BEGIN;
} else if (continuation.check != node_id) {
log_debug("No continuation for phrase with start=%d, yielding tokens\n", phrase_start);
state = SEARCH_STATE_NO_MATCH;
phrase_start = phrase_len = 0;
node_id = last_node_id = start_node_id;
node = last_node = trie_get_node(self, start_node_id);
} else { } else {
log_debug("Has continuation, node_id=%d\n", continuation_id); log_debug("Has continuation, node_id=%d\n", continuation_id);
last_node = node = continuation; last_node = node = continuation;

View File

@@ -9,6 +9,6 @@ CFLAGS = $(CFLAGS_BASE)
TESTS = test_libpostal TESTS = test_libpostal
noinst_PROGRAMS = test_libpostal noinst_PROGRAMS = test_libpostal
test_libpostal_SOURCES = test.c test_expand.c test_transliterate.c test_numex.c test_libpostal_SOURCES = test.c test_expand.c test_transliterate.c test_numex.c test_trie.c
test_libpostal_LDADD = ../src/libpostal.la test_libpostal_LDADD = ../src/libpostal.la
test_libpostal_CFLAGS = $(CFLAGS_O3) test_libpostal_CFLAGS = $(CFLAGS_O3)

View File

@@ -3,6 +3,7 @@
SUITE_EXTERN(libpostal_expansion_tests); SUITE_EXTERN(libpostal_expansion_tests);
SUITE_EXTERN(libpostal_transliteration_tests); SUITE_EXTERN(libpostal_transliteration_tests);
SUITE_EXTERN(libpostal_numex_tests); SUITE_EXTERN(libpostal_numex_tests);
SUITE_EXTERN(libpostal_trie_tests);
GREATEST_MAIN_DEFS(); GREATEST_MAIN_DEFS();
@@ -13,5 +14,6 @@ int main(int argc, char **argv) {
RUN_SUITE(libpostal_expansion_tests); RUN_SUITE(libpostal_expansion_tests);
RUN_SUITE(libpostal_transliteration_tests); RUN_SUITE(libpostal_transliteration_tests);
RUN_SUITE(libpostal_numex_tests); RUN_SUITE(libpostal_numex_tests);
RUN_SUITE(libpostal_trie_tests);
GREATEST_MAIN_END(); GREATEST_MAIN_END();
} }

59
test/test_trie.c Normal file
View File

@@ -0,0 +1,59 @@
#include <stdio.h>
#include <stdlib.h>
#include <stdbool.h>
#include <stdarg.h>
#include "greatest.h"
#include "../src/scanner.h"
#include "../src/trie.h"
#include "../src/trie_search.h"
SUITE(libpostal_trie_tests);
static greatest_test_res test_trie_add_get(trie_t *trie, char *key, uint32_t data) {
bool added = trie_add(trie, key, data);
ASSERT(added);
uint32_t trie_data;
bool fetched = trie_get_data(trie, key, &trie_data);
ASSERT(fetched);
ASSERT_EQ(data, trie_data);
PASS();
}
static greatest_test_res test_trie_setup(trie_t *trie) {
CHECK_CALL(test_trie_add_get(trie, "st", 1));
CHECK_CALL(test_trie_add_get(trie, "street", 2));
CHECK_CALL(test_trie_add_get(trie, "st rt", 3));
CHECK_CALL(test_trie_add_get(trie, "st rd", 3));
CHECK_CALL(test_trie_add_get(trie, "state route", 4));
CHECK_CALL(test_trie_add_get(trie, "maine", 5));
PASS();
}
TEST test_trie(void) {
trie_t *trie = trie_new();
ASSERT(trie != NULL);
CHECK_CALL(test_trie_setup(trie));
char *input = "main st r 20";
token_array *tokens = tokenize_keep_whitespace(input);
phrase_array *phrases = trie_search_tokens(trie, input, tokens);
ASSERT(phrases != NULL);
ASSERT(phrases->n == 1);
phrase_t phrase = phrases->a[0];
ASSERT(phrase.start == 2);
ASSERT(phrase.len == 1);
trie_destroy(trie);
PASS();
}
GREATEST_SUITE(libpostal_trie_tests) {
RUN_TEST(test_trie);
}