Merge pull request #297 from oschwald/greg/fix-more-leaks

Fix several more memory leaks and a segfault

src/address_parser.c

@@ -1755,8 +1755,6 @@ libpostal_address_parser_response_t *address_parser_parse(char *address, char *l
             label = strdup(ADDRESS_PARSER_LABEL_POSTAL_CODE);
         }
 
-        response = address_parser_response_new();
-
         // Implicit: if most_common is not one of the above, ignore and parse regularly
         if (label != NULL) {
             char **single_label = malloc(sizeof(char *));
@@ -1764,6 +1762,8 @@ libpostal_address_parser_response_t *address_parser_parse(char *address, char *l
             char **single_component = malloc(sizeof(char *));
             single_component[0] = strdup(normalized);
 
+            response = address_parser_response_new();
+
             response->num_components = 1;
             response->labels = single_label;
             response->components = single_component;

src/expand.c

@@ -1491,8 +1491,12 @@ void expand_alternative_phrase_option(cstring_array *strings, khash_t(str_set) *
             for (; !string_tree_iterator_done(iter); string_tree_iterator_next(iter)) {
                 char_array_clear(temp_string);
                 string_tree_iterator_foreach_token(iter, token, {
-                    log_debug("token=%s\n", token);
-                    char_array_append(temp_string, token);
+                    if (token == NULL) {
+                        log_debug("token=NULL\n");
+                    } else {
+                        log_debug("token=%s\n", token);
+                        char_array_append(temp_string, token);
+                    }
                 })
                 char_array_terminate(temp_string);
 

src/numex.c

@@ -1232,6 +1232,7 @@ char *replace_numeric_expressions(char *str, char *lang) {
 
     if (!have_valid_numex) {
         numex_result_array_destroy(results);
+        char_array_destroy(replacement);
         return NULL;
     }
 

test/test_expand.c

@@ -135,6 +135,15 @@ TEST test_expansions(void) {
     CHECK_CALL(test_expansion_contains_with_languages("Marktstrasse", "markt strasse", options, 1, "de"));
     CHECK_CALL(test_expansion_contains_with_languages("Hoofdstraat", "hoofdstraat", options, 1, "nl"));
     CHECK_CALL(test_expansion_contains_with_languages("มงแตร", "มงแตร", options, 1, "th"));
+
+    PASS();
+}
+
+TEST test_expansion_for_invalid_input(void) {
+    size_t num_expansions;
+
+    // This is tested as the input caused a segfault in expand_alternative_phrase_option
+    libpostal_expand_address("bob-ad1233@blah.co", libpostal_get_default_options(), &num_expansions);
     PASS();
 }
 
@@ -314,6 +323,7 @@ SUITE(libpostal_expansion_tests) {
     RUN_TEST(test_po_box_root_expansions);
     RUN_TEST(test_expansions_language_classifier);
     RUN_TEST(test_expansions_no_options);
+    RUN_TEST(test_expansion_for_invalid_input);
 
     libpostal_teardown();
     libpostal_teardown_language_classifier();

test/test_trie.c

@@ -49,6 +49,8 @@ TEST test_trie(void) {
     ASSERT(phrase.start == 2);
     ASSERT(phrase.len == 1);
 
+    phrase_array_destroy(phrases);
+    token_array_destroy(tokens);
     trie_destroy(trie);
 
     PASS();