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

[expansion] Fixing cases like ML King where a global (all languages) expansion subsumes the specific language expansion (like English)

Browse Source
This commit is contained in:
Al
2015-12-07 18:09:20 -05:00
parent 9bab70909d
commit 0d8d396108
1 changed files with 8 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/libpostal.c
View File

@@ -23,7 +23,7 @@ typedef struct phrase_language {
VECTOR_INIT(phrase_language_array, phrase_language_t) VECTOR_INIT(phrase_language_array, phrase_language_t)
#define ks_lt_phrase_language(a, b) ((a).phrase.start < (b).phrase.start) #define ks_lt_phrase_language(a, b) ((a).phrase.start < (b).phrase.start || ((a).phrase.start == (b).phrase.start && (a).phrase.len > (b).phrase.len))
KSORT_INIT(phrase_language_array, phrase_language_t, ks_lt_phrase_language) KSORT_INIT(phrase_language_array, phrase_language_t, ks_lt_phrase_language)
@@ -167,6 +167,12 @@ string_tree_t *add_string_alternatives(char *str, normalize_options_t options) {
for (int i = 0; i < phrases->n; i++) { for (int i = 0; i < phrases->n; i++) {
phrase_lang = phrases->a[i]; phrase_lang = phrases->a[i];
phrase_t phrase = phrase_lang.phrase;
if (phrase.start < start) {
continue;
}
char_array_clear(key); char_array_clear(key);
char_array_cat(key, phrase_lang.language); char_array_cat(key, phrase_lang.language);
@@ -174,8 +180,6 @@ string_tree_t *add_string_alternatives(char *str, normalize_options_t options) {
size_t namespace_len = key->n; size_t namespace_len = key->n;
phrase_t phrase = phrase_lang.phrase;
end = phrase.start; end = phrase.start;
for (int j = start; j < end; j++) { for (int j = start; j < end; j++) {
@@ -196,7 +200,7 @@ string_tree_t *add_string_alternatives(char *str, normalize_options_t options) {
token_t token; token_t token;
if (value.components & options.address_components) { if ((value.components & options.address_components) > 0) {
key->n = namespace_len; key->n = namespace_len;
for (int j = phrase.start; j < phrase.start + phrase.len; j++) { for (int j = phrase.start; j < phrase.start + phrase.len; j++) {
token = tokens->a[j]; token = tokens->a[j];