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

[expansion] Changes to address_expansion struct to allow for multiple dictionaries per record. Only adding unique canonical strings to the string array

Browse Source
This commit is contained in:
Al
2015-07-22 20:35:29 -04:00
parent 454be89121
commit 27af28eacf
3 changed files with 82 additions and 34 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
src/address_dictionary.c
View File

@@ -18,12 +18,24 @@ address_expansion_array *address_dictionary_get_expansions(char *key) {
return k != kh_end(address_dict->expansions) ? kh_value(address_dict->expansions, k) : NULL;
}
int32_t address_dictionary_next_canonical_index(void) {
if (address_dict == NULL || address_dict->canonical == NULL) return -1;
return (int32_t)cstring_array_num_strings(address_dict->canonical);
}
bool address_dictionary_add_canonical(char *canonical) {
if (address_dict == NULL || address_dict->canonical == NULL) return false;
cstring_array_add_string(address_dict->canonical, canonical);
return true;
}
char *address_dictionary_get_canonical(uint32_t index) {
if (address_dict == NULL || address_dict->canonical == NULL || index > cstring_array_num_strings(address_dict->canonical)) return NULL;
return cstring_array_get_string(address_dict->canonical, index);
}
bool address_dictionary_add_expansion(char *key, char *canonical, char *language, uint16_t dictionary_id, uint16_t address_components) {
bool address_dictionary_add_expansion(char *key, address_expansion_t expansion) {
int ret;
log_debug("key=%s\n", key);
@@ -33,22 +45,7 @@ bool address_dictionary_add_expansion(char *key, char *canonical, char *language
expansion_value_t value;
value.value = 0;
if (canonical == NULL) {
canonical_index = -1;
value.canonical = 1;
} else {
canonical_index = (int32_t) cstring_array_num_strings(address_dict->canonical);
cstring_array_add_string(address_dict->canonical, canonical);
value.canonical = 0;
}
address_expansion_t expansion;
expansion.canonical_index = canonical_index;
strcpy(expansion.language, language);
expansion.dictionary_id = dictionary_id;
expansion.address_components = address_components;
value.canonical = expansion.canonical_index == -1;
if (expansions == NULL) {
expansions = address_expansion_array_new_size(1);
@@ -57,16 +54,14 @@ bool address_dictionary_add_expansion(char *key, char *canonical, char *language
kh_value(address_dict->expansions, k) = expansions;
value.count = 1;
value.components = address_components;
value.components = expansion.address_components;
log_debug("value.count=%d, value.components=%d\n", value.count, value.components);
trie_add(address_dict->trie, key, value.value);
} else {
uint32_t node_id = trie_get(address_dict->trie, key);
log_debug("node_id=%d\n", node_id);
if (node_id != NULL_NODE_ID) {
if (!trie_get_data_at_index(address_dict->trie, node_id, &value.value)) {
log_warn("get_data_at_index returned false\n");
return false;
@@ -79,7 +74,7 @@ bool address_dictionary_add_expansion(char *key, char *canonical, char *language
}
value.count++;
value.components |= address_components;
value.components |= expansion.address_components;
if (!trie_set_data_at_index(address_dict->trie, node_id, value.value)) {
log_warn("set_data_at_index returned false for node_id=%d and value=%d\n", node_id, value.value);
@@ -187,10 +182,16 @@ static bool address_expansion_read(FILE *f, address_expansion_t *expansion) {
return false;
}
if (!file_read_uint16(f, &expansion->dictionary_id)) {
if (!file_read_uint32(f, (uint32_t *)&expansion->num_dictionaries)) {
return false;
}
for (int i = 0; i < expansion->num_dictionaries; i++) {
if (!file_read_uint16(f, (uint16_t *)expansion->dictionary_ids + i)) {
return false;
}
}
if (!file_read_uint16(f, &expansion->address_components)) {
return false;
}
@@ -207,12 +208,21 @@ static bool address_expansion_write(FILE *f, address_expansion_t expansion) {
if (!file_write_uint32(f, (uint32_t)expansion.canonical_index) ||
!file_write_uint32(f, language_len) ||
!file_write_chars(f, expansion.language, language_len) ||
!file_write_uint16(f, expansion.dictionary_id) ||
!file_write_uint16(f, expansion.address_components)
!file_write_uint32(f, expansion.num_dictionaries)
) {
return false;
}
for (int i = 0; i < expansion.num_dictionaries; i++) {
if (!file_write_uint16(f, expansion.dictionary_ids[i])) {
return false;
}
}
if (!file_write_uint16(f, expansion.address_components)) {
return false;
}
return true;
}

7
src/address_dictionary.h
View File

@@ -35,7 +35,8 @@ typedef union expansion_value {
typedef struct address_expansion {
int32_t canonical_index;
char language[MAX_LANGUAGE_LEN];
uint16_t dictionary_id;
uint32_t num_dictionaries;
uint16_t dictionary_ids[MAX_DICTIONARY_TYPES];
uint16_t address_components;
} address_expansion_t;
@@ -56,7 +57,9 @@ bool address_dictionary_init(void);
phrase_array *search_address_dictionaries(char *str, char *lang);
address_expansion_array *address_dictionary_get_expansions(char *key);
char *address_dictionary_get_canonical(uint32_t index);
bool address_dictionary_add_expansion(char *key, char *canonical, char *language, uint16_t dictionary_id, uint16_t address_components);
int32_t address_dictionary_next_canonical_index(void);
bool address_dictionary_add_canonical(char *canonical);
bool address_dictionary_add_expansion(char *key, address_expansion_t expansion);
void address_dictionary_destroy(address_dictionary_t *self);

51
src/address_dictionary_builder.c
View File

@@ -26,6 +26,8 @@ int main(int argc, char **argv) {
khash_t(int_int) *dictionary_components = kh_init(int_int);
khash_t(str_int) *canonical_indices = kh_init(str_int);
khiter_t k;
for (int g = 0; g < NUM_DICTIONARY_TYPES; g++) {
@@ -41,26 +43,57 @@ int main(int argc, char **argv) {
address_language_index_t lang_index = expansion_languages[i];
char *language = lang_index.language;
log_info("Doing language: %s\n", language);
for (int j = lang_index.index; j < lang_index.index + lang_index.len; j++) {
address_expansion_rule_t expansion_rule = expansion_rules[j];
uint16_t dictionary_id = (uint16_t) expansion_rule.dictionary;
k = kh_get(int_int, dictionary_components, (uint32_t)dictionary_id);
if (k == kh_end(dictionary_components)) {
log_error("Invalid dictionary_type: %d\n", dictionary_id);
exit(EXIT_FAILURE);
uint16_t address_components = 0;
address_expansion_t expansion;
strcpy(expansion.language, language);
expansion.num_dictionaries = expansion_rule.num_dictionaries;
for (int d = 0; d < expansion_rule.num_dictionaries; d++) {
uint16_t dictionary_id = (uint16_t) expansion_rule.dictionaries[d];
expansion.dictionary_ids[d] = dictionary_id;
k = kh_get(int_int, dictionary_components, (uint32_t)dictionary_id);
if (k == kh_end(dictionary_components)) {
log_error("Invalid dictionary_type: %d\n", dictionary_id);
exit(EXIT_FAILURE);
}
address_components |= (uint16_t) kh_value(dictionary_components, k);
}
uint16_t address_components = (uint16_t) kh_value(dictionary_components, k);
expansion.address_components = address_components;
char *canonical = NULL;
if (expansion_rule.canonical_index != -1) {
canonical = canonical_strings[expansion_rule.canonical_index];
}
if (canonical == NULL) {
expansion.canonical_index = -1;
} else {
k = kh_get(str_int, canonical_indices, canonical);
if (k != kh_end(canonical_indices)) {
expansion.canonical_index = kh_value(canonical_indices, k);
} else {
expansion.canonical_index = address_dictionary_next_canonical_index();
if (!address_dictionary_add_canonical(canonical)) {
log_error("Error adding canonical string: %s\n", canonical);
exit(EXIT_FAILURE);
}
}
}
// Add the phrase itself to the base namespace for existence checks
if (!address_dictionary_add_expansion(expansion_rule.phrase, canonical, language, dictionary_id, address_components)) {
if (!address_dictionary_add_expansion(expansion_rule.phrase, expansion)) {
log_error("Could not add expansion {%s}\n", expansion_rule.phrase);
exit(EXIT_FAILURE);
}
@@ -73,7 +106,7 @@ int main(int argc, char **argv) {
char_array_cat(key, expansion_rule.phrase);
char *token = char_array_get_string(key);
if (!address_dictionary_add_expansion(token, canonical, language, dictionary_id, address_components)) {
if (!address_dictionary_add_expansion(token, expansion)) {
log_error("Could not add language expansion {%s, %s}\n", language, expansion_rule.phrase);
exit(EXIT_FAILURE);
}
@@ -86,5 +119,7 @@ int main(int argc, char **argv) {
kh_destroy(int_int, dictionary_components);
kh_destroy(str_int, canonical_indices);
address_dictionary_module_teardown();
}