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

[api/compat] PR #465 redefined the language classifier response struct in the API and was casting between incompatible pointer types. Using the exported struct throughout.

Browse Source
This commit is contained in:
Al
2025-01-30 01:45:18 -05:00
parent 7245a178cf
commit 0540d7c7e3
9 changed files with 19 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/expand.c
View File

@@ -1567,7 +1567,7 @@ cstring_array *expand_address_phrase_option(char *input, libpostal_normalize_opt
size_t len = strlen(input); size_t len = strlen(input);
language_classifier_response_t *lang_response = NULL; libpostal_language_classifier_response_t *lang_response = NULL;
if (options.num_languages == 0) { if (options.num_languages == 0) {
lang_response = classify_languages(input); lang_response = classify_languages(input);
@@ -1627,7 +1627,7 @@ cstring_array *expand_address_phrase_option(char *input, libpostal_normalize_opt
kh_destroy(str_set, unique_strings); kh_destroy(str_set, unique_strings);
if (lang_response != NULL) { if (lang_response != NULL) {
language_classifier_response_destroy(lang_response); libpostal_language_classifier_response_destroy(lang_response);
} }
char_array_destroy(temp_string); char_array_destroy(temp_string);

8
src/language_classifier.c
View File

@@ -46,7 +46,7 @@ language_classifier_t *get_language_classifier(void) {
return language_classifier; return language_classifier;
} }
void language_classifier_response_destroy(language_classifier_response_t *self) { void language_classifier_response_destroy(libpostal_language_classifier_response_t *self) {
if (self == NULL) return; if (self == NULL) return;
if (self->languages != NULL) { if (self->languages != NULL) {
free(self->languages); free(self->languages);
@@ -59,7 +59,7 @@ void language_classifier_response_destroy(language_classifier_response_t *self)
free(self); free(self);
} }
language_classifier_response_t *classify_languages(char *address) { libpostal_language_classifier_response_t *classify_languages(char *address) {
language_classifier_t *classifier = get_language_classifier(); language_classifier_t *classifier = get_language_classifier();
if (classifier == NULL) { if (classifier == NULL) {
@@ -88,7 +88,7 @@ language_classifier_response_t *classify_languages(char *address) {
size_t n = classifier->num_labels; size_t n = classifier->num_labels;
double_matrix_t *p_y = double_matrix_new_zeros(1, n); double_matrix_t *p_y = double_matrix_new_zeros(1, n);
language_classifier_response_t *response = NULL; libpostal_language_classifier_response_t *response = NULL;
bool model_exp = false; bool model_exp = false;
if (classifier->weights_type == MATRIX_DENSE) { if (classifier->weights_type == MATRIX_DENSE) {
model_exp = logistic_regression_model_expectation(classifier->weights.dense, x, p_y); model_exp = logistic_regression_model_expectation(classifier->weights.dense, x, p_y);
@@ -129,7 +129,7 @@ language_classifier_response_t *classify_languages(char *address) {
free(indices); free(indices);
response = malloc(sizeof(language_classifier_response_t)); response = malloc(sizeof(libpostal_language_classifier_response_t));
response->num_languages = num_languages; response->num_languages = num_languages;
response->languages = languages; response->languages = languages;
response->probs = probs; response->probs = probs;

15
src/language_classifier.h
View File

@@ -6,6 +6,8 @@
#include <stdint.h> #include <stdint.h>
#include <stdbool.h> #include <stdbool.h>
#include "libpostal.h"
#include "collections.h" #include "collections.h"
#include "language_features.h" #include "language_features.h"
#include "logistic_regression.h" #include "logistic_regression.h"
@@ -29,21 +31,14 @@ typedef struct language_classifier {
} weights; } weights;
} language_classifier_t; } language_classifier_t;
typedef struct language_classifier_response {
size_t num_languages;
char **languages;
double *probs;
} language_classifier_response_t;
// General usage // General usage
language_classifier_t *language_classifier_new(void); language_classifier_t *language_classifier_new(void);
language_classifier_t *get_language_classifier(void); language_classifier_t *get_language_classifier(void);
language_classifier_t *get_language_classifier_country(void); language_classifier_t *get_language_classifier_country(void);
language_classifier_response_t *classify_languages(char *address); libpostal_language_classifier_response_t *classify_languages(char *address);
void language_classifier_response_destroy(language_classifier_response_t *self); void language_classifier_response_destroy(libpostal_language_classifier_response_t *self);
void language_classifier_destroy(language_classifier_t *self); void language_classifier_destroy(language_classifier_t *self);
@@ -58,4 +53,4 @@ bool language_classifier_module_setup(char *dir);
void language_classifier_module_teardown(void); void language_classifier_module_teardown(void);
#endif #endif

2
src/language_classifier_cli.c
View File

@@ -29,7 +29,7 @@ int main(int argc, char **argv) {
} }
language_classifier_response_t *response = classify_languages(address); libpostal_language_classifier_response_t *response = classify_languages(address);
if (response == NULL) { if (response == NULL) {
printf("Could not classify language\n"); printf("Could not classify language\n");
exit(EXIT_FAILURE); exit(EXIT_FAILURE);

2
src/language_classifier_test.c
View File

@@ -34,7 +34,7 @@ double test_accuracy(char *filename) {
continue; continue;
} }
language_classifier_response_t *response = classify_languages(address); libpostal_language_classifier_response_t *response = classify_languages(address);
if (response == NULL || response->num_languages == 0) { if (response == NULL || response->num_languages == 0) {
printf("%s\tNULL\t%s\n", language, address); printf("%s\tNULL\t%s\n", language, address);
continue; continue;

2
src/libpostal.c
View File

@@ -119,7 +119,7 @@ char **libpostal_near_dupe_hashes_languages(size_t num_components, char **labels
char **libpostal_place_languages(size_t num_components, char **labels, char **values, size_t *num_languages) { char **libpostal_place_languages(size_t num_components, char **labels, char **values, size_t *num_languages) {
language_classifier_response_t *lang_response = place_languages(num_components, labels, values); libpostal_language_classifier_response_t *lang_response = place_languages(num_components, labels, values);
if (lang_response == NULL) { if (lang_response == NULL) {
*num_languages = 0; *num_languages = 0;
return NULL; return NULL;

2
src/near_dupe.c
View File

@@ -670,7 +670,7 @@ cstring_array *near_dupe_hashes_languages(size_t num_components, char **labels,
libpostal_normalize_options_t normalize_options = libpostal_get_default_options(); libpostal_normalize_options_t normalize_options = libpostal_get_default_options();
language_classifier_response_t *lang_response = NULL; libpostal_language_classifier_response_t *lang_response = NULL;
if (num_languages == 0) { if (num_languages == 0) {
lang_response = place_languages(num_components, labels, values); lang_response = place_languages(num_components, labels, values);

4
src/place.c
View File

@@ -17,10 +17,10 @@ static inline bool is_address_text_component(char *label) {
); );
} }
language_classifier_response_t *place_languages(size_t num_components, char **labels, char **values) { libpostal_language_classifier_response_t *place_languages(size_t num_components, char **labels, char **values) {
if (num_components == 0 || values == NULL || labels == NULL) return NULL; if (num_components == 0 || values == NULL || labels == NULL) return NULL;
language_classifier_response_t *lang_response = NULL; libpostal_language_classifier_response_t *lang_response = NULL;
char *label; char *label;
char *value; char *value;

4
src/place.h
View File

@@ -32,7 +32,7 @@ typedef struct place {
char *website; char *website;
} place_t; } place_t;
language_classifier_response_t *place_languages(size_t num_components, char **labels, char **values); libpostal_language_classifier_response_t *place_languages(size_t num_components, char **labels, char **values);
place_t *place_new(void); place_t *place_new(void);
@@ -40,4 +40,4 @@ place_t *place_from_components(size_t num_components, char **labels, char **valu
void place_destroy(place_t *place); void place_destroy(place_t *place);
#endif #endif