Datum postal_normalize(PG_FUNCTION_ARGS)
{
text *address = PG_GETARG_TEXT_P(0);
size_t arr_nelems, i;
libpostal_normalize_options_t options = libpostal_get_default_options();
char **expansions = libpostal_expand_address(text_to_cstring(address), options, &arr_nelems);
ArrayType *arr;
Datum *arr_elems = palloc(sizeof(Datum) * arr_nelems);
Oid elem_type = TEXTOID;
int16 elem_len;
char elem_byval;
char elem_align;
get_typlenbyvalalign(elem_type, &elem_len, &elem_byval, &elem_align);
for (i = 0; i < arr_nelems; i++)
{
arr_elems[i] = PointerGetDatum(cstring_to_text(expansions[i]));
}
/* Array construction takes a full copy of the input */
arr = construct_array(arr_elems, arr_nelems, elem_type, elem_len, elem_byval, elem_align);
/* Clean up unmanaged memory */
libpostal_expansion_array_destroy(expansions, arr_nelems);
PG_RETURN_ARRAYTYPE_P(arr);
}
JNIEXPORT jobjectArray JNICALL Java_com_mapzen_jpostal_AddressExpander_libpostalExpand
(JNIEnv *env, jclass cls, jstring jAddress, jobject jOptions) {
const char *address = (*env)->GetStringUTFChars(env, jAddress, 0);
size_t num_expansions = 0;
libpostal_normalize_options_t options = libpostal_get_default_options();
jfieldID fid;
jclass optionsCls = (*env)->GetObjectClass(env, jOptions);
fid = (*env)->GetFieldID(env, optionsCls, "languages", "[Ljava/lang/String;");
if (fid == 0) {
return NULL;
}
jobject jLanguages = (*env)->GetObjectField(env, jOptions, fid);
size_t num_languages = 0;
char **languages = NULL;
int i;
if (jLanguages != NULL) {
jsize jNumLanguages = (*env)->GetArrayLength(env, jLanguages);
languages = malloc(sizeof(char *) * jNumLanguages);
jboolean is_copy = JNI_FALSE;
num_languages = (size_t)jNumLanguages;
for (i = 0; i < jNumLanguages; i++) {
jstring jLanguage = (*env)->GetObjectArrayElement(env, jLanguages, i);
const char *lang = (*env)->GetStringUTFChars(env, jLanguage, &is_copy);
char *language = strdup(lang);
languages[i] = language;
(*env)->ReleaseStringUTFChars(env, jLanguage, lang);
}
options.languages = languages;
options.num_languages = num_languages;
}
fid = (*env)->GetFieldID(env, optionsCls, "addressComponents", "S");
if (fid == 0) {
return NULL;
}
options.address_components = (uint16_t) (*env)->GetShortField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "latinAscii", "Z");
if (fid == 0) {
return NULL;
}
options.latin_ascii = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "transliterate", "Z");
if (fid == 0) {
return NULL;
}
options.transliterate = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "stripAccents", "Z");
if (fid == 0) {
return NULL;
}
options.strip_accents = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "decompose", "Z");
if (fid == 0) {
return NULL;
}
options.decompose = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "lowercase", "Z");
if (fid == 0) {
return NULL;
}
options.lowercase = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "trimString", "Z");
if (fid == 0) {
return NULL;
}
options.trim_string = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "dropParentheticals", "Z");
if (fid == 0) {
return NULL;
}
options.drop_parentheticals = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "replaceNumericHyphens", "Z");
if (fid == 0) {
return NULL;
}
options.replace_numeric_hyphens = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "deleteNumericHyphens", "Z");
if (fid == 0) {
return NULL;
}
options.delete_numeric_hyphens = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "splitAlphaFromNumeric", "Z");
if (fid == 0) {
return NULL;
}
options.split_alpha_from_numeric = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "replaceWordHyphens", "Z");
if (fid == 0) {
return NULL;
}
options.replace_word_hyphens = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "deleteWordHyphens", "Z");
if (fid == 0) {
return NULL;
}
options.delete_word_hyphens = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "deleteFinalPeriods", "Z");
if (fid == 0) {
return NULL;
}
options.delete_final_periods = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "deleteAcronymPeriods", "Z");
if (fid == 0) {
return NULL;
}
options.delete_acronym_periods = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "dropEnglishPossessives", "Z");
if (fid == 0) {
return NULL;
}
options.drop_english_possessives = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "deleteApostrophes", "Z");
if (fid == 0) {
return NULL;
}
options.delete_apostrophes = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "expandNumex", "Z");
if (fid == 0) {
return NULL;
}
options.expand_numex = (*env)->GetBooleanField(env, jOptions, fid);
fid = (*env)->GetFieldID(env, optionsCls, "romanNumerals", "Z");
if (fid == 0) {
return NULL;
}
options.roman_numerals = (*env)->GetBooleanField(env, jOptions, fid);
char **expansions = libpostal_expand_address((char *)address, options, &num_expansions);
(*env)->ReleaseStringUTFChars(env, jAddress, address);
jobjectArray ret = (jobjectArray)(*env)->NewObjectArray(env,
num_expansions,
(*env)->FindClass(env, "java/lang/String"),
(*env)->NewStringUTF(env, ""));
if (num_expansions > 0) {
for (size_t i = 0; i < num_expansions; i++) {
(*env)->SetObjectArrayElement(env, ret, i, (*env)->NewStringUTF(env, expansions[i]));
}
}
if (expansions != NULL) {
libpostal_expansion_array_destroy(expansions, num_expansions);
}
if (languages != NULL) {
for (i = 0; i < num_languages; i++) {
free(languages[i]);
}
free(languages);
}
return ret;
}