JNIEXPORT jstring JNICALL Java_org_mozilla_charsetdetector_CharsetDetector_get_1encoding(
JNIEnv *env, jclass jclazz, jstring file) {
const char * path;
jboolean iscopy;
path = env->GetStringUTFChars(file, &iscopy);
char buf[4096];
char encoding[CHARDET_MAX_ENCODING_NAME];
size_t len;
int res = 0;
chardet_t det = NULL;
FILE* fp = NULL;
chardet_create(&det);
fp = fopen(path, "rb");
do {
len = fread(buf, 1, sizeof(buf), fp);
res = chardet_handle_data(det, buf, len);
} while (res==CHARDET_RESULT_OK && feof(fp)==0);
fclose(fp);
chardet_data_end(det);
int result = chardet_get_charset(det, encoding, CHARDET_MAX_ENCODING_NAME);
chardet_destroy(det);
env->ReleaseStringUTFChars(file, path);
if (result == CHARDET_RESULT_OK) {
return env->NewStringUTF(encoding);
}
return 0;
}
char * GetLocalEncoding(const char* in_str, unsigned int str_len){
chardet_t chardect=NULL;
char *out_encode=new char[CHARDET_MAX_ENCODING_NAME];
if(chardet_create(&chardect)==CHARDET_RESULT_OK){
if(chardet_handle_data(chardect, in_str, (unsigned int)str_len) == CHARDET_RESULT_OK){
if(chardet_data_end(chardect) == CHARDET_RESULT_OK){
chardet_get_charset(chardect, out_encode, CHARDET_MAX_ENCODING_NAME);
}
}
}
if(chardect){
chardet_destroy(chardect);
return out_encode;
}
else{
return NULL;
}
}
int main() {
int res;
chardet_t * detector = chardet_create();
char const * s = "Hello World";
res = chardet_feed(detector, s, strlen(s));
assert(res == 0);
chardet_stat_t const ** stats;
chardet_get_stats(detector, &stats);
printf("Probabilities:\n");
for (int i = 0; stats[i] != NULL; i++) {
char const * encoding = chardet_stat_get_encoding(stats[i]);
float proba = chardet_stat_get_probability(stats[i]);
printf(" - %s: %5.3f\n", encoding, proba);
}
chardet_destroy(detector);
}
~Private() {
if (!m_det)
return;
chardet_destroy(m_det);
m_det = NULL;
}