ICU charset detection

icu-detect.cpp

// compile with g++ icu-detect.cpp -I /usr/local/include -L /usr/local/lib -licuuc -licudata -licui18n
#include <cstdio>
#include <string>
#include <cassert>
#include "unicode/ucsdet.h"
#include "unicode/uclean.h"

int main(int argc, char **argv)
{
        using namespace std;

        UErrorCode u_glob_status = U_ZERO_ERROR;
        char buf[4096];
        FILE* fp(0);
        size_t read=0;
        string data;

        u_init(&u_glob_status);
        fp = fopen(argv[1], "rb");

        assert(argc > 1);
        assert(U_SUCCESS(u_glob_status));
        assert(fp != 0);

        while(0 < (read = fread(buf, 1, 4096, fp)))
                data.append(buf, read);
        fclose(fp);
        
        UErrorCode uerr = U_ZERO_ERROR;
        UCharsetDetector *ucd = ucsdet_open ( &uerr );
        ucsdet_setText(ucd, data.c_str(), data.size(), &uerr);
        UCharsetMatch const * match = ucsdet_detect(ucd, &uerr);
        printf("Name: %s\n", ucsdet_getName(match, &uerr));
        printf("Lang: %s\n", ucsdet_getLanguage(match, &uerr));
        printf("Confidence: %u\n", ucsdet_getConfidence(match, &uerr));
        ucsdet_close(ucd);
        u_cleanup(); // keep valgrind happy!!
        return 0;
}