utf8 decoder

get_utf8_char.c

int get_utf8_char(FILE *stream)
{
        int a, b, c, d;
        
        a = fgetc(stream);
        if (ferror(stream)) goto error;
        if (a == EOF) { return EOF; }

        // Interesting see how ASCII and UTF-8 having good relationships
        if ((a & 0x80) == 0) {
                return a;
        }

#define CHECK(condition) if (!(condition)) goto error;
        CHECK((a & 0x40) != 0)

        b = fgetc(stream);
        if (ferror(stream)) goto error;
        CHECK((b & 0xc0) == 0x80)
        b &= 0x3f;

        if ((a & 0x20) == 0) {
                CHECK((a & 0x1e) > 0)
                return ((a & 0x1f) << 6) | b;
        }

        c = fgetc(stream);
        if (ferror(stream)) goto error;
        CHECK((c & 0xc0) == 0x80)

        if ((a & 0x10) == 0) {
                CHECK(((a & 0x0f) | (b & 0x20)) > 0)
                return ((a & 0x0f) << 12) | (b << 6) | (c & 0x3f);
        }

        d = fgetc(stream);
        if (ferror(stream)) goto error;
        CHECK((d & 0xc0) == 0x80)
        c &= 0x3f;

        if ((a & 0x08) == 0) {
                CHECK(((a & 0x07) | (b & 0x30)) > 0)
                return ((a & 0x07) << 18) | (b << 12) | (c << 6) | (d & 0x3f);
        }

        CHECK(0);
#undef  CHECK

error:
        perror("utf8_decoder");
        return 0;
}