Manually parsing HTTP response header

Due to some constraints I had to manually parse an HTTP response coming out of a stream.

I could not use buffered reader or scanner because the perform buffering and end up buffering bytes beyond the response header itself.

I'll be hitting my AWS S3 and only expect 200 or 206 status response.

This seems to work, but I'd appreciate some review.

final InputStream reader = socket.getInputStream();
stringBuilder.setLength(0);
//first fetch header
while (true) {

    final char read = (char) reader.read();
    if (read  4 &&
            stringBuilder.charAt(currentCount - 1) == '\n' &&
            stringBuilder.charAt(currentCount - 2) == '\r' &&
            stringBuilder.charAt(currentCount - 3) == '\n' &&
            stringBuilder.charAt(currentCount - 4) == '\r') {
        break; //response consumed
    }
}

//print the header
final String completeResponse = stringBuilder.toString();
Log.i("Downloader", completeResponse);

This is a bit surprising:

stringBuilder.setLength(0);

Seeing the posted code starting with this is fishy,
as it suggests that there is more code before it,
and that the method is not decomposed enough,
and should be multiple methods instead.

This is a bit surprising:

final char read = (char) reader.read();
    if (read < 1) {

The read() method returns -1 in case of EOF.
I suggest to change this check to `read

• A method that consumes the remaining lines

StackExchange Code Review Q#101306, answer score: 5