Download binary file from OKHTTP

The best option to download (based on source code "okio")

private void download(@NonNull String url, @NonNull File destFile) throws IOException {
    Request request = new Request.Builder().url(url).build();
    Response response = okHttpClient.newCall(request).execute();
    ResponseBody body = response.body();
    long contentLength = body.contentLength();
    BufferedSource source = body.source();

    BufferedSink sink = Okio.buffer(Okio.sink(destFile));
    Buffer sinkBuffer = sink.buffer();

    long totalBytesRead = 0;
    int bufferSize = 8 * 1024;
    for (long bytesRead; (bytesRead = source.read(sinkBuffer, bufferSize)) != -1; ) {
        sink.emit();
        totalBytesRead += bytesRead;
        int progress = (int) ((totalBytesRead * 100) / contentLength);
        publishProgress(progress);
    }
    sink.flush();
    sink.close();
    source.close();
}

This is how I use Okhttp + Okio libraries while publishing download progress after every chunk download:

public static final int DOWNLOAD_CHUNK_SIZE = 2048; //Same as Okio Segment.SIZE

try {
        Request request = new Request.Builder().url(uri.toString()).build();

        Response response = client.newCall(request).execute();
        ResponseBody body = response.body();
        long contentLength = body.contentLength();
        BufferedSource source = body.source();

        File file = new File(getDownloadPathFrom(uri));
        BufferedSink sink = Okio.buffer(Okio.sink(file));

        long totalRead = 0;
        long read = 0;
        while ((read = source.read(sink.buffer(), DOWNLOAD_CHUNK_SIZE)) != -1) {
            totalRead += read;
            int progress = (int) ((totalRead * 100) / contentLength);
            publishProgress(progress);
        }
        sink.writeAll(source);
        sink.flush();
        sink.close();
        publishProgress(FileInfo.FULL);
} catch (IOException e) {
        publishProgress(FileInfo.CODE_DOWNLOAD_ERROR);
        Logger.reportException(e);
}

Getting ByteStream from OKHTTP

I've been digging around in the Documentation of OkHttp you need to go this way

use this method :

response.body().byteStream() wich will return an InputStream

so you can simply use a BufferedReader or any other alternative

OkHttpClient client = new OkHttpClient();
request = new Request.Builder().url("URL string here")
                     .addHeader("X-CSRFToken", csrftoken)
                     .addHeader("Content-Type", "application/json")
                     .build();
response = getClient().newCall(request).execute();

InputStream in = response.body().byteStream();
BufferedReader reader = new BufferedReader(new InputStreamReader(in));
String result, line = reader.readLine();
result = line;
while((line = reader.readLine()) != null) {
    result += line;
}
System.out.println(result);
response.body().close();

For what it's worth, I would recommend response.body().source() from okio (since OkHttp is already supporting it natively) in order to enjoy an easier way to manipulate a large quantity of data that can come when downloading a file.

@Override
public void onResponse(Call call, Response response) throws IOException {
    File downloadedFile = new File(context.getCacheDir(), filename);
    BufferedSink sink = Okio.buffer(Okio.sink(downloadedFile));
    sink.writeAll(response.body().source());
    sink.close();
}

A couple of advantages taken from the documentation in comparison with InputStream:

This interface is functionally equivalent to InputStream. InputStream requires multiple layers when consumed data is heterogeneous: a DataInputStream for primitive values, a BufferedInputStream for buffering, and InputStreamReader for strings. This class uses BufferedSource for all of the above. Source avoids the impossible-to-implement available() method. Instead callers specify how many bytes they require.

Source omits the unsafe-to-compose mark and reset state that's tracked by InputStream; callers instead just buffer what they need.

When implementing a source, you need not worry about the single-byte read method that is awkward to implement efficiently and that returns one of 257 possible values.

And source has a stronger skip method: BufferedSource.skip(long) won't return prematurely.