How should I handle gzip in okhttp (decompressing)?

For my current project I’m currently using OkHttp to connect to an API endpoint. Initially I assumed that it would use gzip automatically in the request, but it doesn’t seem to enable it?

Because of this, I added an additional header (.addHeader(“Accept-Encoding”, “gzip”)) to tell the server that I would like a gzip compressed result. After adding this, the response is indeed compressed, however, I am wondering if the okhttp is supposed to handle gzip content automatically (Because the description talks about Transparent GZIP)? For now it seems like I have to manually modify my incoming stream before I can use it in my JSON parser (Using LoganSquare)

Question: Any thoughts on how I could implement this in a better way?

Sample:

OkHttpClient client = new OkHttpClient();

Request request =  new Request.Builder().url("https://api.awesomecompany.com/details/item/1000")
                                                .addHeader("X-TOKEN", "Bearer " + Auth.getInstance(mContext).getToken())
                                                .addHeader("Accept-Encoding", "gzip")
                                                .build();

Response response = client.newCall(request).execute();

if (responseCode == 200) {
    // Regular JSON parsing to model
    ItemDetailsModel itemDetailsModel = LoganSquare.parse(response.body().byteStream(), ItemDetailsModel.class);
    long responseSize = response.body().contentLength();    // Size in bytes
    ... use my model

    // Manually decompress GZIP?
    ItemDetailsModel itemDetailsModel = LoganSquare.parse(new GZIPInputStream(response.body().byteStream()), ItemDetailsModel.class);
    long responseSize = response.body().contentLength();    // Value is -1, how to get the compressed size?
}

[falcon4ever]

Hi @JakeWharton,

thanks for the response. That test certainly works.

I think I misunderstood how “Transparent GZIP” works.

For testing purposes I made a small php script that outputs the value of HTTP_ACCEPT_ENCODING (+ some text to make sure there is enough data to be gzipped). The script would return a gzip version if it was requested, else it would return an uncompressed version. I also made a small Android test project with okhttp that would call this php file.

From what I understand, okhttp will always ask for a gzipped result (testing the php page with curl and the app made this clear). However, unless you’ve manually added the entry .addHeader(“Accept-Encoding”, “gzip”), you won’t see the value “Content-Encoding: gzip” in the header of the Response object. And if I added it manually, I will have to decode the response.body().byteStream() myself.

Sorry for the confusion, you can close this issue.

[JakeWharton]

it doesn’t seem to enable it?

What makes you say this?

Accept-Encoding: gzip is added automatically and the gzip’d response will be ungzipped transparently. You can confirm by looking for the Transfer-Encoding: gzip header on the networkResponse() on the Response object.