container.logs().fetch() returning non-ASCII value

Hi Everyone,

I am running ubuntu image with command as date.

docker logs is showing correct output but my program is not able to print anything.

Program:

public class App {
	public static void main(String[] args) throws Exception {
		final Container container = new LocalDocker(new File("/var/run/docker.sock")).containers().create("ubuntutest",
				containerJsonObject());
		container.start();

		while (runningState(container)) {

		}
		final String logs = container.logs().fetch();
		logs.trim();
		System.out.println(logs);
		System.out.println(logs.length());
	}

	private static JsonObject containerJsonObject() {
		JsonObjectBuilder json = Json.createObjectBuilder();
		json.add("AttachStdout", true);
		json.add("AttachStderr", true);
		json.add("Image", "ubuntu");
		json.add("Tty", false);
		json.add("Cmd", "date");
		JsonArrayBuilder builder = Json.createArrayBuilder();
		builder.add("OUT=/scans/firstimg.jpg");
		json.add("Env", builder);
		JsonObjectBuilder hostConfig = Json.createObjectBuilder();
		JsonArrayBuilder binds = Json.createArrayBuilder();
		binds.add("/home/shubham/scans:/scans/");
		hostConfig.add("Binds", binds);
		hostConfig.add("Privileged", true);
		json.add("HostConfig", hostConfig);
		return json.build();
	}

	private static boolean runningState(final Container container) throws IOException {
		return container.inspect().getJsonObject("State").getBoolean("Running");
	}

}

Output of above is:

37

Thanks for any help!!

[tuxji]

Yes, the new 0.0.12 version fixes the problem really well and the new Container.wait() method is wonderful too! Thank you!

[tuxji]

I just found the same problem, and I know what the problem is. When you call GET /containers/{id}/logs to get stdout and stderr logs from a container, the logs are returned using a binary stream format in the response body. The stream format is described in the documentation for the attach endpoint. Briefly summarizing, eight binary bytes are prepended to each line of logging output to tell you 1) whether the next line came from stdout or stderr, and 2) how many bytes long the next line is.

Here are two example lines taken from my Emacs buffer visiting a file saved by wget ‘http://localhost:2375/v1.35/containers/{id}/logs?stdout=true&stderr=true’:

^A^@^@^@^@^@^@,Info: STEM is running. Please be patient...
^B^@^@^@^@^@^@c[main] INFO kbaseRoot set to /app/STEM

The first byte is a ^A or ^B (1 or 2) followed by six zero (null) bytes, then followed by a byte which looks like an ordinary character but really is meant to be the size of the next line, and then the line itself is appended. I think it would be easier for callers if fetch() trimmed these eight bytes from the beginning of each line in the string output it reads from the logs API endpoint, but one could disagree, simply document the stream format in the javadoc, and make callers decode the stream format themselves.