Xqueue external grading system plugin for Tutor

This is a plugin for Tutor that provides the Xqueue external grading system for Open edX platforms. If you don't know what it is, you probably don't need it.

Installation

The plugin is currently bundled with the binary releases of Tutor. If you have installed Tutor from source, you will have to install this plugin from source, too:

pip install tutor-xqueue

Then, to enable this plugin, run:

tutor plugins enable xqueue

You should then run the initialisation scripts. The easiest way to do this is to run tutor local quickstart.

Usage

In the Open edX studio, edit a course and add a new "Advanced blank problem" ("Problem" 🠆 "Advanced" 🠆 "Blank Advanced Problem"). Then, click "Edit" and copy-paste the following in the editor:

<problem>
  <coderesponse queuename="openedx">
    <label>Write a program that prints "hello world".</label>
    <textbox rows="10" cols="80" mode="python" tabsize="4"/>
    <codeparam>
      <initial_display>
        # students write your program here
        print("")
      </initial_display>
      <answer_display>
        print("hello world")
      </answer_display>
      <grader_payload>
        {"name": "hello world"}
      </grader_payload>
    </codeparam>
  </coderesponse>
</problem>

For a problem that includes a file submission, write instead:

<problem>
  <coderesponse queuename="openedx">
    <filesubmission/>
    <codeparam>
      <grader_payload>
        {"name": "file submission"}
      </grader_payload>
    </codeparam>
  </coderesponse>
</problem>

Note that in all cases, the queue name must be "openedx".

Save and publish the created unit. Then, access the unit from the LMS and attempt to answer the problem. The answer is sent to the Xqueue service. If you know how to use the Xqueue API, you can access it at http(s)://xqueue.LMS_HOST (in production) or http://xqueue.local.overhang.io (in development). However, the Xqueue API is a bit awkward to use. Tutor provides a simple command-line interface to interact with the Xqueue service.

Count the number of submissions that need to be graded:

$ tutor xqueue submissions count
{
  "content": 0,
  "return_code": 0
}
Note

By default, tutor xqueue submissions will hit the Xqueue API running at http(s)://xqueue.LMS_HOST. To hit a different server, you should pass the --url=http://xqueue.yourcustomhost.com option to the CLI. Alternatively, and to avoid passing this option every time, you can define the following environment variable:

export TUTOR_XQUEUE_URL=http://xqueue.yourcustomhost.com

Show the first submission that should be graded:

$ tutor xqueue submissions show
{
  "id": 1,
  "key": "692c2896cdfc8bdc2d073bc3b3daf928",
  "body": {
    "student_info": "{\"random_seed\": 1, \"anonymous_student_id\": \"af46c9d6c05627aee45257d155ec0b79\", \"submission_time\": \"20200504101653\"}",
    "grader_payload": "\n        {\"output\": \"hello world\", \"max_length\": 2}\n      ",
    "student_response": "        # students write your program here\r\n        print \"42\"\r\n      "
  },
  "files": {},
  "return_code": 0
}

Grade the submission (in this case, mark it as being correct):

$ tutor xqueue submissions grade 1 692c2896cdfc8bdc2d073bc3b3daf928 0.9 true "Good job!"
{
  "content": "",
  "return_code": 0
}

The submission should then appear as correct with the message that you provided on the command line:

Correct answer

Configuration

  • XQUEUE_AUTH_PASSWORD (default: "{{ 8|random_string }}")
  • XQUEUE_AUTH_USERNAME (default: "lms")
  • XQUEUE_DOCKER_IMAGE (default: "{{ DOCKER_REGISTRY }}overhangio/openedx-xqueue:{{ TUTOR_VERSION }}")
  • XQUEUE_HOST (default: "xqueue.{{ LMS_HOST }}")
  • XQUEUE_MYSQL_PASSWORD (default: "{{ 8|random_string }}")
  • XQUEUE_MYSQL_DATABASE (default: "xqueue"
  • XQUEUE_MYSQL_USERNAME (default: "xqueue")
  • XQUEUE_SECRET_KEY (default: "{{ 24|random_string }}")

These values can be modified with tutor config save --set PARAM_NAME=VALUE commands.

License

This work is licensed under the terms of the GNU Affero General Public License (AGPL).