Combine the data contained in multiple Feedback activities into a unified CSV. For use with our Internships courses in Moodle which have identically structured feedback activities with data we want to transfer to other systems for analysis and display.
poetry install to get python dependencies.
See Moodle's Web Services Overview for their outline of setting up an API user. Normally, we would create a user of the "Web Services" authentication type, put it in the Web Services User role, and give that role the needed capabilities in the right context. However, after doing that, calls to the mod_feedback_get_analysis function still failed with a required_capability_exception. So below we use an account that is one of the site administrators.
- Create a custom service for the app
- Enabled for Authorized users only
- Required capability:
mod/feedback:viewreports
- Add all necessary web service functions to the service (basically, every REST API endpoint the script calls)
- There's a list of functions in the API Documentation
core_course_get_courses_by_fieldmod_feedback_get_feedbacks_by_coursesmod_feedback_get_responses_analysis
- Add a site admin user as an authorized user of the service
- Create a token for the user, select the service, optionally set an expiration date if prudent. You may need to jump to the final page of tokens to see the one you just created.
- Copy example.env to .env and add your values (token, domain, category ID, course IDs to ignore)
If you have configured the .env file as instructed above, simply poetry run python app.py collates feedbacks from all courses in the Internships category and outputs CSV files. If you want to test feedback responses from a single course, you can edit app.py like so:
if __name__ == "__main__":
# courses = get_courses()
feedbacks = get_feedbacks([{"id": 12345}]) # where 12345 is the course IDOnly one of mod_feedback_get_analysis or mod_feedback_get_responses_analysis is needed; I opted for get_responses_analysis. The former gives more information about how the questions are structured and then puts all responses in a data array under the question with little context, while the latter is more focused on the response data, collecting all attempts into an array where each response contains the name of the field and two different values ("print" versus "raw"). We use the "raw" value for all fields except Connection where we use "print" because "raw" is an uninformative integer.