Allow users to configure a personal OpenAI API key

[Melhaya]

Having an API key per user will allow users to utilize the AI capabilities within CodeHarbor (i.e. Unittest generator)

• The generate unittest button is now always visible
• If key is invalid or missing, an alert is shown to the user
• If key is available and valid, a unittest will be created for the task

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.83%. Comparing base (df6988e) to head (dbc61fc).
Report is 7 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1496      +/-   ##
==========================================
+ Coverage   93.77%   93.83%   +0.06%     
==========================================
  Files         122      123       +1     
  Lines        2955     2985      +30     
==========================================
+ Hits         2771     2801      +30     
  Misses        184      184              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Melhaya]

I managed to add an API key text area for users to input their key. When I edit the profile and add a key and update, its still blank. I am not really sure what I need to do to make it persist.

@MrSerth @kkoehn : With the current changes I made, am I on the right track to have each user have their own API key? Once I can persist the update in the profile, the next step will be to actually use that for the unit test generator.

I also noticed that the schema.db had also changed again. do I need to turn it back to how it was? I only added the line t.string "personal_api_key" but noticed that the code also worked without it. I can remove that line.

Update: I was able to persist the API key in the user profile by adjusting the RegistrationsController file. Current challenge is to use the API key given by a user in the unit test generator. Not sure which file to adjust for this. need guidance on how to proceed. Thanks

[kkoehn]

I managed to add an API key text area for users to input their key. When I edit the profile and add a key and update, its still blank. I am not really sure what I need to do to make it persist.

@MrSerth @kkoehn : With the current changes I made, am I on the right track to have each user have their own API key? Once I can persist the update in the profile, the next step will be to actually use that for the unit test generator.

I also noticed that the schema.db had also changed again. do I need to turn it back to how it was? I only added the line t.string "personal_api_key" but noticed that the code also worked without it. I can remove that line.

Update: I was able to persist the API key in the user profile by adjusting the RegistrationsController file. Current challenge is to use the API key given by a user in the unit test generator. Not sure which file to adjust for this. need guidance on how to proceed. Thanks

Good work on figuring out the strong parameters!

To use user-specific API-Keys you will have to touch multiple points:

1. The task_policy should check for the user-key and not the "application-wide key"
2. The "application-wide key" and all remnants should be removed
3. Add the API-Key as parameter to your GptGenerateTests-Service
4. In your generate_test-Action in the tasks_controller supply the API-Key to the service-call

I hope this gives you a way forward, if you need more help, let me know!

[Melhaya]

Everything should now work as intended. There is only one 'issue' that I can't seem to find a solution for.

In the code, I am raising Gpt::InvalidApiKeyError if the user's API key is empty or invalid. the alert message is shown correctly when the API key is invalid. but for some reason, I get a You are not authorized for this action. alert when the API key is empty.

I figured out the issue is in the task_policy.rb

Currently the task policy says the following:

  def generate_test?
    user.present? and user.personal_openai_api_key.present? and update?
  end

When removing and user.personal_openai_api_key.present? , everything works and alerts are shown correctly. But now I can't seem to figure out the task_policy_spec.rb. I think I should remove most of the contexts there since the action of generating unit tests is permitted to any user. the alerts and checks are done somewhere else. I have pushed these changes now for you to see. If it is correct, then this should be ready to merge maybe with some minor cleanups that you might mention. :)

[kkoehn]

Thanks for the quick work with this PR. It looks good in general and seems to work fine, but I found some points that could be improved and had some questions.

[Melhaya]

@kkoehn I don't understand why the tests are currently failing. what changed?

Could you also help me resolve the conflict, please? I have updated my branch but the conflict still exists.

[kkoehn]

All of my review points have been addressed. Good work!