PIP 176: Refactor Doc Bot

PIP

PIP 176: Refactor Doc Bot to Improve Efficiency of Managing Doc

Issues (new)

• P0: doc bot check is pending or waiting and documentation label check is changed to label check after it is set to required.

This blocks many PRs, eg. https://github.com/apache/pulsar/pull/16298, https://github.com/apache/pulsar/pull/16023

This might be caused by GitHub Action limitations https://stackoverflow.com/questions/52200096/github-pull-request-waiting-for-status-to-be-reported

Can you please take a look at it at your earliest convenience? Thank you!

• P1: when the PR description is empty and the PR author selects labels from Labels, the doc bot should work normally.

• P2: configure the doc bot to other branches except for master - [x] 2.10. Obsever the results and set for other branches if it works well - [ ] 2.9 - [ ] 2.8 - [ ] 2.7

Issues can’t be resolved

• P2: Doc Bot is triggered after updating PR title. (can not be resolved now due to GitHub limitations)

Issues (old)

When fixing https://github.com/apache/pulsar/issues/15732, @maxsxu found that the doc Bot was implemented by modifying dist rather than src, which makes it hard to maintain and scale.

Consequently, he suggests refactoring the doc Bot along with resolving the following issues in Bot:

Scenario 1

Issue

This PR was marked with doc-related labels and then removed. No doc-related label is marked for this PR.

Expected behavior

In this case, this PR should be labeled w/ doc-label-missing.

Scenario 2

Issue

Doc Bot only works for master change.

Expected behavior

Doc Bot should work for all branches and label all PRs.

Scenario 3

Issue

Bot labels this PR with several doc labels.

Expected behavior

Bot should label a PR with only a doc label.

Improvements

Set documentation label check as one of the GitHub checks.

References

Pulsar Bot

• Instructions
• Implementation process
• Flowchart
• Script file

[Anonymitaet]

@maxsxu I’ve added more improvements as we discussed just now

[maxsxu]

Thanks @Anonymitaet brings this out.

More context from the previous maintainer:

Current doc-label-check inspired from https://github.com/harupy/auto-labeling. During the doc bot development, just been modifying the compiled dist/index.js to quickly satisfy requirements.

So from all perspectives, refactoring is necessary. I’ll be working on this to make the doc Bot more maintainable and scalable.