feat(ncu-ci): parse comments to find a safe commit

[aduh95]

Currently, the request-ci only works for approved PRs. This is annoying because:

• collaborators cannot approve their own PRs, and have to wait for someone else to approve their PR for them, or start the CI manually.
• it gives an incentive to approve a PR just to run the CI rather than because the collaborator actually supports it.

There have been suggestions to use PR author, or PR head repo owner, or commit signature, or a combination of the three – but I don't really like that idea, because:

• it doesn't help for PRs from non-collaborators.
• we should care about who is requesting the CI, not who's proposing the change.

IIUC comments are not editable by triagers, so should be a safe way to determine a safe SHA for unapproved PRs.

[bjohansebas]

IIUC comments are not editable by triagers, so should be a safe way to determine a safe SHA for unapproved PRs.

yes, only people with write access have the ability to edit comments

[targos]

Would you like a review before updating the tests?

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 79.82%. Comparing base (ef135f5) to head (733bd44). Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #920      +/-   ##
==========================================
+ Coverage   79.70%   79.82%   +0.12%     
==========================================
  Files          39       39              
  Lines        4642     4650       +8     
==========================================
+ Hits         3700     3712      +12     
+ Misses        942      938       -4     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

:rocket: New features to boost your workflow:

• :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• :package: JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[aduh95]

AFAICT, this allows to request CI for a commit SHA that is not part of the pull request. Is that a bug or a feature?

The commit SHA provided by the collaborator with be passed as COMMIT_SHA_CHECK , Jenkins checks that GIT_REMOTE_REF (i.e. refs/pull/${prID}/head) resolves to that expected commit SHA, otherwise the CI fails. So if someone pass the SHA of a commit not in the PR branch, it won't match the tip of the PR head, and it won't work