clear-pr-merge-commit-message - Preserve "Signed-off-by" lines

[agriyakhetarpal]

Hi! This PR addresses one part of #9330 by not removing "Signed-off-by" lines (which are usually required by DCO). I've added some test cases too, which were fairly straightforward.

Test URLs

Tested on Dependabot's PRs opened on one of the projects I help maintain: pyodide/pyodide-cli#60

Screenshot

[fregante]

Thank you!

[fregante]

As for the first Co-authored-by line, I always found that to be noise, together with what Copilot does. In a future PR I'd probably like to detect them and drop them on PRs that are already authored by them. What do you think?

[agriyakhetarpal]

As for the first Co-authored-by line, I always found that to be noise, together with what Copilot does. In a future PR I'd probably like to detect them and drop them on PRs that are already authored by them. What do you think?

I agree with you, and I will gladly take up your offer to contribute further and put together more PRs! :) Do we have a helper somewhere to find who the author of a PR is?

Edit: I'm guessing yes, since one of the features IIRC, does check and adds "Now you can release this change"

Edit 2: okay, we can probably export this:

refined-github/source/github-helpers/index.ts

Lines 211 to 216 in 78ebd32

	function getConversationAuthor(): string | undefined {
	return $optional([
	'.js-command-palette-pull-body .author', // PR conversation
	'[data-testid="issue-body-header-author"]', // Issue conversation
	])?.textContent;
	}

and then pass the author name into cleanCommitMessage.

[fregante]

Yes, so I think it's only safe to handle these two bots explicitly. There's no straightforward way to match the email to the user without API. There's only exception would be those "privacy emails" that do include the username. We can also drop those if they match the PR author.

[agriyakhetarpal]

Okay, sounds great. If I understand correctly, privacy emails of the format digits+username@users.noreply.github.com by extracting username should cover all cases, including Dependabot and Copilot, because I can see Dependabot uses 49699333+dependabot[bot]@users.noreply.github.com. It's the same for pre-commit.ci (pre-commit-ci[bot]), Renovate (renovate[bot]), and automated PRs by github-actions[bot] as well, as they also have privacy emails, AFAICS. So I'd say we don't need any explicit bot names?

[fregante]

If that's the case, then yes.

1. Get PR author
2. Remove Co-authored-by lines that match the PR author

That's all.