don't interpret HEAD as a faulty branch

[dimbleby]

with this requirement

caput = { git = "https://github.com/radiocosmology/caput.git", extras = ["compression"] }

and using dulwich as the git client, we get:

Failed to clone https://github.com/radiocosmology/caput.git at 'HEAD', verify ref exists on remote.

What seems to happen is we go through the branch

poetry/src/poetry/vcs/git/backend.py

Lines 69 to 75 in 16046d9

	elif (
	self.revision.encode("utf-8") in remote_refs.refs
	or f"refs/heads/{self.revision}".encode() in remote_refs.refs
	):
	# this is most likely a ref spec or a branch incorrectly specified
	self.branch = self.revision
	self.revision = None

, correcting "HEAD" to be treated as a branch with that name - and that later goes wrong.

I don't really understand what's going on in this code, maybe someone who does can contribute a testcase

Another fix that I tried was

--- a/src/poetry/puzzle/provider.py
+++ b/src/poetry/puzzle/provider.py
@@ -111,7 +111,7 @@ def _get_package_from_git(
     package = Provider.get_package_from_directory(path)
     package._source_type = "git"
     package._source_url = url
-    package._source_reference = rev or tag or branch or "HEAD"
+    package._source_reference = rev or tag or branch
     package._source_resolved_reference = revision
     package._source_subdirectory = subdirectory

which solved this case, but did quite a bit more damage to the unit tests.

[abn]

@dimbleby I have raised #7028 with a slightly different fix since it seems HEAD should really be treated as a branch from what I can tell. Added some test cases too. Your changes helped track down the issue fast! 🎉

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.