dulwich: fix clone with file:// protocol on windows

- clone: add test for file://<path> and shallow_branch

Author: pmrowla

scmrepo/git/backend/dulwich/__init__.py

@@ -159,6 +159,14 @@ def clone(
         from dulwich.porcelain import NoneStream
         from dulwich.porcelain import clone as git_clone
 
+        parsed = urlparse(url)
+        if os.name == "nt":
+            # NOTE: Dulwich client code doesn't handle mixed path seps when
+            # passed a file:// URL (i.e. file://C:\foo\bar) but we can just
+            # strip the prefix since Dulwich uses the same clone behavior
+            # for both paths and file:// URLs (as opposed to C-git)
+            if parsed.scheme == "file":
+                url = url[7:]
         try:
             clone_from = partial(
                 git_clone,
@@ -174,11 +182,10 @@ def clone(
                 # NOTE: dulwich only supports shallow/depth for non-local
                 # clones. This differs from CLI git, where depth is used for
                 # file:// URLs but not direct local paths
-                parsed = urlparse(url)
-                if not parsed.scheme or parsed.scheme == "file":
-                    depth = 0
-                else:
+                if parsed.scheme in ("git", "git+ssh", "ssh", "http", "https"):
                     depth = 1
+                else:
+                    depth = 0
                 repo = clone_from(
                     depth=depth, branch=os.fsencode(shallow_branch)
                 )

tests/test_git.py

@@ -1,6 +1,6 @@
 import os
 import shutil
-from typing import Any, Dict, Iterator, Type
+from typing import Any, Dict, Iterator, Optional, Type
 
 import pytest
 from asyncssh import SFTPClient
@@ -885,15 +885,24 @@ async def test_git_ssh(
     assert (tmp_dir / "foo").read_text() == "foo"
 
 
+@pytest.mark.parametrize("scheme", ["", "file://"])
+@pytest.mark.parametrize("shallow_branch", [None, "master"])
 def test_clone(
-    tmp_dir: TmpDir, scm: Git, git: Git, tmp_dir_factory: TempDirFactory
+    tmp_dir: TmpDir,
+    scm: Git,
+    git: Git,
+    tmp_dir_factory: TempDirFactory,
+    scheme: str,
+    shallow_branch: Optional[str],
 ):
     tmp_dir.gen("foo", "foo")
     scm.add_commit("foo", message="init")
     rev = scm.get_rev()
 
     target_dir = tmp_dir_factory.mktemp("git-clone")
-    git.clone(str(tmp_dir), (target_dir))
+    git.clone(
+        f"{scheme}{tmp_dir}", str(target_dir), shallow_branch=shallow_branch
+    )
     target = Git(str(target_dir))
     assert target.get_rev() == rev
     assert (target_dir / "foo").read_text() == "foo"