fix: strip credentials from the recorded git origin url

src/dvsim/utils/git.py

@@ -5,6 +5,7 @@
 """Git utility functions."""
 
 from pathlib import Path
+from urllib.parse import urlsplit, urlunsplit
 
 from git import Repo
 
@@ -13,6 +14,26 @@
 __all__ = ("repo_root",)
 
 
+def strip_url_credentials(url: str) -> str:
+    """Remove any username/password from a URL, keeping the rest of it intact.
+
+    A CI checkout leaves its access token in the remote URL as ``https://oauth2:<token>@host/org/repo``.
+    That URL is recorded in the run metadata and published in the reports, which get archived, so the
+    token would outlive the run that leaked it.
+
+    An ssh URL keeps its user: ``git@`` names an ssh account rather than an identity to authenticate
+    as, so dropping it gives a URL that no longer addresses the remote. A password is stripped
+    whatever the scheme carries it.
+    """
+    parts = urlsplit(url)
+    userinfo, _, host = parts.netloc.rpartition("@")
+    if not userinfo:
+        return url
+    if ":" not in userinfo and parts.scheme not in ("http", "https"):
+        return url
+    return urlunsplit((parts.scheme, host, parts.path, parts.query, parts.fragment))
+
+
 def repo_root(path: Path) -> Path | None:
     """Given a sub dir in a git repo provide the root path.
 
@@ -56,7 +77,11 @@ def git_is_dirty(path: Path | None = None) -> bool:
 
 
 def git_origin_url(path: Path | None = None) -> str | None:
-    """Get the git remote origin url, or None if no ``origin`` remote is configured."""
+    """Get the git remote origin url, or None if no ``origin`` remote is configured.
+
+    Any credentials the remote carries are stripped, so that the url is safe to record in run
+    metadata and reports.
+    """
     root = repo_root(path=path or Path.cwd())
 
     if root is None:
@@ -68,7 +93,7 @@ def git_origin_url(path: Path | None = None) -> str | None:
     if "origin" not in [remote.name for remote in r.remotes]:
         return None
 
-    return r.remote("origin").url
+    return strip_url_credentials(r.remote("origin").url)
 
 
 def git_https_url_with_commit(path: Path | None = None) -> str | None:

tests/utils/test_git.py

@@ -132,6 +132,66 @@ def test_git_origin_url(tmp_path: Path) -> None:
             equal_to(url),
         )
 
+    @staticmethod
+    @pytest.mark.parametrize(
+        ("url", "expected"),
+        [
+            (
+                "https://github.com/lowRISC/test.git",
+                "https://github.com/lowRISC/test.git",
+            ),
+            ("git@github.com:lowRISC/test.git", "git@github.com:lowRISC/test.git"),
+            # An ssh user is part of the address rather than a credential, so it stays. Dropping
+            # it would leave a url that no longer reaches the remote.
+            (
+                "ssh://git@github.com/lowRISC/test.git",
+                "ssh://git@github.com/lowRISC/test.git",
+            ),
+            (
+                "git+ssh://git@github.com/lowRISC/test.git",
+                "git+ssh://git@github.com/lowRISC/test.git",
+            ),
+            (
+                "https://oauth2:ghs_secrettoken@github.com/lowRISC/test.git",
+                "https://github.com/lowRISC/test.git",
+            ),
+            (
+                "https://someuser@github.com/lowRISC/test.git",
+                "https://github.com/lowRISC/test.git",
+            ),
+            # A password is a credential whatever the scheme carries it.
+            (
+                "ssh://user:secretpw@github.com/lowRISC/test.git",
+                "ssh://github.com/lowRISC/test.git",
+            ),
+        ],
+        ids=[
+            "plain_https",
+            "ssh_scp_form",
+            "ssh_url_form",
+            "git_ssh_scheme",
+            "token",
+            "user_only",
+            "ssh_with_password",
+        ],
+    )
+    def test_git_origin_url_strips_credentials(tmp_path: Path, url: str, expected: str) -> None:
+        """A token in the remote url never reaches the caller, whatever the url's shape.
+
+        The url is recorded in the run metadata and published in the reports, so a credential
+        left in it outlives the run. Both ssh forms carry an '@' without being credentialed, and
+        have to survive untouched or the recorded url stops addressing the remote.
+        """
+        r = Repo.init(path=tmp_path)
+
+        file = tmp_path / "a"
+        file.write_text("file to commit")
+        r.index.add([file])
+        r.index.commit("initial commit")
+        r.create_remote("origin", url)
+
+        assert_that(git_origin_url(tmp_path), equal_to(expected))
+
     @staticmethod
     @pytest.mark.parametrize(
         ("url", "expected"),
@@ -141,6 +201,10 @@ def test_git_origin_url(tmp_path: Path) -> None:
                 "https://github.com/lowRISC/test.git",
                 "https://github.com/lowRISC/test/tree/{commit}",
             ),
+            (
+                "https://oauth2:ghs_secrettoken@github.com/lowRISC/test.git",
+                "https://github.com/lowRISC/test/tree/{commit}",
+            ),
         ],
     )
     def test_git_https_url_with_commit(tmp_path: Path, url: str, expected: str) -> None: