fix(data status): handle missing DVC repo at Git HEAD (#10862)

Avoid raising errors in `data_status()` when Git HEAD does not contain a DVC
repository. Instead, fall back to an empty index.

Tests now cover both unborn and committed Git repos, with and without subdirs.

Author: skshetry

dvc/repo/data.py

@@ -325,6 +325,7 @@ def _diff_head_to_index(
     granular: bool = False,
     with_renames: bool = False,
 ) -> DiffResult:
+    from dvc.exceptions import NotDvcRepoError
     from dvc.scm import RevError
     from dvc_data.index import DataIndex
 
@@ -338,6 +339,11 @@ def _diff_head_to_index(
     except RevError:
         logger.debug("failed to switch to '%s'", head)
         head_view = DataIndex()
+    except NotDvcRepoError as exc:
+        # NOTE: this only gets raised on subdir repos at the moment,
+        # which looks like a bug in `repo.switch`.
+        logger.warning(exc)
+        head_view = DataIndex()
 
     with ui.progress(desc="Calculating diff between head/index", unit="entry") as pb:
         return _diff(

tests/func/test_data_status.py

@@ -138,18 +138,25 @@ def test_tracked_directory_deep(M, tmp_dir, dvc, scm):
     }
 
 
-def test_new_empty_git_repo(M, tmp_dir, scm):
-    dvc = Repo.init()
+@pytest.mark.parametrize("git_repo_state", ["unborn", "committed"])
+@pytest.mark.parametrize("subdir", [True, False])
+def test_new_dvc_repo(M, tmp_dir, scm, subdir, git_repo_state):
+    if git_repo_state == "committed":
+        tmp_dir.scm_gen("test", "test", commit="init")
+
+    is_empty = git_repo_state == "unborn"
+    dir_ = tmp_dir / "sub" if subdir else tmp_dir
+    dvc = Repo.init(dir_, subdir=subdir)
     assert dvc.data_status() == {
         **EMPTY_STATUS,
-        "git": M.dict(is_empty=True, is_dirty=True),
+        "git": M.dict(is_dirty=True, is_empty=is_empty),
     }
 
-    tmp_dir.gen("foo", "foo")
-    dvc.add(["foo"])
+    dir_.gen("foo", "foo")
+    dvc.add([dir_ / "foo"])
     assert dvc.data_status() == {
         **EMPTY_STATUS,
-        "git": M.dict(is_empty=True, is_dirty=True),
+        "git": M.dict(is_empty=is_empty, is_dirty=True),
         "committed": {"added": ["foo"]},
     }