data status: fix untracked files output for subdir

dvc/repo/data.py

@@ -215,6 +215,17 @@ class Status(TypedDict):
     git: GitInfo
 
 
+def _transform_git_paths_to_dvc(repo: "Repo", files: List[str]):
+    """Transform files rel. to Git root to DVC root, and drop outside files."""
+    rel = repo.fs.path.relpath(repo.root_dir, repo.scm.root_dir).rstrip("/")
+    if rel in (os.curdir, ""):
+        return files
+
+    prefix = rel + os.sep
+    length = len(prefix)
+    return [file[length:] for file in files if file.startswith(prefix)]
+
+
 def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
     from scmrepo.exceptions import SCMError
 
@@ -235,12 +246,14 @@ def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
         unchanged &= set(committed_diff.pop("unchanged", []))
 
     git_info = _git_info(repo.scm, untracked_files=untracked_files)
+    untracked = git_info.get("untracked", [])
+    untracked = _transform_git_paths_to_dvc(repo, untracked)
     # order matters here
     return Status(
         not_in_cache=not_in_cache,
         committed=committed_diff,
         uncommitted=uncommitted_diff,
-        untracked=git_info.get("untracked", []),
+        untracked=untracked,
         unchanged=list(unchanged),
         git=git_info,
     )

tests/func/test_data_status.py

@@ -1,6 +1,7 @@
 from os.path import join
 
 from dvc.repo import Repo
+from dvc.testing.tmp_dir import make_subrepo
 from dvc.utils.fs import remove
 
 EMPTY_STATUS = {
@@ -146,3 +147,21 @@ def test_skip_uncached_pipeline_outputs(tmp_dir, dvc, run_copy_metrics):
 def test_output_with_newly_added_stage(tmp_dir, dvc):
     dvc.stage.add(deps=["bar"], outs=["foo"], name="copy", cmd="cp foo bar")
     assert dvc.data_status() == {**EMPTY_STATUS, "not_in_cache": ["foo"]}
+
+
+def test_subdir(M, tmp_dir, scm):
+    subrepo = tmp_dir / "sub"
+    make_subrepo(subrepo, scm)
+
+    with subrepo.chdir():
+        subrepo.dvc_gen({"dir": {"foo": "foo"}}, commit="add dir")
+        subrepo.dvc_gen("bar", "bar", commit="add foo")
+        subrepo.gen("untracked", "untracked")
+
+        dvc = subrepo.dvc
+        assert dvc.data_status(granular=True, untracked_files="all") == {
+            **EMPTY_STATUS,
+            "git": M.instance_of(dict),
+            "unchanged": M.unordered("bar", join("dir", "foo")),
+            "untracked": ["untracked"],
+        }