data status: test cleanup and improvements (#8185)

dvc/repo/data.py

@@ -161,7 +161,6 @@ def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
         unstaged=unstaged,
         untracked=untracked,
         is_empty=empty_repo,
-        # TODO: fix is_dirty when untracked_files="no"
         is_dirty=any([staged, unstaged, untracked]),
     )
 
@@ -244,8 +243,8 @@ def _diff_head_to_index(
 
 class Status(TypedDict):
     not_in_cache: List[str]
-    committed: Dict[str, Any]
-    uncommitted: Dict[str, Any]
+    committed: Dict[str, List[str]]
+    uncommitted: Dict[str, List[str]]
     untracked: List[str]
     unchanged: List[str]
     git: GitInfo

tests/func/test_data_status.py

@@ -20,12 +20,10 @@ def test_file(M, tmp_dir, dvc, scm):
     remove(tmp_dir / "foo")
 
     expected = {
+        **EMPTY_STATUS,
         "committed": {"modified": ["foo"]},
         "uncommitted": {"deleted": ["foo"]},
         "git": M.dict(),
-        "not_in_cache": [],
-        "unchanged": [],
-        "untracked": [],
     }
     assert dvc.data_status() == expected
     assert dvc.data_status(granular=True) == expected
@@ -39,12 +37,10 @@ def test_directory(M, tmp_dir, dvc, scm):
     tmp_dir.gen("untracked", "untracked")
 
     assert dvc.data_status() == {
+        **EMPTY_STATUS,
         "committed": {"modified": [join("dir", "")]},
         "uncommitted": {"modified": [join("dir", "")]},
         "git": M.dict(),
-        "not_in_cache": [],
-        "unchanged": [],
-        "untracked": [],
     }
 
     assert dvc.data_status(granular=True, untracked_files="all") == {
@@ -123,8 +119,6 @@ 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"]}
-    # TODO: discuss what the output should be
     assert dvc.data_status() == EMPTY_STATUS
     assert dvc.data_status(granular=True) == EMPTY_STATUS
 
@@ -178,16 +172,10 @@ def test_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
     assert dvc.data_status(granular=True, untracked_files="all") == {
         **EMPTY_STATUS,
         "untracked": M.unordered("foobar.dvc", "dir.dvc", ".gitignore"),
-        # "committed": {
-        #   "added": M.unordered(
-        #       "foobar", join("dir", "foo"), join("dir", "bar")
-        # )},
         "committed": {"added": M.unordered("foobar", join("dir", ""))},
         "uncommitted": {
             "unknown": M.unordered(join("dir", "foo"), join("dir", "bar"))
         },
-        # even though we don't have cache obj, we can repurpose
-        # what we have in the workspace
         "not_in_cache": M.unordered(
             "foobar",
             join("dir", ""),
@@ -230,7 +218,6 @@ def test_git_committed_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
         **EMPTY_STATUS,
         "not_in_cache": M.unordered("foobar", join("dir", "")),
         "git": M.dict(),
-        # TODO: clarify what is unchanged
         "unchanged": M.unordered("foobar", join("dir", "")),
     }
     assert dvc.data_status(granular=True) == {
@@ -243,7 +230,6 @@ def test_git_committed_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
             "unknown": M.unordered(join("dir", "foo"), join("dir", "bar"))
         },
         "git": M.dict(),
-        # TODO: clarify what is unchanged
         "unchanged": M.unordered("foobar", join("dir", "")),
     }
 

tests/unit/command/test_data_status.py

@@ -14,15 +14,21 @@ from tests.func.parsing.test_errors import escape_ansi
 def mocked_status():
     yield Status(
         not_in_cache=["notincache"],
-        committed={"added": ["dir/bar", "dir/foo"]},
+        committed={
+            "added": ["dir/bar", "dir/foo"],
+            "deleted": ["dir/baz"],
+            "modified": ["dir/foobar"],
+            "unknown": ["dir/unknown1"],
+        },
         uncommitted={
             "added": ["dir/baz"],
             "modified": ["dir/bar"],
             "deleted": ["dir/foobar"],
+            "unknown": ["dir2/unknown2"],
         },
         untracked=["untracked"],
         unchanged=["dir/foo"],
-        git={"is_dirty": True},
+        git={"is_dirty": True, "is_empty": False},
     )
 
 
@@ -113,6 +119,8 @@ DVC committed changes:
   (git commit the corresponding dvc files to update the repo)
         added: dir/bar
         added: dir/foo
+        deleted: dir/baz
+        modified: dir/foobar
 
 DVC uncommitted changes:
   (use "dvc commit <file>..." to track changes)