From 2374c721055efc73696a28a39e96656c36ec2985 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Saugat=20Pachhai=20=28=E0=A4=B8=E0=A5=8C=E0=A4=97=E0=A4=BE?=
 =?UTF-8?q?=E0=A4=A4=29?= <suagatchhetri@outlook.com>
Date: Wed, 24 Aug 2022 19:12:05 +0545
Subject: [PATCH] status: add unknown state

---
 dvc/commands/data.py           |  1 +
 dvc/repo/data.py               | 20 +++++++++++++++++---
 tests/func/test_data_status.py | 13 ++++++++++---
 3 files changed, 28 insertions(+), 6 deletions(-)

diff --git a/dvc/commands/data.py b/dvc/commands/data.py
index 6a98acaeb..e26899fc4 100644
--- a/dvc/commands/data.py
+++ b/dvc/commands/data.py
@@ -51,6 +51,7 @@ class CmdDataStatus(CmdBase):
                     file: state
                     for state, files in stage_status.items()
                     for file in files
+                    if state != "unknown"
                 }
             if not items:
                 continue
diff --git a/dvc/repo/data.py b/dvc/repo/data.py
index 4c5311a48..26884c25b 100644
--- a/dvc/repo/data.py
+++ b/dvc/repo/data.py
@@ -99,8 +99,20 @@ def _granular_diff(
     return dict(output)
 
 
+def _get_obj_items(root: str, obj: Optional["HashFile"]) -> List[str]:
+    if not obj:
+        return []
+
+    from dvc_data.objects.tree import Tree
+
+    sep = os.path.sep
+    if isinstance(obj, Tree):
+        return [sep.join([root, *key]) for key, _, _ in obj]
+    return [root]
+
+
 def _diff(
-    root,
+    root: str,
     old_oid: Optional["HashInfo"],
     old_obj: Optional["HashFile"],
     new_oid: Optional["HashInfo"],
@@ -108,12 +120,14 @@ def _diff(
     odb: "HashFileDB",
     with_dirs: bool = False,
     granular: bool = False,
-):
+) -> Dict[str, List[str]]:
     if not granular:
         return _shallow_diff(root, old_oid, new_oid, odb)
     if (old_oid and not old_obj) or (new_oid and not new_obj):
         # we don't have enough information to give full details
-        return _shallow_diff(root, old_oid, new_oid, odb)
+        unknown = _get_obj_items(root, new_obj or old_obj)
+        shallow_diff = _shallow_diff(root, old_oid, new_oid, odb)
+        return {**shallow_diff, "unknown": unknown}
     return _granular_diff(root, old_obj, new_obj, odb, with_dirs=with_dirs)
 
 
diff --git a/tests/func/test_data_status.py b/tests/func/test_data_status.py
index 47b54d979..89f9a15a7 100644
--- a/tests/func/test_data_status.py
+++ b/tests/func/test_data_status.py
@@ -171,7 +171,6 @@ def test_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
         **EMPTY_STATUS,
         "untracked": M.unordered("foobar.dvc", "dir.dvc", ".gitignore"),
         "committed": {"added": M.unordered("foobar", join("dir", ""))},
-        "uncommitted": {},
         "not_in_cache": M.unordered("foobar", join("dir", "")),
         "git": M.dict(),
     }
@@ -184,7 +183,9 @@ def test_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
         #       "foobar", join("dir", "foo"), join("dir", "bar")
         # )},
         "committed": {"added": M.unordered("foobar", join("dir", ""))},
-        "uncommitted": {},
+        "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(
@@ -238,6 +239,9 @@ def test_git_committed_missing_cache_workspace_exists(M, tmp_dir, dvc, scm):
             "foobar",
             join("dir", ""),
         ),
+        "uncommitted": {
+            "unknown": M.unordered(join("dir", "foo"), join("dir", "bar"))
+        },
         "git": M.dict(),
         # TODO: clarify what is unchanged
         "unchanged": M.unordered("foobar", join("dir", "")),
@@ -303,7 +307,10 @@ def test_missing_dir_object_from_head(M, tmp_dir, dvc, scm):
     }
     assert dvc.data_status(granular=True) == {
         **EMPTY_STATUS,
-        "committed": {"modified": [join("dir", "")]},
+        "committed": {
+            "modified": [join("dir", "")],
+            "unknown": [join("dir", "foobar")],
+        },
         "git": M.dict(),
     }