From 3427e115ebadaa16dad63afdfd0517bd7b32667d Mon Sep 17 00:00:00 2001
From: skshetry <18718008+skshetry@users.noreply.github.com>
Date: Wed, 25 Jan 2023 18:09:38 +0545
Subject: [PATCH] data status: check if change.old.hash_info is None or not
 (#8878)

data:status: check if change.old.hash_info is None or not
---
 dvc/repo/data.py               |  1 +
 tests/func/test_data_status.py | 28 +++++++++++++++++++++++++---
 2 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/dvc/repo/data.py b/dvc/repo/data.py
index 7b035eb23..2dbd40575 100644
--- a/dvc/repo/data.py
+++ b/dvc/repo/data.py
@@ -93,6 +93,7 @@ def _diff(
         if (
             with_missing
             and change.old
+            and change.old.hash_info
             and not change.old.odb.exists(change.old.hash_info.value)
         ):
             # NOTE: emulating previous behaviour
diff --git a/tests/func/test_data_status.py b/tests/func/test_data_status.py
index 6793f9fce..87d0b3206 100644
--- a/tests/func/test_data_status.py
+++ b/tests/func/test_data_status.py
@@ -176,10 +176,32 @@ def test_skip_uncached_pipeline_outputs(tmp_dir, dvc, run_copy_metrics):
     )
 
 
-def test_output_with_newly_added_stage(tmp_dir, dvc):
+def test_outs_with_no_hashes(M, tmp_dir, dvc, scm):
+    dvc.stage.add(single_stage=True, outs=["bar"])
     dvc.stage.add(deps=["bar"], outs=["foo"], name="copy", cmd="cp foo bar")
-    assert dvc.data_status() == EMPTY_STATUS
-    assert dvc.data_status(granular=True) == EMPTY_STATUS
+
+    expected_output = {
+        **EMPTY_STATUS,
+        "committed": {"added": M.unordered("bar", "foo")},
+        "git": M.dict(),
+    }
+    assert dvc.data_status() == expected_output
+    assert dvc.data_status(granular=True) == expected_output
+
+
+def test_outs_with_no_hashes_and_with_uncommitted_files(M, tmp_dir, dvc, scm):
+    tmp_dir.gen({"bar": "bar", "foo": "foo"})
+    dvc.stage.add(single_stage=True, outs=["bar"])
+    dvc.stage.add(deps=["bar"], outs=["foo"], name="copy", cmd="cp foo bar")
+
+    expected_output = {
+        **EMPTY_STATUS,
+        "uncommitted": {"added": M.unordered("bar", "foo")},
+        "committed": {"added": M.unordered("bar", "foo")},
+        "git": M.dict(),
+    }
+    assert dvc.data_status() == expected_output
+    assert dvc.data_status(granular=True) == expected_output
 
 
 def test_subdir(M, tmp_dir, scm):