dvc: add: set top level md5 to None (#3846)

This way it won't be included into produced dvcfiles (we don't need it
for `dvc add`).

dvc/stage/__init__.py

@@ -356,7 +356,11 @@ class Stage(params.StageParams):
         return get_dump(self)
 
     def compute_md5(self):
-        m = compute_md5(self)
+        # `dvc add`ed files don't need stage md5
+        if self.is_data_source and not (self.is_import or self.is_repo_import):
+            m = None
+        else:
+            m = compute_md5(self)
         logger.debug(f"Computed {self} md5: '{m}'")
         return m
 

tests/func/test_add.py

@@ -44,10 +44,9 @@ def test_add(tmp_dir, dvc):
     assert len(stage.deps) == 0
     assert stage.cmd is None
     assert stage.outs[0].info["md5"] == md5
-    assert stage.md5 == "411854a9fc55b3cebda28704e41b23b3"
+    assert stage.md5 is None
 
     assert load_stage_file("foo.dvc") == {
-        "md5": "411854a9fc55b3cebda28704e41b23b3",
         "outs": [{"md5": "acbd18db4cc2f85cedef654fccc4a4d8", "path": "foo"}],
     }
 

tests/func/test_commit.py

@@ -70,6 +70,7 @@ def test_commit_changed_md5(tmp_dir, dvc):
         dvc.commit(stage.path)
 
     dvc.commit(stage.path, force=True)
+    assert "md5" not in load_stage_file(stage.path)
 
 
 def test_commit_no_exec(tmp_dir, dvc):

tests/unit/stage/test_stage.py

@@ -20,14 +20,14 @@ TEST_STAGE_DICT = {
 
 
 def test_stage_checksum():
-    stage = Stage(None, "path")
+    stage = Stage(None, "path", cmd="mycmd")
 
     with mock.patch.object(stage, "dumpd", return_value=TEST_STAGE_DICT):
         assert stage.compute_md5() == "e9521a22111493406ea64a88cda63e0b"
 
 
 def test_wdir_default_ignored():
-    stage = Stage(None, "path")
+    stage = Stage(None, "path", cmd="mycmd")
     d = dict(TEST_STAGE_DICT, wdir=".")
 
     with mock.patch.object(stage, "dumpd", return_value=d):
@@ -35,7 +35,7 @@ def test_wdir_default_ignored():
 
 
 def test_wdir_non_default_is_not_ignored():
-    stage = Stage(None, "path")
+    stage = Stage(None, "path", cmd="mycmd")
     d = dict(TEST_STAGE_DICT, wdir="..")
 
     with mock.patch.object(stage, "dumpd", return_value=d):
@@ -43,7 +43,7 @@ def test_wdir_non_default_is_not_ignored():
 
 
 def test_meta_ignored():
-    stage = Stage(None, "path")
+    stage = Stage(None, "path", cmd="mycmd")
     d = dict(TEST_STAGE_DICT, meta={"author": "Suor"})
 
     with mock.patch.object(stage, "dumpd", return_value=d):