Merge pull request iterative#3151 from efiop/3087

import: don't forget to set rev_lock for git files

dvc/dependency/repo.py

@@ -11,6 +11,7 @@
 from dvc.exceptions import OutputNotFoundError
 from dvc.exceptions import PathMissingError
 from dvc.utils.fs import fs_copy
+from dvc.scm import SCM
 
 
 class DependencyREPO(DependencyLOCAL):
@@ -105,6 +106,7 @@ def _copy_if_git_file(self, to_path):
         src_full_path = os.path.join(repo_dir, src_path)
         dst_full_path = os.path.abspath(to_path)
         fs_copy(src_full_path, dst_full_path)
+        self.def_repo[self.PARAM_REV_LOCK] = SCM(repo_dir).get_rev()
         return True
 
     def download(self, to):

tests/func/test_import.py

@@ -21,11 +21,15 @@ def test_import(tmp_dir, scm, dvc, erepo_dir):
     with erepo_dir.chdir():
         erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
 
-    dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
+    stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
 
     assert os.path.isfile("foo_imported")
     assert (tmp_dir / "foo_imported").read_text() == "foo content"
     assert scm.repo.git.check_ignore("foo_imported")
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev_lock": erepo_dir.scm.get_rev(),
+    }
 
 
 @pytest.mark.parametrize("src_is_dvc", [True, False])
@@ -39,13 +43,17 @@ def test_import_git_file(erepo_dir, tmp_dir, dvc, scm, src_is_dvc):
 
     erepo_dir.scm_gen({src: "hello"}, commit="add a regular file")
 
-    tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
+    stage = tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
 
     assert (tmp_dir / dst).is_file()
     assert filecmp.cmp(
         fspath(erepo_dir / src), fspath(tmp_dir / dst), shallow=False
     )
     assert tmp_dir.scm.repo.git.check_ignore(fspath(tmp_dir / dst))
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev_lock": erepo_dir.scm.get_rev(),
+    }
 
 
 @pytest.mark.parametrize("src_is_dvc", [True, False])
@@ -59,22 +67,30 @@ def test_import_git_dir(erepo_dir, tmp_dir, dvc, scm, src_is_dvc):
 
     erepo_dir.scm_gen({src: {"file.txt": "hello"}}, commit="add a dir")
 
-    tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
+    stage = dvc.imp(fspath(erepo_dir), src, dst)
 
     assert (tmp_dir / dst).is_dir()
     trees_equal(fspath(erepo_dir / src), fspath(tmp_dir / dst))
     assert tmp_dir.scm.repo.git.check_ignore(fspath(tmp_dir / dst))
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev_lock": erepo_dir.scm.get_rev(),
+    }
 
 
 def test_import_dir(tmp_dir, scm, dvc, erepo_dir):
     with erepo_dir.chdir():
         erepo_dir.dvc_gen({"dir": {"foo": "foo content"}}, commit="create dir")
 
-    dvc.imp(fspath(erepo_dir), "dir", "dir_imported")
+    stage = dvc.imp(fspath(erepo_dir), "dir", "dir_imported")
 
     assert os.path.isdir("dir_imported")
     trees_equal(fspath(erepo_dir / "dir"), "dir_imported")
     assert scm.repo.git.check_ignore("dir_imported")
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev_lock": erepo_dir.scm.get_rev(),
+    }
 
 
 def test_import_non_cached(erepo_dir, tmp_dir, dvc, scm):
@@ -90,23 +106,34 @@ def test_import_non_cached(erepo_dir, tmp_dir, dvc, scm):
     erepo_dir.scm.add([fspath(erepo_dir / src)])
     erepo_dir.scm.commit("add a non-cached output")
 
-    tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
+    stage = tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
 
     assert (tmp_dir / dst).is_file()
     assert filecmp.cmp(
         fspath(erepo_dir / src), fspath(tmp_dir / dst), shallow=False
     )
     assert tmp_dir.scm.repo.git.check_ignore(dst)
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev_lock": erepo_dir.scm.get_rev(),
+    }
 
 
 def test_import_rev(tmp_dir, scm, dvc, erepo_dir):
+    rev = None
     with erepo_dir.chdir(), erepo_dir.branch("branch", new=True):
         erepo_dir.dvc_gen("foo", "foo content", commit="create foo on branch")
+        rev = erepo_dir.scm.get_rev()
 
-    dvc.imp(fspath(erepo_dir), "foo", "foo_imported", rev="branch")
+    stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported", rev="branch")
 
     assert (tmp_dir / "foo_imported").read_text() == "foo content"
     assert scm.repo.git.check_ignore("foo_imported")
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev": "branch",
+        "rev_lock": rev,
+    }
 
 
 def test_pull_imported_stage(tmp_dir, dvc, erepo_dir):

tests/func/test_update.py

@@ -1,19 +1,36 @@
+import pytest
+
+from dvc.stage import Stage
 from dvc.compat import fspath
 from dvc.external_repo import clean_repos
 
 
-def test_update_import(tmp_dir, dvc, erepo_dir):
+@pytest.mark.parametrize("cached", [True, False])
+def test_update_import(tmp_dir, dvc, erepo_dir, cached):
+    old_rev = None
     with erepo_dir.branch("branch", new=True), erepo_dir.chdir():
-        erepo_dir.dvc_gen("version", "branch", "add version file")
+        gen = erepo_dir.dvc_gen if cached else erepo_dir.scm_gen
+        gen("version", "branch", "add version file")
+        old_rev = erepo_dir.scm.get_rev()
 
     stage = dvc.imp(fspath(erepo_dir), "version", "version", rev="branch")
 
     imported = tmp_dir / "version"
     assert imported.is_file()
     assert imported.read_text() == "branch"
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev": "branch",
+        "rev_lock": old_rev,
+    }
 
+    new_rev = None
     with erepo_dir.branch("branch", new=False), erepo_dir.chdir():
-        erepo_dir.dvc_gen("version", "updated", "update version content")
+        gen = erepo_dir.dvc_gen if cached else erepo_dir.scm_gen
+        gen("version", "updated", "update version content")
+        new_rev = erepo_dir.scm.get_rev()
+
+    assert old_rev != new_rev
 
     # Caching in external repos doesn't see upstream updates within single
     # cli call, so we need to clean the caches to see the changes.
@@ -26,6 +43,13 @@ def test_update_import(tmp_dir, dvc, erepo_dir):
     assert imported.is_file()
     assert imported.read_text() == "updated"
 
+    stage = Stage.load(dvc, stage.path)
+    assert stage.deps[0].def_repo == {
+        "url": fspath(erepo_dir),
+        "rev": "branch",
+        "rev_lock": new_rev,
+    }
+
 
 def test_update_import_url(tmp_dir, dvc, tmp_path_factory):
     import_src = tmp_path_factory.mktemp("import_url_source")