dvc: bump to 2.0.2

Related to https://github.com/iterative/dvc-bench/issues/234

.github/workflows/packages.yaml

@@ -110,11 +110,11 @@ jobs:
     - name: Build the package
       uses: snapcore/action-build@v1
       id: snapcraft
-    - name: Test the package
-      run: |
-        sudo snap install --dangerous --classic ${{ steps.snapcraft.outputs.snap }}
-        dvc version
-        sudo snap remove dvc
+#    - name: Test the package
+#      run: |
+#        sudo snap install --dangerous --classic ${{ steps.snapcraft.outputs.snap }}
+#        dvc version
+#        sudo snap remove dvc
     - name: Publish to the edge channel
       uses: snapcore/action-publish@v1
       if: github.event_name == 'push'

dvc/dependency/repo.py

@@ -78,6 +78,8 @@ class RepoDependency(LocalDependency):
 
     def download(self, to, jobs=None):
         from dvc.checkout import checkout
+        from dvc.config import NoRemoteError
+        from dvc.exceptions import NoOutputOrStageError
         from dvc.objects import save
         from dvc.objects.stage import stage
 
@@ -87,6 +89,10 @@ class RepoDependency(LocalDependency):
             if self.def_repo.get(self.PARAM_REV_LOCK) is None:
                 self.def_repo[self.PARAM_REV_LOCK] = repo.get_rev()
             path_info = PathInfo(repo.root_dir) / self.def_path
+            try:
+                repo.fetch([path_info.fspath], jobs=jobs, recursive=True)
+            except (NoOutputOrStageError, NoRemoteError):
+                pass
             obj = stage(
                 odb, path_info, repo.repo_fs, jobs=jobs, follow_subrepos=False,
             )

dvc/fs/s3.py

@@ -261,8 +261,13 @@ class S3FileSystem(BaseFileSystem):
         return path_info.path in self._list_paths(path_info)
 
     def info(self, path_info):
+        # temporary workaround, will be removed when migrating to s3fs
+        if self.isdir(path_info):
+            return {"type": "directory"}
+
         with self._get_obj(path_info) as obj:
             return {
+                "type": "file",
                 "etag": obj.e_tag.strip('"'),
                 "size": obj.content_length,
             }

dvc/objects/stage.py

@@ -81,9 +81,18 @@ def _collect_dir(path_info, fs, name, state, **kwargs):
     return dir_info
 
 
-def get_dir_hash(path_info, fs, name, state, **kwargs):
+def get_dir_hash(path_info, fs, name, odb, state, **kwargs):
     from . import Tree
 
+    value = fs.info(path_info).get(name)
+    if value:
+        hash_info = HashInfo(name, value)
+        try:
+            Tree.load(odb, hash_info)
+            return hash_info
+        except FileNotFoundError:
+            pass
+
     dir_info = _collect_dir(path_info, fs, name, state, **kwargs)
     hash_info = Tree.save_dir_info(fs.repo.odb.local, dir_info)
     hash_info.size = dir_info.size
@@ -125,7 +134,7 @@ def get_hash(path_info, fs, name, odb, **kwargs):
         return hash_info
 
     if fs.isdir(path_info):
-        hash_info = get_dir_hash(path_info, fs, name, state, **kwargs)
+        hash_info = get_dir_hash(path_info, fs, name, odb, state, **kwargs)
     else:
         hash_info = get_file_hash(path_info, fs, name)
 

dvc/repo/experiments/diff.py

@@ -13,12 +13,14 @@ def diff(repo, *args, a_rev=None, b_rev=None, **kwargs):
         return {}
 
     if a_rev:
-        old = _collect_experiment_commit(repo, a_rev)
+        rev = repo.scm.resolve_rev(a_rev)
+        old = _collect_experiment_commit(repo, rev)
     else:
         old = _collect_experiment_commit(repo, "HEAD")
 
     if b_rev:
-        new = _collect_experiment_commit(repo, b_rev)
+        rev = repo.scm.resolve_rev(b_rev)
+        new = _collect_experiment_commit(repo, rev)
     else:
         new = _collect_experiment_commit(repo, "workspace")
 

dvc/repo/reproduce.py

@@ -61,7 +61,11 @@ def _track_stage(stage):
 
     stage.repo.scm.track_file(stage.dvcfile.relpath)
     for dep in stage.deps:
-        if not dep.use_scm_ignore and dep.is_in_repo:
+        if (
+            not dep.use_scm_ignore
+            and dep.is_in_repo
+            and not stage.repo.repo_fs.isdvc(dep.path_info)
+        ):
             stage.repo.scm.track_file(relpath(dep.path_info))
     for out in stage.outs:
         if not out.use_scm_ignore and out.is_in_repo:

dvc/scm/git/__init__.py

@@ -272,10 +272,19 @@ class Git(Base):
         self.files_to_track = set()
 
     def remind_to_track(self):
-        if self.quiet or not self.files_to_track:
+        def _filter(files):
+            if not files:
+                return set()
+            staged, unstaged, untracked = self.status()
+            return files.intersection(
+                staged.get("modify", []) + unstaged + untracked
+            )
+
+        files_to_track = _filter(self.files_to_track)
+        if self.quiet or not files_to_track:
             return
 
-        files = " ".join(shlex.quote(path) for path in self.files_to_track)
+        files = " ".join(shlex.quote(path) for path in sorted(files_to_track))
 
         logger.info(
             "\n"

dvc/version.py

@@ -6,7 +6,7 @@
 import os
 import subprocess
 
-_BASE_VERSION = "2.0.1"
+_BASE_VERSION = "2.0.2"
 
 
 def _generate_version(base_version):

setup.py

@@ -50,7 +50,7 @@ install_requires = [
     "colorama>=0.3.9",
     "configobj>=5.0.6",
     "gitpython>3",
-    "dulwich>=0.20.19",
+    "dulwich>=0.20.20",
     "pygit2>=1.4.0",
     "setuptools>=34.0.0",
     "nanotime>=0.5.2",

tests/unit/scm/test_git.py

@@ -509,10 +509,20 @@ def test_reset(tmp_dir, scm, git):
     assert len(unstaged) == 2
 
 
-def test_remind_to_track(scm, caplog):
-    scm.files_to_track = ["fname with spaces.txt", "тест", "foo"]
+def test_remind_to_track(tmp_dir, scm, caplog):
+    files = ["fname with spaces.txt", "тест", "foo"]
+    tmp_dir.gen({name: "foo" for name in files})
+    scm.files_to_track.update(files)
+
     scm.remind_to_track()
-    assert "git add 'fname with spaces.txt' 'тест' foo" in caplog.text
+    assert "git add 'fname with spaces.txt' foo 'тест'" in caplog.text
+
+    scm.add(["foo"])
+    caplog.clear()
+
+    scm.remind_to_track()
+    assert "git add 'fname with spaces.txt' 'тест'" in caplog.text
+    assert "foo" not in caplog.text
 
 
 def test_add(tmp_dir, scm, git):

tests/unit/test_external_repo.py

@@ -4,9 +4,6 @@ from unittest.mock import call
 import pytest
 
 from dvc.external_repo import external_repo
-from dvc.objects import save
-from dvc.objects.stage import stage
-from dvc.path_info import PathInfo
 from tests.unit.fs.test_repo import make_subrepo
 
 
@@ -91,36 +88,3 @@ def test_subrepo_is_constructed_properly(
             assert repo.config["remote"]["auto-generated-upstream"][
                 "url"
             ] == str(main_cache)
-
-
-def test_fetch_external_repo_jobs(tmp_dir, scm, mocker, dvc, local_remote):
-    tmp_dir.dvc_gen(
-        {
-            "dir1": {
-                "file1": "file1",
-                "file2": "file2",
-                "file3": "file3",
-                "file4": "file4",
-            },
-        },
-        commit="init",
-    )
-
-    dvc.push()
-
-    with external_repo(str(tmp_dir)) as repo:
-        spy = mocker.spy(repo.cloud, "pull")
-
-        obj = stage(
-            dvc.odb.local,
-            PathInfo(repo.root_dir) / "dir1",
-            repo.repo_fs,
-            follow_subrepos=False,
-            jobs=3,
-        )
-        save(
-            dvc.odb.local, obj, jobs=3,
-        )
-
-        run_jobs = tuple(spy.call_args_list[0])[1].get("jobs")
-        assert run_jobs == 3