dvc: don't use realpath where not needed

dvc/config.py

@@ -103,7 +103,7 @@ class Config(dict):
         self.fs = fs or self.wfs
 
         if dvc_dir:
-            self.dvc_dir = self.fs.path.abspath(self.fs.path.realpath(dvc_dir))
+            self.dvc_dir = self.fs.path.abspath(dvc_dir)
 
         self.load(
             validate=validate, config=config, remote=remote, remote_config=remote_config

dvc/exceptions.py

@@ -71,6 +71,10 @@ class OutputNotFoundError(DvcException):
         )
 
 
+class StageNotFoundError(DvcException):
+    pass
+
+
 class StagePathAsOutputError(DvcException):
     """Thrown if directory that stage is going to be saved in is specified as
     an output of another stage.

dvc/repo/__init__.py

@@ -421,7 +421,7 @@ class Repo:
 
         fs = fs or localfs
         root = root or os.curdir
-        root_dir = fs.path.realpath(root)
+        root_dir = fs.path.abspath(root)
 
         if not fs.isdir(root_dir):
             raise NotDvcRepoError(f"directory '{root}' does not exist")

dvc/repo/experiments/executor/base.py

@@ -607,7 +607,7 @@ class BaseExecutor(ABC):
             old_cwd = os.getcwd()
 
             for path in copy_paths or []:
-                cls._copy_path(os.path.realpath(path), os.path.join(dvc.root_dir, path))
+                cls._copy_path(os.path.abspath(path), os.path.join(dvc.root_dir, path))
 
             if info.wdir:
                 os.chdir(os.path.join(dvc.scm.root_dir, info.wdir))

dvc/repo/graph.py

@@ -31,6 +31,9 @@ def check_acyclic(graph: "DiGraph") -> None:
 
 def get_pipeline(pipelines, node):
     found = [i for i in pipelines if i.has_node(node)]
+    if not found:
+        return None
+
     assert len(found) == 1
     return found[0]
 
@@ -60,6 +63,9 @@ def collect_pipeline(stage: "Stage", graph: "DiGraph") -> Iterator["Stage"]:
     import networkx as nx
 
     pipeline = get_pipeline(get_pipelines(graph), stage)
+    if not pipeline:
+        return iter([])
+
     return nx.dfs_postorder_nodes(pipeline, stage)
 
 

dvc/repo/init.py

@@ -37,7 +37,7 @@ def init(root_dir=os.curdir, no_scm=False, force=False, subdir=False):  # noqa:
             "Cannot initialize repo with `--no-scm` and `--subdir`"
         )
 
-    root_dir = os.path.realpath(root_dir)
+    root_dir = os.path.abspath(root_dir)
     dvc_dir = os.path.join(root_dir, Repo.DVC_DIR)
 
     try:

dvc/repo/stage.py

@@ -37,11 +37,19 @@ StageSet = Set["Stage"]
 
 
 def _collect_with_deps(stages: StageList, graph: "DiGraph") -> StageSet:
+    from dvc.exceptions import StageNotFoundError
     from dvc.repo.graph import collect_pipeline
 
     res: StageSet = set()
     for stage in stages:
-        res.update(collect_pipeline(stage, graph=graph))
+        pl = list(collect_pipeline(stage, graph=graph))
+        if not pl:
+            raise StageNotFoundError(
+                f"Stage {stage} is not found in the project. "
+                "Check that there are no symlinks in the parents "
+                "leading up to it within the project."
+            )
+        res.update(pl)
     return res
 
 
@@ -211,7 +219,7 @@ class StageLoad:
         self, path: Optional[str] = None, name: Optional[str] = None
     ) -> str:
         if path:
-            return self.repo.fs.path.realpath(path)
+            return self.repo.fs.path.abspath(path)
 
         path = PROJECT_FILE
         logger.debug("Assuming '%s' to be a stage inside '%s'", name, path)

dvc/stage/utils.py

@@ -32,14 +32,14 @@ def check_stage_path(repo, path, is_wdir=False):
         wdir_or_path="stage working dir" if is_wdir else "file path", path=path
     )
 
-    real_path = os.path.realpath(path)
+    real_path = os.path.abspath(path)
     if not os.path.exists(real_path):
         raise StagePathNotFoundError(error_msg.format("does not exist"))
 
     if not os.path.isdir(real_path):
         raise StagePathNotDirectoryError(error_msg.format("is not directory"))
 
-    proj_dir = os.path.realpath(repo.root_dir)
+    proj_dir = os.path.abspath(repo.root_dir)
     if real_path != proj_dir and not path_isin(real_path, proj_dir):
         raise StagePathOutsideError(error_msg.format("is outside of DVC repo"))
 

dvc/utils/__init__.py

@@ -218,23 +218,13 @@ def _visual_center(line, width):
 
 
 def relpath(path, start=os.curdir):
-    path = os.fspath(path)
+    path = os.path.abspath(os.fspath(path))
     start = os.path.abspath(os.fspath(start))
 
     # Windows path on different drive than curdir doesn't have relpath
-    if os.name == "nt":
-        # Since python 3.8 os.realpath resolves network shares to their UNC
-        # path. So, to be certain that relative paths correctly captured,
-        # we need to resolve to UNC path first. We resolve only the drive
-        # name so that we don't follow any 'real' symlinks on the path
-        def resolve_network_drive_windows(path_to_resolve):
-            drive, tail = os.path.splitdrive(path_to_resolve)
-            return os.path.join(os.path.realpath(drive), tail)
+    if os.name == "nt" and not os.path.commonprefix([start, path]):
+        return path
 
-        path = resolve_network_drive_windows(os.path.abspath(path))
-        start = resolve_network_drive_windows(start)
-        if not os.path.commonprefix([start, path]):
-            return path
     return os.path.relpath(path, start)
 
 

tests/func/test_stage.py

@@ -230,17 +230,31 @@ def test_parent_repo_collect_stages(tmp_dir, scm, dvc):
 
 @pytest.mark.parametrize("with_deps", (False, True))
 def test_collect_symlink(tmp_dir, dvc, with_deps):
+    from dvc.exceptions import StageNotFoundError
+
     tmp_dir.gen({"data": {"foo": "foo contents"}})
     foo_path = os.path.join("data", "foo")
     dvc.add(foo_path)
 
     data_link = tmp_dir / "data_link"
     data_link.symlink_to("data")
-    stage = list(
-        dvc.stage.collect(target=str(data_link / "foo.dvc"), with_deps=with_deps)
-    )[0]
 
-    assert stage.addressing == f"{foo_path}.dvc"
+    if with_deps:
+        # NOTE: with_deps means that we'll need to collect and use dvcfiles in the repo
+        # and we currently don't follow symlinks when collecting those, so it will not
+        # be able to find the target stage.
+        with pytest.raises(StageNotFoundError):
+            dvc.stage.collect(target=str(data_link / "foo.dvc"), with_deps=with_deps)
+    else:
+        stage = list(
+            dvc.stage.collect(target=str(data_link / "foo.dvc"), with_deps=with_deps)
+        )[0]
+
+        assert stage.addressing == os.path.join("data_link", "foo.dvc")
+
+    stage = list(dvc.stage.collect(target=f"{foo_path}.dvc", with_deps=with_deps))[0]
+
+    assert stage.addressing == os.path.join("data", "foo.dvc")
 
 
 def test_stage_strings_representation(tmp_dir, dvc, run_copy):

tests/unit/utils/test_utils.py

@@ -40,19 +40,8 @@ def test_fix_env_pyenv(path, orig):
 
 
 @pytest.mark.skipif(os.name != "nt", reason="Windows specific")
-def test_relpath_windows(monkeypatch):
-    """test that relpath correctly generated when run on a
-    windows network share. The drive mapped path is mapped
-    to a UNC path by os.path.realpath"""
-
-    def dummy_realpath(path):
-        return path.replace("x:", "\\\\server\\share")
-
-    monkeypatch.setattr(os.path, "realpath", dummy_realpath)
-    assert (
-        relpath("x:\\dir1\\dir2\\file.txt", "\\\\server\\share\\dir1")
-        == "dir2\\file.txt"
-    )
+def test_relpath_windows():
+    assert relpath("x:\\dir1\\dir2\\file.txt", "x:\\dir1") == "dir2\\file.txt"
 
     assert (
         relpath("y:\\dir1\\dir2\\file.txt", "\\\\server\\share\\dir1")