From 3adede8427d7eec96fe79afab83af1b4f1992b3d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Daniele=20Trifir=C3=B2?= <daniele@iterative.ai>
Date: Thu, 29 Sep 2022 11:34:07 +0200
Subject: [PATCH] repo: use realpath to in _get_filepath when loading stages

---
 dvc/repo/stage.py        |  5 ++---
 tests/func/test_stage.py | 17 +++++++++++++++++
 2 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/dvc/repo/stage.py b/dvc/repo/stage.py
index 049e00f95..84b4629fe 100644
--- a/dvc/repo/stage.py
+++ b/dvc/repo/stage.py
@@ -225,10 +225,9 @@ class StageLoad:
         path, name = parse_target(target)
         return self.load_one(path=path, name=name)
 
-    @staticmethod
-    def _get_filepath(path: str = None, name: str = None) -> str:
+    def _get_filepath(self, path: str = None, name: str = None) -> str:
         if path:
-            return path
+            return self.repo.fs.path.realpath(path)
 
         path = PIPELINE_FILE
         logger.debug("Assuming '%s' to be a stage inside '%s'", name, path)
diff --git a/tests/func/test_stage.py b/tests/func/test_stage.py
index 82950102b..73e0699e9 100644
--- a/tests/func/test_stage.py
+++ b/tests/func/test_stage.py
@@ -265,6 +265,23 @@ def test_collect_repo_ignored_dir_unignored_pattern(tmp_dir, dvc, scm):
     assert dvc.stage.collect_repo() == [stage]
 
 
+@pytest.mark.parametrize("with_deps", (False, True))
+def test_collect_symlink(tmp_dir, dvc, with_deps):
+    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"
+
+
 def test_stage_strings_representation(tmp_dir, dvc, run_copy):
     tmp_dir.dvc_gen("foo", "foo")
     stage1 = run_copy("foo", "bar", single_stage=True)