custom check for subrepo (#950)

Author: PythonFZ

tests/integration/test_nested_repo.py

@@ -62,6 +62,37 @@ def test_subrepo(proj_path):
     assert node_loaded.outs == {"param1": 3, "param2": 4}
 
 
+def test_from_rev_in_subrepo(proj_path):
+    """Test loading a node with from_rev when cwd is inside a subrepo.
+
+    This tests the scenario where:
+    1. The DVC repo root is at the parent directory
+    2. The project is in a subdirectory
+    3. The user is inside that subdirectory
+    4. The user calls zntrack.from_rev("NodeName") without specifying a path
+
+    This should work because the zntrack.json and other files are in the
+    current directory. The path should NOT be duplicated.
+    """
+    directory = pathlib.Path("subrepo")
+    directory.mkdir(parents=True, exist_ok=True)
+    os.chdir(directory)
+
+    project = zntrack.Project()
+    with project:
+        _ = zntrack.examples.ParamsToOuts(
+            params={"param1": 1, "param2": 2},
+        )
+    project.repro()
+
+    # Now we're inside the subdirectory and want to load the node
+    # This should work - we shouldn't need to specify the path since
+    # zntrack.json is in the current directory
+    node_loaded = zntrack.from_rev("ParamsToOuts")
+    assert node_loaded.params == {"param1": 1, "param2": 2}
+    assert node_loaded.outs == {"param1": 1, "param2": 2}
+
+
 def test_subrepo_external_node(proj_path):
     """Test subrepo functionality with external dataclasses."""
     # Create external node module file that can be imported

zntrack/from_rev.py

@@ -76,6 +76,25 @@ def from_rev(
         # DVC Issue where stage.path is relative to the DVC root
         # if rev is given, if only remote is given, it is relative to file system
         # so we need to use path_in_repo instead.
+
+        # When using local filesystem (no remote/rev), check if the current working
+        # directory is already inside the stage's path. If so, use an empty path
+        # to avoid path duplication issues.
+        if remote is None and rev is None and fs.repo is not None:
+            repo_root = pathlib.Path(fs.repo.root_dir).resolve()
+            cwd = pathlib.Path.cwd().resolve()
+            try:
+                cwd_relative = cwd.relative_to(repo_root)
+                # If cwd is at or inside the stage's path, adjust accordingly
+                if cwd_relative == path or str(cwd_relative).startswith(str(path) + "/"):
+                    # We're inside the stage directory, use empty path
+                    path = pathlib.Path()
+                elif str(path).startswith(str(cwd_relative) + "/"):
+                    # Stage is inside cwd, use relative path from cwd
+                    path = pathlib.Path(str(path)[len(str(cwd_relative)) + 1 :])
+            except ValueError:
+                # cwd is not inside repo_root, keep the path as-is
+                pass
     except AttributeError:
         raise ValueError("Stage is not a ZnTrack pipeline stage.")