From 28021445a71b8c8e7cc1f7fdac6d89457095b81d Mon Sep 17 00:00:00 2001
From: Joanne Bogart <jrb@slac.stanford.edu>
Date: Wed, 23 Oct 2024 17:01:03 -0700
Subject: [PATCH] allow duplicate relative_path in case dataset is not copied

---
 src/dataregistry/registrar/dataset.py         | 27 +++++++++++++------
 .../test_register_dataset_real_data.py        |  8 ++++--
 2 files changed, 25 insertions(+), 10 deletions(-)

diff --git a/src/dataregistry/registrar/dataset.py b/src/dataregistry/registrar/dataset.py
index 85034977..e6bbabe5 100644
--- a/src/dataregistry/registrar/dataset.py
+++ b/src/dataregistry/registrar/dataset.py
@@ -414,6 +414,7 @@ def register(
 
         # Make sure the relative_path in the `root_dir` is avaliable
         if kwargs_dict["location_type"] in ["dataregistry", "dummy"]:
+            will_copy = kwargs_dict["old_location"]
             previous_datasets = self._find_previous(
                 None,
                 None,
@@ -433,17 +434,27 @@ def register(
                     root_dir=self._root_dir,
                 )
 
+                warned = False
                 if get_dataset_status(previous_datasets[-1].status, "archived"):
-                    raise ValueError(
-                        f"Relative path {dest} is reserved "
-                        f"for archived datasetid={previous_datasets[-1].dataset_id}"
-                    )
+                    if will_copy:
+                        raise ValueError(
+                            f"Relative path {dest} is reserved "
+                            f"for archived datasetid={previous_datasets[-1].dataset_id}"
+                        )
+                    else:
+                        print(f"Warning: found existing entry with path {kwargs_dict['relative_path']}")
+                        warned = True
 
                 if not get_dataset_status(previous_datasets[-1].status, "deleted"):
-                    raise ValueError(
-                        f"Relative path {dest} is taken by "
-                        f"datasetid={previous_datasets[-1].dataset_id}"
-                    )
+                    if will_copy:
+                        raise ValueError(
+                            f"Relative path {dest} is taken by "
+                            f"datasetid={previous_datasets[-1].dataset_id}"
+                        )
+                    else:
+                        if not warned:
+                            print(f"Warning: found existing entry with path {kwargs_dict['relative_path']}")
+
 
         # Make sure there is not already a database entry with this
         # name/version combination
diff --git a/tests/end_to_end_tests/test_register_dataset_real_data.py b/tests/end_to_end_tests/test_register_dataset_real_data.py
index 015b54ca..22210731 100644
--- a/tests/end_to_end_tests/test_register_dataset_real_data.py
+++ b/tests/end_to_end_tests/test_register_dataset_real_data.py
@@ -54,6 +54,7 @@ def test_copy_data(dummy_file, data_org):
     [
         ("file", "file1.txt"),
         ("directory", "dummy_dir"),
+        ("same_directory", "dummy_dir")
     ],
 )
 def test_on_location_data(dummy_file, data_org, data_path):
@@ -84,9 +85,12 @@ def test_on_location_data(dummy_file, data_org, data_path):
         [f],
     )
 
-    assert len(results["dataset.data_org"]) == 1
+    if data_org == "same_directory":
+        assert len(results["dataset.data_org"]) == 2
+    else:
+        assert len(results["dataset.data_org"]) == 1
 
-    assert results["dataset.data_org"][0] == data_org
+    assert results["dataset.data_org"][0].endswith(data_org)
     assert results["dataset.nfiles"][0] == 1
     assert results["dataset.total_disk_space"][0] > 0