Add check for UUIDs when checking for existence of archives

payu/metadata.py

@@ -155,9 +155,8 @@ def set_experiment_name(self,
                   self.experiment_name)
             return
 
-        # Legacy experiment name and archive path
+        # Legacy experiment name
         legacy_name = self.control_path.name
-        legacy_archive_path = self.lab_archive_path / legacy_name
 
         if not self.enabled:
             # Metadata/UUID generation is disabled, so leave UUID out of
@@ -167,18 +166,13 @@ def set_experiment_name(self,
                   f"Experiment name used for archival: {self.experiment_name}")
             return
 
-        # Branch-UUID experiment name and archive path
         branch_uuid_experiment_name = self.new_experiment_name()
-        archive_path = self.lab_archive_path / branch_uuid_experiment_name
-
-        if is_new_experiment or archive_path.exists():
+        if is_new_experiment or self.has_archive(branch_uuid_experiment_name):
             # Use branch-UUID aware experiment name
             self.experiment_name = branch_uuid_experiment_name
-        elif legacy_archive_path.exists():
+        elif self.has_archive(legacy_name):
             # Use legacy CONTROL-DIR experiment name
             self.experiment_name = legacy_name
-            print(f"Pre-existing archive found at: {legacy_archive_path}. "
-                  f"Experiment name will remain: {legacy_name}")
         elif keep_uuid:
             # Use same experiment UUID and use branch-UUID name for archive
             self.experiment_name = branch_uuid_experiment_name
@@ -190,6 +184,25 @@ def set_experiment_name(self,
             )
             self.set_new_uuid(is_new_experiment=True)
 
+    def has_archive(self, experiment_name: str) -> bool:
+        """Return True if archive under the experiment name exists and
+        if it exists, check for a non-matching UUID in archive metadata."""
+        archive_path = self.lab_archive_path / experiment_name
+
+        if archive_path.exists():
+            # Check if the UUID in the archive metadata matches the
+            # UUID in metadata
+            archive_metadata_path = archive_path / METADATA_FILENAME
+            if archive_metadata_path.exists():
+                archive_metadata = YAML().load(archive_metadata_path)
+                if (UUID_FIELD in archive_metadata and
+                        archive_metadata[UUID_FIELD] != self.uuid):
+                    print("Mismatch of UUIDs between metadata and an archive "
+                          f"metadata found at: {archive_metadata_path}")
+                    return False
+            print(f"Found experiment archive: {archive_path}")
+        return archive_path.exists()
+
     def set_new_uuid(self, is_new_experiment: bool = False) -> None:
         """Generate a new uuid and set experiment name"""
         self.uuid_updated = True

test/test_metadata.py

@@ -261,6 +261,52 @@ def test_set_experiment_and_uuid(uuid_exists, keep_uuid, is_new_experiment,
     assert metadata.uuid == expected_uuid
 
 
+@pytest.mark.parametrize(
+    "archive_metadata_exists, archive_uuid, expected_result",
+    [
+        # A legacy archive exists, but there's no corresponding metadata
+        # in archive
+        (
+            False, None, True
+        ),
+        # Archive metadata exists but has no UUID
+        (
+            True, None, True
+        ),
+        # Archive metadata exists with same UUID
+        (
+            True, "3d18b3b6-dd19-49a9-8d9e-c7fa8582f136", True
+        ),
+        # Archive metadata exists with different UUID
+        (
+            True, "cb793e91-6168-4ed2-a70c-f6f9ccf1659b", False
+        ),
+    ]
+)
+def test_has_archive(archive_metadata_exists, archive_uuid, expected_result):
+    # Setup config and metadata
+    write_config(config)
+    with cd(ctrldir):
+        metadata = Metadata(archive_dir)
+    metadata.uuid = "3d18b3b6-dd19-49a9-8d9e-c7fa8582f136"
+
+    # Setup archive and it's metadata file
+    archive_path = archive_dir / "ctrl"
+    archive_path.mkdir(parents=True)
+
+    if archive_metadata_exists:
+        archive_metadata = {}
+
+        if archive_uuid is not None:
+            archive_metadata["experiment_uuid"] = archive_uuid
+
+        with open(archive_path / 'metadata.yaml', 'w') as file:
+            YAML().dump(archive_metadata, file)
+
+    result = metadata.has_archive("ctrl")
+    assert result == expected_result
+
+
 def test_set_configured_experiment_name():
     # Set experiment in config file
     test_config = copy.deepcopy(config)
@@ -287,6 +333,7 @@ def test_set_configured_experiment_name():
 )
 def test_new_experiment_name(branch, expected_name):
     # Test configured experiment name is the set experiment name
+    write_config(config)
     with cd(ctrldir):
         metadata = Metadata(archive_dir)