Resource policies duplicity

assigned one group to a resource policy based on the action type

src/pump/_collection.py

@@ -20,6 +20,8 @@ class collections:
     ]
 
     TYPE = 3
+    ITEM = "ITEM"
+    BITSTREAM = "BITSTREAM"
 
     def __init__(self, col_file_str: str, com2col_file_str: str, metadata_file_str: str):
         self._col = read_json(col_file_str)
@@ -33,6 +35,7 @@ def __init__(self, col_file_str: str, com2col_file_str: str, metadata_file_str:
 
         self._logos = {}
         self._groups_id2uuid = {}
+        self._groups_uuid2type = {}
 
         if len(self._col) == 0:
             _logger.info(f"Empty input collections: [{col_file_str}].")
@@ -81,6 +84,10 @@ def imported_groups(self):
     def groups_id2uuid(self):
         return self._groups_id2uuid
 
+    @property
+    def groups_uuid2type(self):
+        return self._groups_uuid2type
+
     @time_method
     def import_to(self, dspace, handles, metadatas, coms):
         expected = len(self)
@@ -150,6 +157,7 @@ def import_to(self, dspace, handles, metadatas, coms):
                     resp = dspace.put_collection_bitstream_read_group(col_uuid)
                     self._groups_id2uuid.setdefault(str(group_col), []).append(resp['id'])
                     self._imported["group"] += 1
+                    self._groups_uuid2type[resp['id']] = collections.BITSTREAM
                 except Exception as e:
                     _logger.error(
                         f'put_collection_bitstream_read_group: [{col_id}] failed [{str(e)}]')
@@ -158,6 +166,7 @@ def import_to(self, dspace, handles, metadatas, coms):
                     resp = dspace.put_collection_item_read_group(col_uuid)
                     self._groups_id2uuid.setdefault(str(group_col), []).append(resp['id'])
                     self._imported["group"] += 1
+                    self._groups_uuid2type[resp['id']] = collections.ITEM
                 except Exception as e:
                     _logger.error(
                         f'put_collection_item_read_group: [{col_id}] failed [{str(e)}]')
@@ -172,6 +181,7 @@ def serialize(self, file_str: str):
             "logos": self._logos,
             "groups_id2uuid": self._groups_id2uuid,
             "imported": self._imported,
+            "groups_uuid2type": self._groups_uuid2type,
         }
         serialize(file_str, data)
 
@@ -183,3 +193,4 @@ def deserialize(self, file_str: str):
         self._logos = data["logos"]
         self._groups_id2uuid = data["groups_id2uuid"]
         self._imported = data["imported"]
+        self._groups_uuid2type = data["groups_uuid2type"]

src/pump/_resourcepolicy.py

@@ -19,6 +19,8 @@ def __init__(self, resourcepolicy_file_str: str):
             "respol": 0,
         }
 
+    DEFAULT_BITSTREAM_READ = "DEFAULT_BITSTREAM_READ"
+
     def __len__(self):
         return len(self._respol)
 
@@ -92,11 +94,36 @@ def import_to(self, env, dspace, repo):
             # get group if it is not none
             eg_id = res_policy['epersongroup_id']
             if eg_id is not None:
-                group_list1 = repo.groups.uuid(eg_id)
-                group_list2 = repo.collections.group_uuid(eg_id)
-                group_list = set(group_list1 + group_list2)
+                # groups created with coll and comm are already in the group
+                group_list = repo.groups.uuid(eg_id)
                 if len(group_list) == 0:
                     continue
+                if len(group_list) > 1:
+                    if len(group_list) != 2:
+                        raise RuntimeError(
+                            f'Unexpected size of mapped groups to group [{eg_id}]: {len(group_list)}. '
+                            f'Expected size: 2.')
+                    group_types = repo.collections.groups_uuid2type
+                    # Determine the target type based on the action
+                    target_type = (
+                        repo.collections.BITSTREAM
+                        if dspace_actions[actionId] == resourcepolicies.DEFAULT_BITSTREAM_READ
+                        else repo.collections.ITEM
+                    )
+                    # Filter group_list to find the appropriate group based on type using list comprehension
+                    group_type_list = [
+                        group for group in group_list
+                        if group in group_types and group_types[group] == target_type
+                    ]
+
+                    if len(group_type_list) != 1:
+                        raise RuntimeError(
+                            f'Unexpected size of filtered groups for group [{eg_id}] '
+                            f'of type [{target_type}]: {len(group_type_list)}. Expected size: 1.'
+                        )
+
+                    group_list = group_type_list
+
                 imported_groups = 0
                 for group in group_list:
                     params['group'] = group

src/repo_import.py

@@ -278,15 +278,14 @@ def deserialize(resume: bool, obj, cache_file: str) -> bool:
     repo.diff(repo.usermetadatas)
     _logger.info(import_sep)
 
-    # before importing of resource policies we have to delete all
-    # created data
-    repo.raw_db_7.delete_resource_policy()
-
-    # import bitstreams
+    # import resourcepolicy
     cache_file = env["cache"]["resourcepolicy"]
     if deserialize(args.resume, repo.resourcepolicies, cache_file):
         _logger.info(f"Resuming resourcepolicies [{repo.resourcepolicies.imported}]")
     else:
+        # before importing of resource policies we have to delete all
+        # created data
+        repo.raw_db_7.delete_resource_policy()
         repo.resourcepolicies.import_to(env, dspace_be, repo)
         repo.resourcepolicies.serialize(cache_file)
     repo.diff(repo.resourcepolicies)