Merge pull request #693 from hubmapconsortium/Derek-Furst/external-process

Derek furst/external process

Author: yuanzhou

src/schema/provenance_schema.yaml

@@ -654,6 +654,15 @@ ENTITIES:
         indexed: true
         description: "The title of the publication."
         required_on_create: true # Only required for create via POST, not update via PUT
+      creation_action:
+        type: string
+        transient: true
+        generated: true
+        immutable: true
+        indexed: true
+        on_read_trigger: get_creation_action_activity
+        on_index_trigger: get_creation_action_activity
+        description: "The activity that was performed."
       dataset_type:
         before_create_trigger: set_publication_dataset_type
         before_property_create_validators:

src/schema/schema_neo4j_queries.py

@@ -112,6 +112,38 @@ def get_entity(neo4j_driver, uuid):
     return result
 
 
+
+"""
+Get the uuids for each entity in a list that doesn't belong to a certain entity type. Uuids are ordered by type
+
+Parameters
+----------
+neo4j_driver : neo4j.Driver object
+    The neo4j database connection pool
+direct_ancestor_uuids : list
+    List of the uuids to be filtered
+entity_type : string
+    The entity to be excluded
+
+Returns
+-------
+dict
+    A dictionary of entity uuids that don't pass the filter, grouped by entity_type
+"""
+def filter_ancestors_by_type(neo4j_driver, direct_ancestor_uuids, entity_type):
+    query = (f"MATCH (e:Entity) "
+             f"WHERE e.uuid in {direct_ancestor_uuids} AND toLower(e.entity_type) <> '{entity_type.lower()}' "
+             f"RETURN e.entity_type AS entity_type, collect(e.uuid) AS uuids")
+    logger.info("======filter_ancestors_by_type======")
+    logger.info(query)
+
+    with neo4j_driver.session() as session:
+        records = session.run(query).data()
+          
+    return records if records else None
+   
+
+
 """
 Get all children by uuid
 

src/schema/schema_validators.py

@@ -579,13 +579,19 @@ def validate_publication_date(property_key, normalized_entity_type, request, exi
     The json data in request body, already after the regular validations
 """
 def validate_creation_action(property_key, normalized_entity_type, request, existing_data_dict, new_data_dict):
-    accepted_creation_action_values = ["central process", "lab process"]
+    accepted_creation_action_values = ["central process", "lab process", "external process"]
     creation_action = new_data_dict[property_key].lower()
     if creation_action and creation_action not in accepted_creation_action_values:
         raise ValueError("Invalid {} value. Accepted values are: {}".format(property_key, ", ".join(accepted_creation_action_values)))
     if creation_action == '':
         raise ValueError(f"The property {property_key} cannot be empty, when specified.")
-
+    if creation_action == 'external process':
+        direct_ancestor_uuids = new_data_dict.get('direct_ancestor_uuids')
+        entity_types_dict = schema_neo4j_queries.filter_ancestors_by_type(schema_manager.get_neo4j_driver_instance(), direct_ancestor_uuids, "dataset")
+        if entity_types_dict:
+            raise ValueError(f"If 'creation_action' field is given, all ancestor uuids must belong to datasets. The following entities belong to non-dataset entities \
+                             {entity_types_dict}")
+        
 
 """
 Validate the provided value of the activity creation action before updating direct ancestors. Certain values prohibited