Merge pull request #966 from hubmapconsortium/Derek-Furst/resurrect-publication-collection-association

Derek furst/resurrect publication collection association

Author: yuanzhou

entity-api-spec.yaml

@@ -937,6 +937,10 @@ components:
           type: string
           readOnly: true
           description: 'The email address of the person or process authenticated when creating the object.'
+        associated_publication:
+          type: object
+          description: 'The publication associated with the given collection'
+          readOnly: true
         created_by_user_sub:
           type: string
           readOnly: true

src/app.py

@@ -5403,8 +5403,6 @@ def delete_cache(entity_uuid, entity_type):
             dataset_upload_dict = schema_neo4j_queries.get_dataset_upload(neo4j_driver_instance, entity_uuid)
 
         # For Publication, also delete cache of the associated collection
-        # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated.
-        # Still keep it in the code until further decision - Zhou
         if entity_type == 'Publication':
             publication_collection_dict = schema_neo4j_queries.get_publication_associated_collection(neo4j_driver_instance, entity_uuid)
 

src/schema/provenance_schema.yaml

@@ -308,6 +308,15 @@ ENTITIES:
         indexed: true
         description: "The displayname of globus group which the user who created this entity is a member of"
         before_create_trigger: set_group_name #same as group_uuid, except set group_name
+      associated_publication:
+        type: json_string #dict
+        generated: true
+        indexed: true
+        transient: true
+        description: "A JSON containing the UUID, HuBMAP_ID, and Title for the associated publication"
+        on_read_trigger: get_collection_associated_publication
+        on_index_trigger: get_collection_associated_publication
+
 
 
   ############################################# Dataset #############################################
@@ -786,9 +795,6 @@ ENTITIES:
         type: string
         indexed: true
         description: 'A DOI pointing to an Organ Mapping Antibody Panel relevant to this publication'
-      
-      # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated.
-      # Still keep it in the code until further decision - Zhou
       associated_collection:
         type: json_string # dict
         generated: true

src/schema/schema_neo4j_queries.py

@@ -839,8 +839,6 @@ def get_parent_activity_uuid_from_entity(neo4j_driver, entity_uuid):
     the uuid of the associated collection
 """
 def link_publication_to_associated_collection(neo4j_driver, entity_uuid, associated_collection_uuid):
-    # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated.
-    # Still keep it in the code until further decision - Zhou
     try:
         with neo4j_driver.session() as session:
             tx = session.begin_transaction()
@@ -1109,9 +1107,6 @@ def get_next_revision_uuids(neo4j_driver, uuid):
 """
 def get_collection_associated_datasets(neo4j_driver, uuid, property_key = None):
     results = []
-
-    # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated. 
-    # Still keep it in the code until further decision - Zhou
     if property_key:
         query = (f"MATCH (e:Entity)-[:IN_COLLECTION|:USES_DATA]->(c:Collection) "
                  f"WHERE c.uuid = '{uuid}' "
@@ -1210,9 +1205,6 @@ def get_dataset_collections(neo4j_driver, uuid, property_key = None, properties_
 """
 def get_publication_associated_collection(neo4j_driver, uuid):
     result = {}
-
-    # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated.
-    # Still keep it in the code until further decision - Zhou
     query = (f"MATCH (p:Publication)-[:USES_DATA]->(c:Collection) "
              f"WHERE p.uuid = '{uuid}' "
              f"RETURN c as {record_field_name}")
@@ -1229,6 +1221,41 @@ def get_publication_associated_collection(neo4j_driver, uuid):
 
     return result
 
+
+"""
+Get the associated collection for a given publication
+
+Parameters
+----------
+neo4j_driver : neo4j.Driver object
+    The neo4j database connection pool
+uuid : str
+    The uuid of publication
+property_key : str
+    A target property key for result filtering
+
+Returns
+-------
+dict
+    A dictionary representation of the chosen values
+"""
+def get_collection_associated_publication(neo4j_driver, uuid):
+    result = {}
+    query = (f"MATCH (p:Publication)-[:USES_DATA]->(c:Collection) "
+             f"WHERE c.uuid = '{uuid}' "
+             f"RETURN {{uuid: p.uuid, hubmap_id: p.hubmap_id, title: p.title}} AS publication")
+
+    logger.info("=====get_collection_associated_publication() query======")
+    logger.debug(query)
+
+    with neo4j_driver.session() as session:
+        record = session.run(query).single()
+        if record:
+            result = record["publication"]
+    return result
+
+
+
 """
 Get the associated Upload for a given dataset
 
@@ -2057,8 +2084,6 @@ def delete_ancestor_linkages_tx(neo4j_driver, entity_uuid, ancestor_uuids):
     The uuid to target publication
 """
 def _delete_publication_associated_collection_linkages_tx(tx, uuid):
-    # NOTE: As of 5/30/2025, the [:USES_DATA] workaround has been deprecated.
-    # Still keep it in the code until further decision - Zhou
     query = (f"MATCH (p:Publication)-[r:USES_DATA]->(c:Collection) "
              f"WHERE p.uuid = '{uuid}' "
              f"DELETE r")

src/schema/schema_triggers.py

@@ -830,6 +830,45 @@ def get_publication_associated_collection(property_key, normalized_type, request
     return property_key, schema_manager.normalize_entity_result_for_response(collection_dict)
 
 
+"""
+TriggerTypeEnum.ON_READ
+
+Trigger event method of getting the associated publication for this collection
+
+Parameters
+----------
+property_key : str
+    The target property key
+normalized_type : str
+    One of the types defined in the schema yaml: Dataset
+request_args: ImmutableMultiDict
+    The Flask request.args passed in from application request
+user_token: str
+    The user's globus nexus token
+existing_data_dict : dict
+    A dictionary that contains all existing entity properties
+new_data_dict : dict
+    A merged dictionary that contains all possible input data to be used
+
+Returns
+-------
+str: The target property key
+dict: A dictionary representation of the associated publication with all the normalized information
+"""
+def get_collection_associated_publication(property_key, normalized_type, request_args, user_token, existing_data_dict, new_data_dict):
+    if 'uuid' not in existing_data_dict:
+        raise KeyError("Missing 'uuid' key in 'existing_data_dict' during calling 'get_collection_associated_publication()' trigger method.")
+
+    logger.info(f"Executing 'get_collection_associated_publication()' trigger method on uuid: {existing_data_dict['uuid']}")
+
+    publication_dict = schema_neo4j_queries.get_collection_associated_publication(schema_manager.get_neo4j_driver_instance(), existing_data_dict['uuid'])
+
+    # Get rid of the entity node properties that are not defined in the yaml schema
+    # as well as the ones defined as `exposed: false` in the yaml schema
+    return property_key, publication_dict
+
+
+
 """
 TriggerTypeEnum.ON_READ