Merge pull request #983 from hubmapconsortium/Derek-Furst/bulk-ids

Derek furst/bulk ids

Author: yuanzhou

entity-api-spec.yaml

@@ -3140,3 +3140,46 @@ paths:
           description: The given dataset is unpublished and the user does not have the authorization to view it.
         '500':
           description: Internal error
+  '/entities/batch-ids':
+    post:
+      summary: Retrieve the HuBMAP ID and UUID for each entity id provided in a list. 
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              type: array
+              description: List of entity identifiers (HuBMAP IDs or UUIDs)
+              items:
+                type: string
+      responses:
+        '200':
+          description: Mapping of input entity identifiers to their resolved HuBMAP ID and UUID. If no items are found, will simply return empty. 
+          content:
+            application/json:
+              schema:
+                type: object
+                additionalProperties:
+                  type: object
+                  properties:
+                    hubmap_id:
+                      type: string
+                      description: Canonical HuBMAP ID for the entity
+                    uuid:
+                      type: string
+                      format: uuid
+                      description: UUID for the entity
+                  required:
+                    - hubmap_id
+                    - uuid
+              examples:
+                batchidsexample:
+                  value:
+                    HBM123.ABCD.456:
+                      hubmap_id: "HBM123.ABCD.456"
+                      uuid: "a8098c1a-f86e-11da-bd1a-00112444be1e"
+                    abcd1234-ef56-gh78-ij90-klmnop123456:
+                      hubmap_id: "HBM987.WXYZ.321"
+                      uuid: "abcd1234-ef56-gh78-ij90-klmnop123456"
+        '500': 
+          description: Internal Error

src/app.py

@@ -4319,6 +4319,33 @@ def entity_bulk_update():
     return jsonify(list(uuids)), 202
 
 
+"""
+Retrieve ids (uuid, hubmap_id) for a given id
+
+Request Body
+------------
+JSON array of entity identifiers (either uuids or HuBMAP IDs)
+
+Example:
+[
+    "HBM123.ABCD.456",
+    "a1234b56-c78d-90de-fg1h-23456789i01j"    
+]
+
+Returns
+-------
+json
+    JSON object keyed by the entity identifier. Each value is a mapping containing the HuBMAP ID and UUID for that entity
+"""
+@app.route('/entities/batch-ids', methods = ['POST'])
+def get_batch_ids():
+    validate_token_if_auth_header_exists(request)
+    require_json(request)
+    json_data_dict = request.get_json()
+    ids = app_neo4j_queries.get_batch_ids(neo4j_driver_instance, json_data_dict)    
+    return jsonify(ids)
+
+
 ####################################################################################################
 ## Internal Functions
 ####################################################################################################

src/app_neo4j_queries.py

@@ -1181,3 +1181,36 @@ def get_entities_by_uuid(neo4j_driver,
             return None
 
         return records
+
+"""
+Get the uuid and hubmap_id for each entity in a list of ids.
+
+Parameters
+----------
+neo4j_driver : neo4j.Driver object
+    The neo4j database connection pool
+id_list : list
+    The list of ids
+
+Returns
+-------
+Dictionary containing the uuid and hubmap_id of each entity in the list, keyed by that entities original id given
+"""
+def get_batch_ids(neo4j_driver, id_list):
+    query = """
+    MATCH (e) 
+    WHERE e.uuid IN $id_list OR e.hubmap_id IN $id_list
+    WITH e, [id IN $id_list WHERE id = e.uuid OR id = e.hubmap_id][0] AS original_id
+    RETURN original_id, e.uuid AS uuid, e.hubmap_id AS hubmap_id
+    """
+
+    result_map = {}
+
+    with neo4j_driver.session() as session:
+        result = session.run(query, id_list=id_list)
+        for record in result:
+            result_map[record['original_id']] = {
+                "uuid": record['uuid'],
+                "hubmap_id": record['hubmap_id']
+            }
+    return result_map