Merge pull request #264 from sennetconsortium/maxsibilla/ingest-metadata

Maxsibilla/ingest metadata

VERSION

@@ -1 +1 @@
-1.4.1
+1.4.2

entity-api-spec.yaml

@@ -1,7 +1,7 @@
 openapi: 3.0.0
 info:
   description: 'The SenNet Entity API is a standard RESTful web service with create, update and read operations for the standard SenNet provenance graph entities.'
-  version: 1.4.1
+  version: 1.4.2
   title: SenNet Entity API
   contact:
     name: SenNet Help Desk

src/app.py

@@ -4601,10 +4601,9 @@ def check_for_metadata(entity_type, user_token):
     # Parse incoming json string into json data(python dict object)
     json_data_dict = request.get_json()
 
-    # Convert `ingest_metadata` from pipeline to `metadata`
+    # ingest_metadata only exists on Datasets and we do not currently validate that
     if 'ingest_metadata' in json_data_dict:
-        json_data_dict['metadata'] = json_data_dict['ingest_metadata']
-        del json_data_dict['ingest_metadata']
+        return json_data_dict
 
     if 'metadata' in json_data_dict:
         # If metadata is empty then just proceed. The portal is more than likely trying to reset this intentionally
@@ -4614,10 +4613,6 @@ def check_for_metadata(entity_type, user_token):
         if isinstance(json_data_dict['metadata'], list):
             json_data_dict['metadata'] = json_data_dict['metadata'][0]
 
-        # TODO: add a validation check (validate_metadata). For now we want this to accept pipeline Dataset metadata
-        if entity_type == 'Dataset':
-            return json_data_dict
-
         if json_data_dict['metadata']:
             file_row = json_data_dict['metadata'].get('file_row')
             if 'pathname' in json_data_dict['metadata']:

src/schema/provenance_schema.yaml

@@ -450,6 +450,9 @@ ENTITIES:
       pipeline_message:
         # Generated from the Ingest Pipeline during processing of a derived dataset
         type: string
+      ingest_metadata:
+        type: json_string # dict
+        description: "The metadata returned from the processing at data submission time."
       local_directory_rel_path:
         # Example: protected/<TMC>/<uuid>
         type: string