Fix: AnVIL summary is lacking file size totals (#7201, PR #7239)

Author: achave11-ucsc

src/azul/plugins/metadata/anvil/service/aggregation.py

@@ -28,6 +28,10 @@ class AnvilSummaryAggregationStage(AnvilAggregationStage):
     def prepare_request(self, request: Search) -> Search:
         request = super().prepare_request(request)
         request = request.extra(size=0)
+        if self.entity_type == 'files':
+            request.aggs.metric('totalFileSize',
+                                'sum',
+                                field='contents.files.size_')
         return request
 
     def process_response(self, response: MutableJSON) -> MutableJSON:

src/azul/plugins/metadata/anvil/service/response.py

@@ -52,7 +52,8 @@ def aggs_by_authority(self) -> Mapping[str, Sequence[str]]:
                 'diagnoses.phenotype'
             ],
             'files': [
-                'files.file_format'
+                'files.file_format',
+                'totalFileSize'
             ]
         }
 
@@ -80,6 +81,7 @@ def bucket_count(field, bucket_key):
             'donorSpecies': bucket_count('donors.organism_type', 'species'),
             'fileCount': doc_count('files.file_format'),
             'fileFormats': bucket_count('files.file_format', 'format'),
+            'totalFileSize': response['totalFileSize']['value'],
         }
 
 

test/service/test_response_anvil.py

@@ -2443,7 +2443,8 @@ def test_summary(self):
                     'count': 1,
                     'format': '.vcf.gz'
                 }
-            ]
+            ],
+            'totalFileSize': 3519867231.0
         }
         url = str(self.base_url.set(path='/index/summary'))
         self._assertResponse(url, expected_response)