Extract and refactor methods

Author: nadove-ucsc

test/integration_test.py

@@ -775,13 +775,8 @@ def _get_one_outer_file(self, catalog: CatalogName) -> JSON:
         return one(hits)
 
     def _source_spec(self, catalog: CatalogName, entity: JSON) -> SourceSpec:
-        if config.is_hca_enabled(catalog):
-            field = 'sourceSpec'
-        elif config.is_anvil_enabled(catalog):
-            field = 'source_spec'
-        else:
-            assert False, catalog
-        return TDRSourceSpec.parse(one(entity['sources'])[field])
+        source = self._source_from_response(catalog, one(entity['sources']))
+        return source.spec
 
     def _file_size_facet(self, catalog: CatalogName) -> str:
         if config.is_hca_enabled(catalog):
@@ -1125,27 +1120,30 @@ def _test_repository_files(self, catalog: CatalogName):
             outer_file, inner_file = self._get_one_inner_file(catalog)
             file_url = inner_file['url']
             if file_url:
-                file_url = furl(file_url)
-                # FIXME: Use _check_endpoint() instead
-                #        https://github.com/DataBiosphere/azul/issues/7373
-                self.assertEqual(file_url.path.segments[0], 'repository')
-                file_url.path.segments.insert(0, 'fetch')
-                response = self._get_url_unchecked(GET, file_url)
-                self.assertEqual(200, response.status)
-                response = json.loads(response.data)
-                while response['Status'] != 302:
-                    self.assertEqual(301, response['Status'])
-                    self.assertNotIn('Retry-After', response)
-                    response = self._get_url_json(GET, furl(response['Location']))
-                self.assertNotIn('Retry-After', response)
-                response = self._get_url(GET, furl(response['Location']), stream=True)
                 source = self._source_spec(catalog, outer_file)
-                self._validate_file_response(response, source, inner_file)
+                self._test_file_download(source, inner_file)
             else:
                 # Phantom files lack DRS URIs and cannot be downloaded
                 self.assertIsNone(file_url, inner_file)
                 self.assertEqual('lungmap', config.catalogs[catalog].atlas, inner_file)
 
+    def _test_file_download(self, source: SourceSpec, file: JSON):
+        file_url = furl(file['url'])
+        # FIXME: Use _check_endpoint() instead
+        #        https://github.com/DataBiosphere/azul/issues/7373
+        self.assertEqual(file_url.path.segments[0], 'repository')
+        file_url.path.segments.insert(0, 'fetch')
+        response = self._get_url_unchecked(GET, file_url)
+        self.assertEqual(200, response.status)
+        response = json.loads(response.data)
+        while response['Status'] != 302:
+            self.assertEqual(301, response['Status'])
+            self.assertNotIn('Retry-After', response)
+            response = self._get_url_json(GET, furl(response['Location']))
+        self.assertNotIn('Retry-After', response)
+        response = self._get_url(GET, furl(response['Location']), stream=True)
+        self._validate_file_response(response, source, file)
+
     def _file_ext(self, file: JSON) -> str:
         # We believe that the file extension is a more reliable indicator than
         # the `format` metadata field. Note that this method preserves multipart
@@ -1289,6 +1287,13 @@ def _prepare_notifications(self,
         notifications.extend(duplicate_bundles)
         return notifications, bundle_fqids
 
+    def _source_from_response(self, catalog: CatalogName, source_json: JSON) -> SourceRef:
+        special_fields = self.metadata_plugin(catalog).special_fields
+        source = dict(id=source_json[special_fields.source_id],
+                      spec=source_json[special_fields.source_spec],
+                      prefix=source_json[special_fields.source_prefix])
+        return self.repository_plugin(catalog).source_ref_cls.from_json(source)
+
     def _get_indexed_bundles(self,
                              catalog: CatalogName,
                              filters: JSON | None = None
@@ -1298,10 +1303,7 @@ def _get_indexed_bundles(self,
         special_fields = self.metadata_plugin(catalog).special_fields
         for hit in hits:
             source, bundle = one(hit['sources']), one(hit['bundles'])
-            source = dict(id=source[special_fields.source_id],
-                          spec=source[special_fields.source_spec],
-                          prefix=source[special_fields.source_prefix])
-            source = self.repository_plugin(catalog).source_ref_cls.from_json(source)
+            source = self._source_from_response(catalog, source)
             bundle_fqid = SourcedBundleFQID(uuid=bundle[special_fields.bundle_uuid],
                                             version=bundle[special_fields.bundle_version],
                                             source=source)