fixup! Fix: Invalid columns in compact manifest for AnVIL (#6110)

src/azul/plugins/metadata/anvil/__init__.py

@@ -133,16 +133,17 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
             'document_id',
             'source_datarepo_row_ids'
         ]
-        # Note that there is a brittle coupling that must be maintained between
-        # the keys here and those used in the custom field name lookup in
-        # `self.manifest_config`.
         return {
             'entity_id': 'entryId',
             'bundles': {
+                # These field paths have a brittle coupling that must be
+                # maintained to the field lookups in `self.manifest_config`.
                 'uuid': self.special_fields.bundle_uuid,
                 'version': self.special_fields.bundle_version
             },
             'sources': {
+                # These field paths have a brittle coupling that must be
+                # maintained to the field lookups in `self.manifest_config`.
                 'id': self.special_fields.source_id,
                 'spec': self.special_fields.source_spec
             },
@@ -199,6 +200,9 @@ def _field_mapping(self) -> MetadataPlugin._FieldMapping:
                     f: f'activities.{f}' for f in [
                         *common_fields,
                         'activity_id',
+                        # This field path has a brittle coupling that must be
+                        # maintained to the field lookup in
+                        # `self.manifest_config`.
                         'activity_table',
                         'activity_type',
                         'assay_type',
@@ -286,17 +290,9 @@ def manifest_config(self) -> ManifestConfig:
             ('contents', 'files', 'version'),
         ]
 
-        # This is used to modify the field names from how they are specified in
-        # the field mapping. Keys are field paths in an ES hit, and values are
-        # the desired manifest column name, or None to omit the column from the
-        # manifest.
-        #
-        # Note that there is a brittle coupling that must be maintained between
-        # the keys here and those used in `self._field_mapping`. Also, the
-        # values should match the related field's path in a response hit from
-        # the `/index/files` endpoint.
-        #
-        custom_field_names = {
+        # Furthermore, renamed values should match the field's path in a
+        # response hit from the `/index/files` endpoint.
+        fields_to_rename_in_manifest = {
             ('bundles', 'uuid'): 'bundles.bundle_uuid',
             ('bundles', 'version'): 'bundles.bundle_version',
             ('sources', 'id'): 'sources.source_id',
@@ -314,16 +310,16 @@ def recurse(mapping: MetadataPlugin._FieldMapping, path: FieldPath):
                     elif new_path in fields_to_omit_from_manifest:
                         result[path][path_element] = None
                         fields_to_omit_from_manifest.remove(new_path)
-                    elif new_path in custom_field_names:
-                        result[path][path_element] = custom_field_names.pop(new_path)
+                    elif new_path in fields_to_rename_in_manifest:
+                        result[path][path_element] = fields_to_rename_in_manifest.pop(new_path)
                     else:
                         result[path][path_element] = name_or_type
                 else:
                     assert False, (path, path_element, name_or_type)
 
         recurse(self._field_mapping, ())
         assert len(fields_to_omit_from_manifest) == 0, fields_to_omit_from_manifest
-        assert len(custom_field_names) == 0, custom_field_names
+        assert len(fields_to_rename_in_manifest) == 0, fields_to_rename_in_manifest
         # The file URL is synthesized from the `uuid` and `version` fields.
         # Above, we already configured these two fields to be omitted from the
         # manifest since they are not informative to the user.

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

@@ -140,8 +140,8 @@ def _make_hit(self, es_hit: JSON) -> MutableJSON:
         return {
             'entryId': es_hit['entity_id'],
             # Note that there is a brittle coupling that must be maintained
-            # between the `sources` and `bundles` field paths and the values in
-            # the custom field name lookup in `Plugin.manifest_config`.
+            # between the `sources` and `bundles` field paths here and the
+            # renamed fields in `Plugin.manifest_config`.
             'sources': list(map(self._make_source, es_hit['sources'])),
             'bundles': list(map(self._make_bundle, es_hit['bundles'])),
             **self._make_contents(es_hit['contents'])