schemas: add SimplifiedSchemaSerializer

Signed-off-by: Anna Trzcinska <[email protected]>

cap/modules/records/utils.py

@@ -71,3 +71,27 @@ def api_url_for(endpoint, pid, **kwargs):
                   **kwargs)
 
     return url_to_api_url(url)
+
+
+def reindex_by_schema_url(schema_url, pid_type):
+    """Reindex all records of given pid_type belonging to that schema."""
+    def _get_json_type():
+        """If postgres db return JSONB, else JSON."""
+        return JSONB if db.session.bind.dialect.name == 'postgresql' else JSON
+
+    indexer = RecordIndexer()
+
+    ids = (x[0] for x in RecordMetadata.query.filter(
+        RecordMetadata.json['$schema'] == cast(
+            schema_url, _get_json_type())).values(RecordMetadata.id))
+
+    filtered_by_pid_type = (x[0] for x in PersistentIdentifier.query.filter(
+        PersistentIdentifier.status == PIDStatus.REGISTERED,
+        PersistentIdentifier.object_type == 'rec', PersistentIdentifier.
+        pid_type == pid_type, PersistentIdentifier.object_uuid.in_(
+            ids)).values(PersistentIdentifier.object_uuid))
+
+    print('{} records will be reindexed...'.format(schema_url))
+
+    indexer.bulk_index(filtered_by_pid_type)
+    indexer.process_bulk_queue(es_bulk_kwargs={'raise_on_error': True})

cap/modules/schemas/serializers.py

@@ -25,10 +25,10 @@
 import copy
 
 from flask import url_for
-from invenio_jsonschemas.proxies import current_jsonschemas
-from marshmallow import Schema, ValidationError, fields, pre_load, validate
 
 from cap.utils import url_to_api_url
+from invenio_jsonschemas.proxies import current_jsonschemas
+from marshmallow import Schema, ValidationError, fields, pre_load, validate
 
 from .validators import JSONSchemaValidator
 
@@ -102,6 +102,15 @@ def filter_out_fields_that_cannot_be_updated(self, data, **kwargs):
         return data
 
 
+class SimplifiedSchemaSerializer(Schema):
+    """Simplified serializer for schema."""
+
+    name = fields.Str(dump_only=True, required=True)
+    version = fields.Str(dump_only=True, required=True,
+                         validate=validate.Regexp(regex=r"(\d+).(\d+).(\d+)"))
+    fullname = fields.Str(dump_only=True)
+
+
 class ResolvedSchemaSerializer(SchemaSerializer):
     """Schema serializer with resolved jsonschemas."""
 
@@ -125,5 +134,6 @@ def get_resolved_record_schema(self, obj):
 
 
 schema_serializer = SchemaSerializer()
+simplified_schema_serializer = SimplifiedSchemaSerializer()
 update_schema_serializer = UpdateSchemaSerializer()
 resolved_schemas_serializer = ResolvedSchemaSerializer()