Feature: allow to exclude relations on object get

Author: psrok1

mwdb/model/object.py

@@ -7,7 +7,7 @@
 from sqlalchemy import and_, cast, distinct, exists, func, or_, select
 from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.exc import IntegrityError
-from sqlalchemy.orm import aliased, column_property, contains_eager
+from sqlalchemy.orm import column_property
 from sqlalchemy.sql.expression import true
 
 from mwdb.core.capabilities import Capabilities
@@ -668,6 +668,21 @@ def _get_or_create(
 
         return created_object, is_new
 
+    def query_visible_parents(self, requestor=None):
+        """
+        Queries for parents visible by specified requestor.
+        """
+        if requestor is None:
+            requestor = g.auth_user
+
+        return (
+            db.session.query(Object)
+            .join(relation, relation.c.parent_id == Object.id)
+            .filter(relation.c.child_id == self.id)
+            .order_by(relation.c.creation_time.desc())
+            .filter(requestor.has_access_to_object(Object.id))
+        )
+
     @classmethod
     def access(cls, identifier, requestor=None):
         """
@@ -686,35 +701,12 @@ def access(cls, identifier, requestor=None):
         if requestor is None:
             requestor = g.auth_user
 
-        obj = cls.get(identifier)
+        obj_query = cls.get(identifier)
+        obj = obj_query.first()
         # If object doesn't exist - it doesn't exist
-        if obj.first() is None:
+        if obj is None:
             return None
 
-        # In that case we want only those parents to which requestor has access.
-        stmtp = (
-            db.session.query(Object)
-            .join(relation, relation.c.parent_id == Object.id)
-            .filter(
-                Object.id.in_(
-                    db.session.query(relation.c.parent_id).filter(
-                        relation.c.child_id == obj.first().id
-                    )
-                )
-            )
-            .order_by(relation.c.creation_time.desc())
-            .filter(requestor.has_access_to_object(Object.id))
-        )
-        stmtp = stmtp.subquery()
-
-        parent = aliased(Object, stmtp)
-
-        obj = (
-            obj.outerjoin(parent, Object.parents)
-            .options(contains_eager(Object.parents, alias=parent))
-            .all()[0]
-        )
-
         # Ok, now let's check whether requestor has explicit access
         if obj.has_explicit_access(requestor):
             return obj

mwdb/resources/blob.py

@@ -8,6 +8,7 @@
 from mwdb.model.object import ObjectTypeConflictError
 from mwdb.schema.blob import (
     BlobCreateRequestSchema,
+    BlobItemAndRelationsResponseSchema,
     BlobItemResponseSchema,
     BlobLegacyCreateRequestSchema,
     BlobListResponseSchema,
@@ -186,6 +187,7 @@ def post(self):
 class TextBlobItemResource(ObjectItemResource, TextBlobUploader):
     ObjectType = TextBlob
     ItemResponseSchema = BlobItemResponseSchema
+    ItemAndRelationsResponseSchema = BlobItemAndRelationsResponseSchema
     CreateRequestSchema = BlobLegacyCreateRequestSchema
 
     def call_specialised_remove_hook(self, text_blob):

mwdb/resources/config.py

@@ -13,6 +13,7 @@
 from mwdb.schema.blob import BlobCreateSpecSchema
 from mwdb.schema.config import (
     ConfigCreateRequestSchema,
+    ConfigItemAndRelationsResponseSchema,
     ConfigItemResponseSchema,
     ConfigLegacyCreateRequestSchema,
     ConfigListResponseSchema,
@@ -166,6 +167,7 @@ class ConfigResource(ObjectResource, ConfigUploader):
     ObjectType = Config
     ListResponseSchema = ConfigListResponseSchema
     ItemResponseSchema = ConfigItemResponseSchema
+    ItemAndRelationsResponseSchema = ConfigItemAndRelationsResponseSchema
 
     @requires_authorization
     def get(self):
@@ -309,6 +311,7 @@ class ConfigItemResource(ObjectItemResource, ConfigUploader):
 
     ObjectType = Config
     ItemResponseSchema = ConfigItemResponseSchema
+    ItemAndRelationsResponseSchema = ConfigItemAndRelationsResponseSchema
     CreateRequestSchema = ConfigLegacyCreateRequestSchema
 
     def call_specialised_remove_hook(self, config):

mwdb/resources/file.py

@@ -11,6 +11,7 @@
 from mwdb.schema.file import (
     FileCreateRequestSchema,
     FileDownloadTokenResponseSchema,
+    FileItemAndRelationsResponseSchema,
     FileItemResponseSchema,
     FileLegacyCreateRequestSchema,
     FileListResponseSchema,
@@ -203,6 +204,7 @@ def post(self):
 class FileItemResource(ObjectItemResource, FileUploader):
     ObjectType = File
     ItemResponseSchema = FileItemResponseSchema
+    ItemAndRelationsResponseSchema = FileItemAndRelationsResponseSchema
     CreateRequestSchema = FileLegacyCreateRequestSchema
 
     def call_specialised_remove_hook(self, file):
@@ -225,6 +227,15 @@ def get(self, identifier):
               schema:
                 type: string
               description: File identifier (SHA256/SHA512/SHA1/MD5)
+            - in: query
+              name: exclude_relations
+              schema:
+                type: integer
+              description: |
+                If set, results doesn't include relations
+                which will be default behavior on next major release of MWDB
+              required: false
+              default: 0
         responses:
             200:
                 description: Information about file

mwdb/resources/object.py

@@ -16,6 +16,8 @@
 from mwdb.schema.object import (
     ObjectCountRequestSchema,
     ObjectCountResponseSchema,
+    ObjectItemAndRelationsResponseSchema,
+    ObjectItemRequestSchema,
     ObjectItemResponseSchema,
     ObjectListRequestSchema,
     ObjectListResponseSchema,
@@ -262,6 +264,7 @@ def get(self):
 class ObjectItemResource(Resource, ObjectUploader):
     ObjectType = Object
     ItemResponseSchema = ObjectItemResponseSchema
+    ItemAndRelationsResponseSchema = ObjectItemAndRelationsResponseSchema
     CreateRequestSchema = None
 
     def call_specialised_remove_hook(self, obj):
@@ -284,6 +287,15 @@ def get(self, identifier):
               schema:
                 type: string
               description: Object identifier
+            - in: query
+              name: exclude_relations
+              schema:
+                type: integer
+              description: |
+                If set, results doesn't include relations
+                which will be default behavior on next major release of MWDB
+              required: false
+              default: 0
         responses:
             200:
                 description: Information about object
@@ -298,10 +310,15 @@ def get(self, identifier):
                 description: |
                     Request canceled due to database statement timeout.
         """
+        args = load_schema(request.args, ObjectItemRequestSchema())
         obj = self.ObjectType.access(identifier)
         if obj is None:
             raise NotFound("Object not found")
-        schema = self.ItemResponseSchema()
+        schema = (
+            self.ItemResponseSchema()
+            if args["exclude_relations"]
+            else self.ItemAndRelationsResponseSchema()
+        )
         return schema.dump(obj)
 
     def _get_upload_args(self, parent_identifier):

mwdb/resources/relations.py

@@ -5,7 +5,7 @@
 from mwdb.core.plugins import hooks
 from mwdb.core.rate_limit import rate_limited_resource
 from mwdb.model import Object, db
-from mwdb.schema.relations import RelationsResponseSchema
+from mwdb.schema.object import RelationsResponseSchema
 
 from . import access_object, logger, requires_authorization, requires_capabilities
 

mwdb/schema/blob.py

@@ -7,6 +7,7 @@
     ObjectLegacyMetakeysMixin,
     ObjectListItemResponseSchema,
     ObjectListResponseSchemaBase,
+    RelationsResponseSchema,
 )
 from .utils import UTCDateTime
 
@@ -47,3 +48,12 @@ class BlobItemResponseSchema(ObjectItemResponseSchema):
     latest_config = fields.Nested(
         ConfigItemResponseSchema, required=True, allow_none=True
     )
+
+
+class BlobItemAndRelationsResponseSchema(
+    BlobItemResponseSchema, RelationsResponseSchema
+):
+    """
+    This is legacy schema that returns object item along with relations
+    It is awfully slow when object is bound with lots of relatives
+    """

mwdb/schema/config.py

@@ -6,6 +6,7 @@
     ObjectLegacyMetakeysMixin,
     ObjectListItemResponseSchema,
     ObjectListResponseSchemaBase,
+    RelationsResponseSchema,
 )
 
 
@@ -47,6 +48,15 @@ class ConfigItemResponseSchema(ObjectItemResponseSchema):
     cfg = fields.Dict(required=True, allow_none=False)
 
 
+class ConfigItemAndRelationsResponseSchema(
+    ConfigItemResponseSchema, RelationsResponseSchema
+):
+    """
+    This is legacy schema that returns object item along with relations
+    It is awfully slow when object is bound with lots of relatives
+    """
+
+
 class ConfigStatsItemResponseSchema(Schema):
     family = fields.Str(required=True, allow_none=False)
     last_upload = fields.Date(required=True, allow_none=False)

mwdb/schema/file.py

@@ -9,6 +9,7 @@
     ObjectLegacyMetakeysMixin,
     ObjectListItemResponseSchema,
     ObjectListResponseSchemaBase,
+    RelationsResponseSchema,
 )
 
 
@@ -71,5 +72,14 @@ class FileItemResponseSchema(ObjectItemResponseSchema):
     )
 
 
+class FileItemAndRelationsResponseSchema(
+    FileItemResponseSchema, RelationsResponseSchema
+):
+    """
+    This is legacy schema that returns object item along with relations
+    It is awfully slow when object is bound with lots of relatives
+    """
+
+
 class FileDownloadTokenResponseSchema(Schema):
     token = fields.Str(required=True, allow_none=False)

mwdb/schema/object.py

@@ -30,6 +30,18 @@ def validate_key(self, data, **kwargs):
             )
 
 
+class ObjectItemRequestSchema(Schema):
+    exclude_relations = fields.Boolean(
+        truthy={
+            "1",
+            "",
+        },
+        falsy={"0"},
+        missing=False,
+        allow_none=False,
+    )
+
+
 class ObjectCountRequestSchema(Schema):
     query = fields.Str(missing=None)
 
@@ -103,12 +115,6 @@ class ObjectItemResponseSchema(Schema):
     upload_time = UTCDateTime(required=True, allow_none=False)
     favorite = fields.Boolean(required=True, allow_none=False)
 
-    parents = fields.Nested(
-        ObjectListItemResponseSchema, many=True, required=True, allow_none=False
-    )
-    children = fields.Nested(
-        ObjectListItemResponseSchema, many=True, required=True, allow_none=False
-    )
     attributes = fields.Nested(
         AttributeItemResponseSchema, many=True, required=True, allow_none=False
     )
@@ -125,5 +131,26 @@ def get_accessible_attributes(self, data, object, **kwargs):
         return {**data, "attributes": attributes_serialized}
 
 
+class RelationsResponseSchema(Schema):
+    parents = fields.Method("get_parents")
+    children = fields.Nested(
+        ObjectListItemResponseSchema, many=True, required=True, allow_none=False
+    )
+
+    def get_parents(self, obj):
+        parents = obj.query_visible_parents().all()
+        schema = ObjectListItemResponseSchema()
+        return schema.dump(parents, many=True)
+
+
+class ObjectItemAndRelationsResponseSchema(
+    ObjectItemResponseSchema, RelationsResponseSchema
+):
+    """
+    This is legacy schema that returns object item along with relations
+    It is awfully slow when object is bound with lots of relatives
+    """
+
+
 class ObjectCountResponseSchema(Schema):
     count = fields.Int()