Mongo's BsonBinaryWriter serializes my complex document on the server, and Json.NET's BsonDataReader deserializes it to a JObject on the client. After deserialization, I replace _idObjectIds with their string representations and then do the following for references:
foreach ( var reference in rootJObject.Descendants().OfType<JValue>().Where( v => v.Type == JTokenType.Bytes ).ToList() )
reference.Replace( new JObject( new JProperty( "$ref", (string)reference ) ) );
This seems to work but relying on v.Type == JTokenType.Bytes to recognize the ObjectIds concerns me. Bytes is too general and could lead to false positives. I propose to improve the deserialization of BSON to provide a more sure way to recognize the ObjectIds.
In the original JSON with which I seeded the Mongo database, I represented ObjectIds in Mongo's strict JSON format of {$oid: "someObjectId"}. Upon deserializing to JObject, I propose an option that the ObjectIds is represented as such instead of the byte representation.
Mongo's
BsonBinaryWriterserializes my complex document on the server, and Json.NET'sBsonDataReaderdeserializes it to aJObjecton the client. After deserialization, I replace_idObjectIdswith their string representations and then do the following for references:This seems to work but relying on
v.Type == JTokenType.Bytesto recognize theObjectIds concerns me.Bytesis too general and could lead to false positives. I propose to improve the deserialization of BSON to provide a more sure way to recognize theObjectIds.In the original JSON with which I seeded the Mongo database, I represented
ObjectIds in Mongo's strict JSON format of{$oid: "someObjectId"}. Upon deserializing toJObject, I propose an option that theObjectIds is represented as such instead of the byte representation.