I am currently working on an app that has a lot of legacy data from the backend. This means types are very unreliable. I can get a boolean like these examples "canJump": "1" "canJump": 0 "canJump": "true" "canJump": false. So I have been pondering about good ways to handle decoding this. I would like to avoid: if let bool = try? container.decode(Bool.self, forKey: .jump) { self.canJump = bool } if let boolAsString = try? container.decode(String.self, forKey: .jump) { if boolAsString == "true" || boolAsString == "false" { self.canJump = Bool(boolAsString) } } etc. in every single custom decoder. So I tried an extension: extension KeyedDecodingContainer { func decodeWithUnknownTypes(_ type: Bool.Type, forKey key: KeyedDecodingContainer<K>.Key) throws -> Bool { do { return try decode(Bool.self, forKey: key) } catch DecodingError.typeMismatch {} do { let string = try decode(String.self, forKey: key) return Bool(string.lowercased()) } catch DecodingError.typeMismatch {} do { let int = try decode(Int.self, forKey: key) if int == 1 { return true } if int == 0 { return false } } throw DecodingError.typeMismatch } } (not entirely finished). The above almost does the trick, but it still leaves me having to implement custom decoders in basically all my backend calls past and future. I would love it if someone has an idea of how I could write it, so I can simply use a Decodable struct, and it would automagically accept 1 as a Bool. Can I maybe override the decode method of KeyedDecodingContainer and still get the original function as a "first try"?