A newer version of this documentation is available.

View Latest

Transcoders & Non-JSON Documents

    +
    The .NET SDK supports common JSON document requirements out-of-the-box. Custom transcoders and serializers provide support for applications needing to perform advanced operations, including supporting non-JSON data.

    The .NET SDK uses the concepts of transcoders and serializers, which are used whenever data is sent to or retrieved from Couchbase Server.

    When sending data to Couchbase, the SDK passes the Object being sent to a transcoder. The transcoder can either reject the Object as being unsupported, or convert it into a byte[] and a Common Flag. The Common Flag specifies whether the data is JSON, a non-JSON string, or raw binary data. It may, but does not have to, use a serializer to perform the byte conversion.

    On retrieving data from Couchbase, the fetched byte[] and Common Flag are passed to a transcoder. The transcoder converts the bytes into a concrete class (the application specifies the required type) if possible. It may use a serializer for this.

    Many applications will not need to be aware of transcoders and serializers, as the defaults support most standard JSON use cases. The information in this page is only needed if the application has an advanced use-case, likely involving either non-JSON data, or a requirement for a particular JSON serialization library.

    Default Behaviour

    The ClusterOptions contains a global transcoder and serializer, which by default are JsonTranscoder and DefaultSerializer.

    DefaultSerializer uses the high-performance JSON library JSON.NET for serializing and deserializing byte arrays to and from concrete objects.

    On sending data to Couchbase, JsonTranscoder will send Objects to its serializer (DefaultSerializer by default) to convert into a byte[]. The serialized bytes are then sent to the Couchbase Server, along with a Common Flag of JSON.

    JsonTranscoder will pass any Object to its serializer, apart from a byte[]. It will reject this with an InvalidArgumentException, as it is ambiguous how it should be handled.

    On retrieving data from Couchbase, JsonTranscoder passes the fetched byte[] and Common Flag to its serializer (DefaultSerializer by default) to convert into a concrete class.

    This table summarizes that information, and this more concise form will be used to describe the other transcoders included in the SDK.

    Item Result Common Flag

    String

    Results of serializer

    JSON

    byte[]

    InvalidArgumentException

    -

    Other Object

    Results of serializer

    JSON

    RawJsonTranscoder

    The RawJsonTranscoder provides the ability for the application to explicitly specify that the data they are storing or retrieving is JSON. This transcoder does not accept a serializer, and always performs straight pass through of the data to the server. This enables the application to avoid unnecessary parsing costs when they are certain they are using JSON data.

    It only accepts Strings and byte[].

    Item Result Common Flag

    String

    Passthrough

    JSON

    byte[]

    Passthrough

    JSON

    Other Object

    InvalidArgumentException

    -

    This transcoder is particularly useful when working with third-party JSON libraries. Here we want to use System.Text.Json for serialization, instead of the default JSON.NET:

    var userBytes = JsonSerializer.SerializeToUtf8Bytes(new User
    {
        Name = "John Smith",
        Age = 27
    }, typeof(User));
    
    await _collection.UpsertAsync("john-smith", userBytes, options => options.Transcoder(new RawJsonTranscoder()));

    Since Gson has already done the serialization work, we don’t want to use the default JsonTranscoder, as this will run the provided String needlessly through DefaultSerializer (Jackson). Instead, RawJsonTranscoder is used, which just passes through the serialized bytes, and stores them in Couchbase with the JSON Common Flag set.

    Similarly, the same transcoder is used on reading the document, so the raw bytes can be retrieved in a String without going through DefaultSerializer (Jackson). Gson can then be used for the deserialization.

    var result =
        await _collection.GetAsync("john-smith", options => options.Transcoder(new RawJsonTranscoder()));
    
    var returnedJson = result.ContentAs<byte[]>();
    var user = JsonSerializer.Deserialize(returnedJson, typeof(User));

    Non-JSON Transcoders

    It is most common to store JSON with Couchbase. However, it is possible to store non-JSON documents, such as raw binary data, perhaps using an concise binary encoding like MessagePack or CBOR, in the Key-Value store.

    It’s important to note that the Couchbase Data Platform includes multiple components other than the Key-Value store — including N1QL and its indexes, FTS, analytics, and eventing — and these are optimized for JSON and will either ignore or provide limited functionality with non-JSON documents.

    Also note that some simple data types can be stored directly as JSON, without recourse to non-JSON transcoding. A valid JSON document can be a simple integer (42), string ("hello"), array ([1,2,3]), boolean (true, false) and the JSON null value.

    RawStringTranscoder

    The RawStringTranscoder provides the ability for the user to explicitly store and retrieve raw string data with Couchbase. It can be used to avoid the overhead of storing the string as JSON, which requires two bytes for double quotes, plus potentially more for escaping characters.

    Note that this transcoder does not accept a serializer, and always performs straight passthrough of the data to the server. It only accepts Strings.

    Item Result Common Flag

    String

    Passthrough

    String

    byte[]

    InvalidArgumentException

    -

    Other Object

    InvalidArgumentException

    -

    Here’s an example of using the RawStringTranscoder:

    var docId = "doc";
    
    await _collection.UpsertAsync<string>(docId, "hello world",
        options =>
        {
            options.Transcoder(new RawStringTranscoder());
            options.Timeout(TimeSpan.FromMinutes(1000));
        });
    
    
    var result = await _collection.GetAsync(docId, options => options.Transcoder(new RawStringTranscoder()));
    
    var returned = result.ContentAs<string>();

    RawBinaryTranscoder

    The RawBinaryTranscoder provides the ability for the user to explicitly store and retrieve raw byte data to Couchbase. The transcoder does not perform any form of real transcoding, and does not take a serializer, but rather passes the data through and assigns the appropriate binary Common Flag.

    Item Result Common Flag

    String

    InvalidArgumentException

    -

    byte[]

    Passthrough

    Binary

    Other Object

    InvalidArgumentException

    -

    Here’s an example of using the RawBinaryTranscoder:

    var docId = "doc";
    
    var bytes = System.Text.Encoding.UTF8.GetBytes("hello world");
    
    await _collection.UpsertAsync(docId, bytes, options => options.Transcoder(new RawBinaryTranscoder()));
    
    var result = await _collection.GetAsync(docId, options => options.Transcoder(new RawBinaryTranscoder()));
    
    var returned = result.ContentAs<byte[]>();

    Custom Transcoders and Serializers

    More advanced transcoding needs can be accomplished if the application implements their own transcoders and serializers.

    Creating a Custom Serializer

    We saw above one example of using Google Gson with the RawJsonTranscoder, but it requires the application to explicitly serialize and deserialize objects each time. By creating a custom Gson serializer, we can avoid this.

    It’s easy to create a serializer. Simply implement the ITypeSerializer interface’s three methods:

    public class DotnetJsonSerializer : ITypeSerializer
    {
        public T Deserialize<T>(ReadOnlyMemory<byte> buffer)
        {
            return JsonSerializer.Deserialize<T>(buffer.Span);
        }
    
        public T Deserialize<T>(Stream stream)
        {
            using var ms = new MemoryStream();
            stream.CopyTo(ms);
            var span = new ReadOnlySpan<byte>(ms.GetBuffer()).Slice(0, (int)ms.Length);
            return JsonSerializer.Deserialize<T>(span);
        }
    
        public ValueTask<T> DeserializeAsync<T>(Stream stream, CancellationToken cancellationToken = default)
        {
            return JsonSerializer.DeserializeAsync<T>(stream, null, cancellationToken);
        }
    
        public void Serialize(Stream stream, object obj)
        {
            using var jsonUtf8Writer = new Utf8JsonWriter(stream);
            JsonSerializer.Serialize(jsonUtf8Writer, obj);
        }
    
        public ValueTask SerializeAsync(Stream stream, object obj, CancellationToken cancellationToken = default)
        {
            return new ValueTask(JsonSerializer.SerializeAsync(stream, obj, null, cancellationToken));
        }
    }

    In this case, there is no need to provide a custom transcoder. The table for JsonTranscoder shows that it already does what we need: for any Object (that’s not a byte[]), it sends it to its serializer, and then stores the result in Couchbase with the JSON Common Flag set. All we need to do is change the serializer, as so:

    var serializer = new DotnetJsonSerializer();
    var transcoder = new JsonTranscoder(serializer);
    
    var user = new User
    {
        Name = "John Smith",
        Age = 27
    };
    
    await _collection.UpsertAsync("john-smith", user, options => options.Transcoder(transcoder));

    And for decoding:

    var serializer = new DotnetJsonSerializer();
    var transcoder = new JsonTranscoder(serializer);
    
    var user = new User
    {
        Name = "John Smith",
        Age = 27
    };
    
    var result = await _collection.GetAsync("john-smith", options => options.Transcoder(transcoder));
    var returnedUser = result.ContentAs<User>();

    Currently its not suggested that a custom JSON serializer be used globally in the Couchbase .NET SDK for anything other than K/V. This is because of streaming optimizations used in Query, FTS and Search that use JSON.NET features.

    var clusterOptions = new ClusterOptions().WithSerializer(new DotnetJsonSerializer());
    var cluster = await Cluster.ConnectAsync("couchbase://localhost", clusterOptions);
    
    
    var results = await cluster.QueryAsync<dynamic>("SELECT * FROM `default`");
    await foreach (var result in results)
    {
        Console.WriteLine(result);
    }

    Creating a Custom Transcoder

    Let’s look at a more complex example: encoding the JSON alternative, MessagePack. MessagePack is a compact binary data representation, so it should be stored with the binary Common Flag. The Common Flag is chosen by the transcoder, and none of the existing transcoders matches our needs (RawBinaryTranscoder does set the binary flag, but it passes data through directly rather than using a serializer). So we need to write one.

    Start by creating a new serializer for MessagePack. This is similar to the GsonSerializer example above:

    public class MsgPackSerializer : ITypeSerializer
    {
        public T Deserialize<T>(ReadOnlyMemory<byte> buffer)
        {
            return MessagePackSerializer.Deserialize<T>(buffer);
        }
    
        public T Deserialize<T>(Stream stream)
        {
            return MessagePackSerializer.Deserialize<T>(stream);
        }
    
        public ValueTask<T> DeserializeAsync<T>(Stream stream, CancellationToken cancellationToken = default)
        {
            return MessagePackSerializer.DeserializeAsync<T>(stream, null, cancellationToken);
        }
    
        public void Serialize(Stream stream, object obj)
        {
            MessagePackSerializer.Serialize(stream, obj);
        }
    
        public ValueTask SerializeAsync(Stream stream, object obj, CancellationToken cancellationToken = default)
        {
            return new ValueTask(MessagePackSerializer.SerializeAsync(stream, obj, null, cancellationToken));
        }
    }

    And now create a transcoder that uses the MsgPackSerializer, and sets the binary Common Flag when storing the data:

    public class MsgPackTranscoder : BaseTranscoder
    {
        public MsgPackTranscoder() : this(new MsgPackSerializer())
        {
        }
    
        internal MsgPackTranscoder(MsgPackSerializer serializer)
        {
            Serializer = serializer;
        }
    
        public override Flags GetFormat<T>(T value)
        {
            var typeCode = Type.GetTypeCode(typeof(T));
            var dataFormat = DataFormat.Binary;
            return new Flags { Compression = Compression.None, DataFormat = dataFormat, TypeCode = typeCode };
        }
    
        public override void Encode<T>(Stream stream, T value, Flags flags, OpCode opcode)
        {
            Serializer.Serialize(stream, value);
        }
    
        public override T Decode<T>(ReadOnlyMemory<byte> buffer, Flags flags, OpCode opcode)
        {
            return Serializer.Deserialize<T>(buffer);
        }
    }

    Note the use of DataFormat.Binary. The other Common Flags that can be used are DataFormat.Json and DataFormat.String.

    Create a POCO properly adjorned with the MessagePack attributes:

    [data-source-url=https://github.com/couchbase/docs-sdk-dotnet/blob/8b18ecd4811cf2c86a06daab920c160edceca6d9/modules/howtos/examples/Couchbase.Examples/Transcoding.cs]

    Now we can use the new transcoder to seamlessly store MessagePack data in Couchbase Server:

    var serializer = new MsgPackSerializer();
    var transcoder = new MsgPackTranscoder(serializer);
    
    var user = new User2
    {
        Name = "John Smith",
        Age = 27
    };
    
    await _collection.UpsertAsync("john-smith", user, options => options.Transcoder(transcoder));

    Further reading