buffer-to-messages

buffer-to-messages (NodeJS)

Small helper class to go about sending messages through and recovering messages from a bufferstream, using length-prefixes.

This is built with TCP sockets in mind, where the order of delivery is fixed, but where the actual packets might be split up or merged arbitrarily.

Setup

Get the package from NPM

npm install buffer-to-messages

Now get the class in your application

var Converter = require('buffer-to-messages');

API & walkthrough

Constructing an instance

/**
 * @param {Integer} [prefixLength] - The length in bytes of the size prefix.
 * This argument is optional. The default length is 2 bytes, but can also be 1
 * or 4 bytes.
 *
 * @param {Function} callback - This will be called every time a new message has
 * been completely processed.
 */
var Converter = function Converter()

Every message needs to be prefixed with a buffer containing it's length. By default the size of that prefix is 2 bytes (16bit) but you can configure it to be 1 byte (8bit) or 4 bytes (32bit) as well.

var prefixLength = 2;

var converter = new Converter(prefixLength, function(message) {
  console.log('got message: ', message.toString());
});

Converter.process

/**
 * Process the given buffer.
 * @param  {Buffer}   buffer - Process a new buffer snippet
 * @param  {Function} [cb] - Override the callback just for this snippet. This
 * is used for testing purposes.
 */
Converter.prototype.process = function(b, cb)

Your messages require a 1, 2 or 4-byte prefix indicating their length. The class provides methods to generate such a prefix from an existing buffer, but for now we'll do it manually.

// A 2-byte length prefix, indicating a message length of 4
converter.process(new Buffer([0]));
converter.process(new Buffer([4]));

// Also process the message,in two parts
converter.process(new Buffer('tes'));
converter.process(new Buffer('t'));

The message has been completely processed, so our callback is executed:

>> got message: test

Converter.flush

/**
 * This flushes the converter instance. It clears all the cached bytes and
 * resets the state. Use this whenever the processor might go out of sync (e.g.
 * when a tcp connection times out and you reconnect)
 */
Converter.prototype.flush = function()

Illustration:

socket.on('end', function() {
  converter.flush();
  // reconnect
});

Converter.createPrefix

/**
 * This creates a new buffer containing the length prefix of the given buffer
 *
 * @param {Buffer} buffer
 * @return {Buffer} - A buffer containing a compatible length-prefix
 */
Converter.prototype.createPrefix = function(buffer)

Illustration:

message = new Buffer('testMessage');
prefix = converter.createPrefix(message);

converter.process(prefix);

converter.process(message);

This will successfully parse your message

>> got message: testMessage

License

MIT