faye/websocket-driver-ruby
1
0
Fork 0
mirror of https://github.com/faye/websocket-driver-ruby.git synced 2025-11-01 13:59:38 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove Faye::WebSocket docs.

Browse Source
This commit is contained in:
James Coglan
2013-04-05 09:11:05 +01:00
parent 6fd2c32b60
commit 35e032aca4
1 changed files with 1 additions and 341 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.rdoc
+1 -341
View File
@@ -1,344 +1,4 @@
= Faye::WebSocket
* Travis CI build: {<img src="https://secure.travis-ci.org/faye/faye-websocket-ruby.png" />}[http://travis-ci.org/faye/faye-websocket-ruby]
* Autobahn tests: {server}[http://faye.jcoglan.com/autobahn/servers/], {client}[http://faye.jcoglan.com/autobahn/clients/]
This is a robust, general-purpose WebSocket implementation extracted from the
{Faye}[http://faye.jcoglan.com] project. It provides classes for easily building
WebSocket servers and clients in Ruby. It does not provide a server itself, but
rather makes it easy to handle WebSocket connections within an existing
{Rack}[http://rack.rubyforge.org/] application. It does not provide any
abstraction other than the standard
{WebSocket API}[http://dev.w3.org/html5/websockets/].
It also provides an abstraction for handling {EventSource}[http://dev.w3.org/html5/eventsource/]
connections, which are one-way connections that allow the server to push data to
the client. They are based on streaming HTTP responses and can be easier to
access via proxies than WebSockets.
Currently, the following web servers are supported, and can be accessed directly
or via HAProxy:
* {Thin}[http://code.macournoyer.com/thin/]
* {Rainbows}[http://rainbows.rubyforge.org/] using EventMachine
* {Goliath}[http://postrank-labs.github.com/goliath/]
The server-side socket can process {draft-75}[http://tools.ietf.org/html/draft-hixie-thewebsocketprotocol-75],
{draft-76}[http://tools.ietf.org/html/draft-hixie-thewebsocketprotocol-76],
{hybi-07}[http://tools.ietf.org/html/draft-ietf-hybi-thewebsocketprotocol-07]
and later versions of the protocol. It selects protocol versions automatically,
supports both +text+ and +binary+ messages, and transparently handles +ping+,
+pong+, +close+ and fragmented messages.
== Handling WebSocket connections in Rack
You can handle WebSockets on the server side by listening for requests using the
<tt>Faye::WebSocket.websocket?</tt> method, and creating a new socket for the
request. This socket object exposes the usual WebSocket methods for receiving
and sending messages. For example this is how you'd implement an echo server:
# app.rb
require 'faye/websocket'
App = lambda do |env|
if Faye::WebSocket.websocket?(env)
ws = Faye::WebSocket.new(env)
ws.onmessage = lambda do |event|
ws.send(event.data)
end
ws.onclose = lambda do |event|
p [:close, event.code, event.reason]
ws = nil
end
# Return async Rack response
ws.rack_response
else
# Normal HTTP request
[200, {'Content-Type' => 'text/plain'}, ['Hello']]
end
end
This is a standard Rack app, so it can be run using a <tt>config.ru</tt> file.
However, so that incoming requests can be properly prepared to process WebSocket
connections, you need to tell <tt>Faye::WebSocket</tt> which adapter to load;
this can be either +thin+, +rainbows+ or +goliath+. If one of these servers is
already loaded before <tt>faye/websocket</tt> is loaded, it will load
appropriate adapters automatically.
# config.ru
require './app'
Faye::WebSocket.load_adapter('thin')
run App
Note that under certain circumstances (notably a draft-76 client connecting
through an HTTP proxy), the WebSocket handshake will not be complete after you
call `Faye::WebSocket.new` because the server will not have received the entire
handshake from the client yet. In this case, calls to `ws.send` will buffer the
message in memory until the handshake is complete, at which point any buffered
messages will be sent to the client.
If you need to detect when the WebSocket handshake is complete, you can use the
`onopen` event.
If the connection's protocol version supports it, you can call <tt>ws.ping()</tt>
to send a ping message and wait for the client's response. This method takes a
message string, and an optional callback that fires when a matching pong message
is received. It returns +true+ iff a ping message was sent. If the client does
not support ping/pong, this method sends no data and returns +false+.
ws.ping 'Mic check, one, two' do
# fires when pong is received
end
== Using the WebSocket client
The client supports both the plain-text +ws+ protocol and the encrypted +wss+
protocol, and has exactly the same interface as a socket you would use in a web
browser. On the wire it identifies itself as hybi-13.
require 'faye/websocket'
require 'eventmachine'
EM.run {
ws = Faye::WebSocket::Client.new('ws://www.example.com/')
ws.onopen = lambda do |event|
p [:open]
ws.send('Hello, world!')
end
ws.onmessage = lambda do |event|
p [:message, event.data]
end
ws.onclose = lambda do |event|
p [:close, event.code, event.reason]
ws = nil
end
}
== Subprotocol negotiation
The WebSocket protocol allows peers to select and identify the application
protocol to use over the connection. On the client side, you can set which
protocols the client accepts by passing a list of protocol names when you
construct the socket:
ws = Faye::WebSocket::Client.new('ws://www.example.com/', ['irc', 'amqp'])
On the server side, you can likewise pass in the list of protocols the server
supports after the other constructor arguments:
ws = Faye::WebSocket.new(env, ['irc', 'amqp'])
If the client and server agree on a protocol, both the client- and server-side
socket objects expose the selected protocol through the <tt>ws.protocol</tt>
property. If they cannot agree on a protocol to use, the client closes the
connection.
== WebSocket API
The WebSocket API consists of several event handlers and a method for sending
messages.
* <b><tt>onopen</tt></b> fires when the socket connection is established. Event
has no attributes.
* <b><tt>onerror</tt></b> fires when the connection attempt fails. Event has no
attributes.
* <b><tt>onmessage</tt></b> fires when the socket receives a message. Event has
one attribute, <b><tt>data</tt></b>, which is either a +String+ (for text
frames) or an +Array+ of byte-sized integers (for binary frames).
* <b><tt>onclose</tt></b> fires when either the client or the server closes the
connection. Event has two optional attributes, <b><tt>code</tt></b> and
<b><tt>reason</tt></b>, that expose the status code and message sent by the
peer that closed the connection.
* <b><tt>send(message)</tt></b> accepts either a +String+ or an +Array+ of
byte-sized integers and sends a text or binary message over the connection to
the other peer.
* <b><tt>close(code, reason)</tt></b> closes the connection, sending the given
status code and reason text, both of which are optional.
* <b><tt>protocol</tt></b> is a string (which may be empty) identifying the
subprotocol the socket is using.
== Handling EventSource connections in Rack
EventSource connections provide a very similar interface, although because they
only allow the server to send data to the client, there is no +onmessage+ API.
EventSource allows the server to push text messages to the client, where each
message has an optional event-type and ID.
# app.rb
require 'faye/websocket'
App = lambda do |env|
if Faye::EventSource.eventsource?(env)
es = Faye::EventSource.new(env)
p [:open, es.url, es.last_event_id]
# Periodically send messages
loop = EM.add_periodic_timer(1) { es.send('Hello') }
es.onclose = lambda do |event|
EM.cancel_timer(loop)
es = nil
end
# Return async Rack response
es.rack_response
else
# Normal HTTP request
[200, {'Content-Type' => 'text/plain'}, ['Hello']]
end
end
The +send+ method takes two optional parameters, <tt>:event</tt> and
<tt>:id</tt>. The default event-type is <tt>'message'</tt> with no ID. For
example, to send a +notification+ event with ID +99+:
es.send('Breaking News!', :event => 'notification', :id => '99')
The +EventSource+ object exposes the following properties:
* <b><tt>url</tt></b> is a string containing the URL the client used to create
the EventSource.
* <b><tt>last_event_id</tt></b> is a string containing the last event ID
received by the client. You can use this when the client reconnects after a
dropped connection to determine which messages need resending.
When you initialize an EventSource with <tt>Faye::EventSource.new</tt>, you can
pass configuration options after the +env+ parameter. Available options are:
* <b><tt>:retry</tt></b> is a number that tells the client how long (in seconds)
it should wait after a dropped connection before attempting to reconnect.
* <b><tt>:ping</tt></b> is a number that tells the server how often (in seconds)
to send 'ping' packets to the client to keep the connection open, to defeat
timeouts set by proxies. The client will ignore these messages.
For example, this creates a connection that pings every 15 seconds and is
retryable every 10 seconds if the connection is broken:
es = Faye::EventSource.new(es, :ping => 15, :retry => 10)
You can send a ping message at any time by calling <tt>es.ping()</tt>. Unlike
WebSocket the client does not send a response to this; it is merely to send some
data over the wire to keep the connection alive.
== Running your socket application
To use this library you must be using an EventMachine-based server; currently
Thin, Rainbows and Goliath are supported.
=== Running the app with Thin
Thin can be started via the command line if you've set up a <tt>config.ru</tt>
file for your application:
thin start -R config.ru -p 9292
Or, you can use +rackup+. In development mode, this adds middlewares that don't
work with async apps, so you must start it in production mode:
rackup config.ru -s thin -E production -p 9292
It can also be started using the <tt>Rack::Handler</tt> interface common to many
Ruby servers. It must be run using EventMachine, and you can configure Thin
further in a block passed to +run+:
require 'eventmachine'
require 'rack'
require 'thin'
require './app'
EM.run {
thin = Rack::Handler.get('thin')
thin.run(App, :Port => 9292) do |server|
# You can set options on the server here, for example to set up SSL:
server.ssl_options = {
:private_key_file => 'path/to/ssl.key',
:cert_chain_file => 'path/to/ssl.crt'
}
server.ssl = true
end
}
=== Running the app with Rainbows
<tt>Faye::WebSocket</tt> can only be run using EventMachine. To begin with,
you'll need a Rainbows config file that tells it to use EventMachine, along with
whatever Rainbows/Unicorn configuration you require.
# rainbows.conf
Rainbows! do
use :EventMachine
end
You can then run your <tt>config.ru</tt> file from the command line. Again,
<tt>Rack::Lint</tt> will complain unless you put the application in production
mode.
rainbows config.ru -c path/to/rainbows.conf -E production -p 9292
Rainbows also has a Ruby API for starting a server:
require 'rainbows'
require './app'
rackup = Unicorn::Configurator::RACKUP
rackup[:port] = 9292
rackup[:set_listener] = true
options = rackup[:options]
options[:config_file] = 'path/to/rainbows.conf'
server = Rainbows::HttpServer.new(App, options)
# This is non-blocking; use server.start.join to block
server.start
=== Running the app with Goliath
Goliath can be made to run arbitrary Rack apps by delegating to them from a
<tt>Goliath::API</tt> instance. A simple server looks like this:
require 'goliath'
require './app'
class EchoServer < Goliath::API
def response(env)
App.call(env)
end
end
<tt>Faye::WebSocket</tt> can also be used inline within a Goliath app:
require 'goliath'
require 'faye/websocket'
class EchoServer < Goliath::API
def response(env)
ws = Faye::WebSocket.new(env)
ws.onmessage = lambda do |event|
ws.send(event.data)
end
ws.rack_response
end
end
= faye-websocket-parser {<img src="https://secure.travis-ci.org/faye/faye-websocket-parser-ruby.png" />}[http://travis-ci.org/faye/faye-websocket-ruby]
== License