123456789_123456789_123456789_123456789_123456789_

Module: ActionController::Live

Relationships & Source Files
Namespace Children
Modules:
Classes:
Exceptions:
Super Chains via Extension / Inclusion / Inheritance
Class Chain:
Defined in: actionpack/lib/action_controller/metal/live.rb,
actionpack/lib/action_controller/test_case.rb

Overview

Mix this module into your controller, and all actions in that controller will be able to stream data to the client as it’s written.

class MyController < ActionController::Base
  include ActionController::Live

  def stream
    response.headers['Content-Type'] = 'text/event-stream'
    100.times {
      response.stream.write "hello world\n"
      sleep 1
    }
  ensure
    response.stream.close
  end
end

There are a few caveats with this module. You cannot write headers after the response has been committed (Response#committed? will return truthy). Calling write or close on the response stream will cause the response object to be committed. Make sure all headers are set before calling write or close on your stream.

You must call close on your stream when you’re finished, otherwise the socket may be left open forever.

The final caveat is that your actions are executed in a separate thread than the main thread. Make sure your actions are thread safe, and this shouldn’t be a problem (don’t share state across threads, etc).

Note that ::Rails includes Rack::ETag by default, which will buffer your response. As a result, streaming responses may not work properly with ::Rack 2.2.x, and you may need to implement workarounds in your application. You can either set the ETag or Last-Modified response headers or remove Rack::ETag from the middleware stack to address this issue.

Here’s an example of how you can set the Last-Modified header if your ::Rack version is 2.2.x:

def stream
  response.headers["Content-Type"] = "text/event-stream"
  response.headers["Last-Modified"] = Time.now.httpdate # Add this line if your Rack version is 2.2.x
  #...
end

Class Method Summary

::ActiveSupport::Concern - Extended

class_methods

Define class methods from given block.

included

Evaluate given block in context of base class, so that you can write class macros here.

prepended

Evaluate given block in context of base class, so that you can write class macros here.

append_features, prepend_features

Instance Attribute Summary

Instance Method Summary

Class Method Details

.live_thread_pool_executor (private)

[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 385

def self.live_thread_pool_executor
  @live_thread_pool_executor ||= Concurrent::CachedThreadPool.new(name: "action_controller.live")
end

Instance Attribute Details

#response_body=(body) (writeonly)

[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 326

def response_body=(body)
  super
  response.close if response
end

Instance Method Details

#log_error(exception) (private)

[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 389

def log_error(exception)
  logger = ActionController::Base.logger
  return unless logger

  logger.fatal do
    message = +"\n#{exception.class} (#{exception.message}):\n"
    message << exception.annotated_source_code.to_s if exception.respond_to?(:annotated_source_code)
    message << "  " << exception.backtrace.join("\n  ")
    "#{message}\n\n"
  end
end

#new_controller_thread (private)

This method is for internal use only.

Spawn a new thread to serve up the controller in. This is to get around the fact that ::Rack isn’t based around IOs and we need to use a thread to stream data from the response bodies. Nobody should call this method except in ::Rails internals. Seriously!

See additional method definition at file actionpack/lib/action_controller/test_case.rb line 28.

[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 377

def new_controller_thread # :nodoc:
  yield
end

#original_new_controller_thread

Disable controller / rendering threads in tests. User tests can access the database on the main thread, so they could open a txn, then the controller thread will open a new connection and try to access data that’s only visible to the main thread’s txn. This is the problem in #23483.

[ GitHub ]

  
# File 'actionpack/lib/action_controller/test_case.rb', line 25

alias_method :original_new_controller_thread, :new_controller_thread

#process(name)

[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 276

def process(name)
  t1 = Thread.current
  locals = t1.keys.map { |key| [key, t1[key]] }

  error = nil
  # This processes the action in a child thread. It lets us return the response
  # code and headers back up the Rack stack, and still process the body in
  # parallel with sending data to the client.
  new_controller_thread {
    ActiveSupport::Dependencies.interlock.running do
      t2 = Thread.current

      # Since we're processing the view in a different thread, copy the thread locals
      # from the main thread to the child thread. :'(
      locals.each { |k, v| t2[k] = v }
      ActiveSupport::IsolatedExecutionState.share_with(t1)

      begin
        super(name)
      rescue => e
        if @_response.committed?
          begin
            @_response.stream.write(ActionView::Base.streaming_completion_on_exception) if request.format == :html
            @_response.stream.call_on_error
          rescue => exception
            log_error(exception)
          ensure
            log_error(e)
            @_response.stream.close
          end
        else
          error = e
        end
      ensure
        # Ensure we clean up any thread locals we copied so that the thread can reused.
        ActiveSupport::IsolatedExecutionState.clear
        locals.each { |k, _| t2[k] = nil }

        @_response.commit!
      end
    end
  }

  ActiveSupport::Dependencies.interlock.permit_concurrent_loads do
    @_response.await_commit
  end

  raise error if error
end

#send_stream(filename:, disposition: "attachment", type: nil)

Sends a stream to the browser, which is helpful when you’re generating exports or other running data where you don’t want the entire file buffered in memory first. Similar to send_data, but where the data is generated live.

Options:

  • :filename - suggests a filename for the browser to use.

  • :type - specifies an HTTP content type. You can specify either a string or a symbol for a registered type with Mime::Type.register, for example :json. If omitted, type will be inferred from the file extension specified in :filename. If no content type is registered for the extension, the default type ‘application/octet-stream’ will be used.

  • :disposition - specifies whether the file will be shown inline or downloaded. Valid values are ‘inline’ and ‘attachment’ (default).

Example of generating a csv export:

send_stream(filename: "subscribers.csv") do |stream|
  stream.write "email_address,updated_at\n"

  @subscribers.find_each do |subscriber|
    stream.write "#{subscriber.email_address},#{subscriber.updated_at}\n"
  end
end
[ GitHub ]

  
# File 'actionpack/lib/action_controller/metal/live.rb', line 355

def send_stream(filename:, disposition: "attachment", type: nil)
  payload = { filename: filename, disposition: disposition, type: type }
  ActiveSupport::Notifications.instrument("send_stream.action_controller", payload) do
    response.headers["Content-Type"] =
      (type.is_a?(Symbol) ? Mime[type].to_s : type) ||
      Mime::Type.lookup_by_extension(File.extname(filename).downcase.delete("."))&.to_s ||
      "application/octet-stream"

    response.headers["Content-Disposition"] =
      ActionDispatch::Http::ContentDisposition.format(disposition: disposition, filename: filename)

    yield response.stream
  end
ensure
  response.stream.close
end