Deployment Zone

Building off my previous ~/.pryrc I wanted to automatically load up my core project Ruby file, spec_helper.rb, and fire off some initialization routines whenever I start a new Pry session in my project's directory.

Because the contents of the local directory's .pryrc is evaluated before the :before_session hook from ~/.pryrc timing is a bit more delicate. I get around this by creating a custom function named _pry_before_session (but you could name it anything you want really) and have the ~/.pryrc's before_session hook execute it if it exists.

So my project's .pryrc:

#~/Projects/ActiveAvro/.pryrc
def _pry_before_session
  require 'active_avro'
  require 'spec_helper'
  ActiveAvroHelper.initialize
end

And my updated ~/.pryrc looks like this:

#~/.pryrc
require 'interactive_editor'

Pry.config.editor = "mate"
# add the current directories /lib and /spec directories to the path if they exist
before_session = Proc.new do |out, target, _pry_|
  dir = `pwd`.chomp
  %w(lib spec test).map{ |d| "#{dir}/#{d}" }.each { |p| $: << p unless !Dir.exists?(p) || $:.include?(p) }
  # if a local .pryrc defines a _pry_before_session function, execute it now
  send(:_pry_before_session) rescue nil
end
Pry.hooks[:before_session] = before_session

I use Pry a lot. If I had to develop without it I'd spend a hell of a lot more time not writing Ruby code. Often I use Pry in a Rails project as part of rails console but I also use it with my non-Rails Ruby projects. I wanted to get around having to append commonly used directories to the $LOAD_PATH each time I fire up a Pry session.

Ruby code you place in your ~/.pryrc is executed when your Pry session begins.

Here's the relevant lines from my ~/.pryrc that add the appropriate paths to $LOAD_PATH during start up.

#~/.pryrc
⋮
# add the current directories /lib and /spec directories to the path if they exist
before_session = Proc.new do |out, target, _pry_|
  dir = `pwd`.chomp
  %w(lib spec test).map{ |d| "#{dir}/#{d}" }.each { |p| $: << p unless !Dir.exists?(p) || $:.include?(p) }
end
Pry.hooks = { :before_session => before_session }

This post basically sums the bug with the Ruby Thrift bindings where the exception message is "Incompatible character encodings: ASCII-8BIT and UTF-8". This problem is a bit of a bitch to hunt down but once you find it its relatively easy to fix.

While I've got a fork with a pull request I'm fairly certain that the Apache software foundation has other... means of accepting patches so this pull request will be largely irrelevant.

Until the problem is fixed and propagated to the thrift gem you can monkey patch this issue yourself:

# encoding: utf-8
module Thrift
  UTF8_ENCODING = "utf-8"
  class BinaryProtocol
    def write_string(str)
      write_i32(str.bytesize)
      trans.write(str)
    end
  end

  class HTTPClientTransport < BaseTransport
    def write(buf)
      puts "write"
      @outbuf << buf.force_encoding(UTF8_ENCODING)
    end
  end
  class FramedTransport < BaseTransport
    def write(buf,sz=nil)
      buf.force_encoding(UTF8_ENCODING)
      return @transport.write(buf) unless @write

      @wbuf << (sz ? buf[0...sz] : buf)
    end
    def flush
      return @transport.flush unless @write
      out = [@wbuf.length].pack('N')
      out.force_encoding(UTF8_ENCODING)
      out << @wbuf
      @transport.write(out)
      @transport.flush
      @wbuf = ''
    end
  end
  class BufferedTransport < BaseTransport
    def write(buf)
      @wbuf << buf.force_encoding(UTF8_ENCODING)
    end

    def flush
      if @wbuf != ''
        @wbuf.force_encoding(UTF8_ENCODING)
        @transport.write(@wbuf)
        @wbuf = ''
      end

      @transport.flush
    end
  end
end

While I can't vouch for the production worthiness of the above code I can say it at least gets me past an aggravating hurdle.

(This was written regarding the RubyMine 4.0 EAP but it probably applies to many of the earlier versions as well.)

When you Run > Edit Configurations in RubyMine you can create a number of different configurations. One of these is a regular Ruby script which can be useful for running such tasks as guard. You select the full path to the *.rb file and then you click the Share configuration check box. You might wonder how you express those paths as relative to the project path; after all you want to share the configuration with your teammates and their paths won't likely match your own.

Well it turns out there's no secret sauce to make this work; RubyMine will substitute the macro value $PROJECT_DIR$ if the script's path is relative to the project you currently have open. In the UI you'll see the fully expanded path, but when you commit your shared project settings the macro value is what is actually recorded in the *.xml file. When your teammates pull down your shared configurations it will be relative to their project's path.

So to get Guard (with Spork) running in RubyMine I did the following:

1. Created a run_guard.rb file in my project's root

#./run_guard.rb
exec 'guard'

2. RubyMine: Run > Edit Configurations
3. Hit the + and add a new Ruby script
4. Name it Guard, select the full path to the run_guard.rb, select the Bundler tab, check run in the context of Bundler (for good measure)
5. .idea is correctly in my .gitignore file... but the shared configs live under .idea/sharedConfigurations/
6. git add -f .idea/sharedConfigurations/Guard.xml and commit, push

Run the new Guard configuration and you'll see it display in one of the output panes at the bottom of the screen by default. When you save your specs it will automatically run as normal (but beware that ⌘S saves all files).

We also found out that pry statements in your RubyMine hosted Guard process won't allow you to interact with the running test session; instead the process will just hang.

Update: some more useful RSpec information with RubyMine:

Built in spec task: If your spec task is dying because you've configured a DRb server via Guard (using Spork) and you don't plan on running the built-in Spork task within RubyMine you'll need to make some changes to your Spork.prefork block and add an environment variable for the Guard script itself (Run > Edit Configurations) named "RUBYMINE_HOME" and point it to your RubyMine's application path (i.e. for me its /Applications/RubyMine EAP.app/). Now when Guard is running your built-in rspec task can execute without dying.

growlnotify: if you still want Growl notifications for your RubyMine hosted Guard you may need to sudo ln -s /usr/local/bin/growlnotify /usr/bin/growlnotify.

I got a request over the weekend to support a client library on the Android platform that cannot submit multipart form data easily. The request was to allow the client to send the raw bytes as the HTTP request body. In order to implement this functionality I of course wanted to spec it out first to make sure I was writing this correctly - it isn't every day that I deal with the raw request body after all.

My spec ended up looking like this...

#./spec/controllers/file_test_api_controller_spec.rb
⋮
  describe "#raw_upload" do
    before do
      # file_path points to a file I used dd to generate that is 11,776 bytes
      bytes = File.open(file_path, 'rb').read
      @request.env['RAW_POST_DATA'] = bytes
      post :raw_upload, { }, 'CONTENT_TYPE' => 'application/octet-stream'
      @request.env.delete('RAW_POST_DATA')
    end
    subject { response }
    its "status" do
      should == 200
    end
    its "body" do
      should =~ /size.*11776/m
    end
  end
⋮

To access this data directly in your controller, you just use request.raw_post.

Its worth noting that my controller replies with a receipt in JSON that includes the size of the received data which is why the spec regexes the body for it - there's no automagic occurring with my controller's response.

This approach does not work with HTTP PUT as the RAW_POST_DATA request environment variable is cleared when a PUT is performed.

Reference: How to send raw post data in a Rails functional test?