Rails3, Cucumber, Capybara, File upload => bad content body (EOFError)? - file-upload

I'm having a hard time getting a simple file upload test working. I'm using Rails 3.0.0 on ruby 1.9.2 with Cucumber and Capybara.
View:
<%= form_tag "/upload/create", :multipart => true do %>
<label for="file">File to Upload:</label>
<%= file_field_tag "file" %>
<%= submit_tag "Upload" %>
<% end %>
Cucumber Step:
When /^I upload the basic file$/ do
visit path_to("upload")
path = File.join(::Rails.root, "somefile")
attach_file("file", path)
click_button("Upload")
end
In my controller, i have commented out everything except for:
def create
file = params[:file]
end
Gemfile snippet:
group :development, :test do
# testing with specs
gem "ZenTest", ">= 4.3.3"
gem "autotest"
gem "rspec-rails", ">= 2.0.0.beta.19", :git => "git://github.com/rspec/rspec-rails.git"
gem "rspec", :git => "git://github.com/rspec/rspec.git"
gem "rspec-core", :git => "git://github.com/rspec/rspec-core.git"
gem "rspec-expectations", :git => "git://github.com/rspec/rspec-expectations.git"
gem "rspec-mocks", :git => "git://github.com/rspec/rspec-mocks.git"
# cucumber stuff
gem 'capybara'
gem 'database_cleaner'
gem 'cucumber-rails'
gem 'cucumber'
gem 'spork'
gem 'launchy' # So you can do Then show me the page
gem 'escape_utils' # needed to fix Cucumber - http://crimpycode.brennonbortz.com/?p=42
end
When I try to run the test, I receive:
(::) failed steps (::)
bad content body (EOFError)
<internal:prelude>:10:in `synchronize'
I appreciate any help or insight. Thanks.

This turned out to be an issue with rack-test and it probably won't be a problem for most until more people adopt Rails3 and Ruby 1.9.x.
Upgrading rack-test to the current master branch fixed the problem.
I'm not sure when rack-test will include these changes in the gem.
See also:
groups.google.com/group/cukes/browse_thread/thread/5028306893c2c54a

I don't have an answer but working on the same problem in the same environ - cukes, capybara, rails 3, 1.9.2.... if I figure this out will let you know. Have you thought of posting on the cucumber google group or the Rails google group? If you don't once I get my act together and cant figure out will post to one of these.
Also, it seems that webrat has the method for attach_file() and thus when I generated cucumber without capybara it had a corollary method in web_steps.rb, but after I added capybara and regenerated cucumber it was gone....

Related

Organizing rspec 2 tests into 'unit' and 'integration' categories in rails

How to organize rspec 2 tests into 'unit' (fast) and 'integration' (slow) categories?
I want to be able to run all unit tests with just rspec command, but not the 'integration' tests.
I want to be able to run only 'integration' tests.
We have groups of the same nature.
We run then one by one both on the local dev boxes and on the CI.
you can simply do
bundle exec rake spec:unit
bundle exec rake spec:integration
bundle exec rake spec:api
This is what our spec.rake looks like
namespace :spec do
RSpec::Core::RakeTask.new(:unit) do |t|
t.pattern = Dir['spec/*/**/*_spec.rb'].reject{ |f| f['/api/v1'] || f['/integration'] }
end
RSpec::Core::RakeTask.new(:api) do |t|
t.pattern = "spec/*/{api/v1}*/**/*_spec.rb"
end
RSpec::Core::RakeTask.new(:integration) do |t|
t.pattern = "spec/integration/**/*_spec.rb"
end
end
One way to do it is to tag your RSpec test cases like this:
it "should do some integration test", :integration => true do
# something
end
When you execute your test cases use this:
rspec . --tag integration
This will execute all the test cases with the tag :integration => true. For more refer to this page.
I had to configure my unit and feature tests as follows:
require 'rspec/rails'
namespace :spec do
RSpec::Core::RakeTask.new(:unit) do |t|
t.pattern = Dir['spec/*/**/*_spec.rb'].reject{ |f| f['/features'] }
end
RSpec::Core::RakeTask.new(:feature) do |t|
t.pattern = "spec/features/**/*_spec.rb"
end
end
Had to add require 'rspec/rails' and change Rspec to RSpec in the answer given by #KensoDev.
Notice at https://github.com/rspec/rspec-rails, they are telling you to place the gem under "group :development, :test" like so,
group :development, :test do
gem 'rspec-rails', '~> 2.0'
end
but if you place this only under :test group only,
group :test do
gem 'rspec-rails', '~> 2.0'
end
then you'll get the above error.
HTH
I suggest to use .rspec file to configure patterns instead of using rake because it's tricky to pass flags to RSpec when using rake.
You can read environment variables in your .rspec file:
<%= if ENV['TEST'] == 'integration' %>
--pattern spec/integration/**/*_spec.rb
<% else %>
<% ENV['TEST'] = 'unit' %>
--pattern spec/unit/**/*_spec.rb
<% end %>
Then you can run TEST=integration rspec to run integration tests or just rspec to run unit tests. The advantage of this approach is that you can still pass flags to it like:
TEST=integration rspec -t login -f doc

Disable cookies for a cucumber test

I would like to be able do something like the following:
#javascript #disallow_cookies
Scenario: Test Something that depends on cookies
How would I write the #disallow_cookies cucumber hook to achieve this?
I am using the following:
gem 'thin', '1.2.11'
gem 'cucumber-rails' , '1.0.4'
gem 'database_cleaner', '0.6.7'
gem 'capybara', '1.1.2'
gem 'capybara-firebug','0.0.10'
gem 'selenium-webdriver', '2.14.0'
Thanks!
This is an old question but here goes. Create a before hook for the tag #disallow_cookies.
Before('#disallow_cookies') do
profile = Selenium::WebDriver::Firefox::Profile.new
profile['network.cookie.cookieBehavior'] = 2 # disables all kind of cookies
Capybara::Selenium::Driver.new(app, :browser => :firefox, :profile => profile)
end
Assuming that you are creating the driver in the general before hook you should be fine. I haven't run this code but it is where I would start the process.

ruby resque without loading rails environment

I have a resque worker which works great but is just too slow. The main reason for this is I'm using activerecord and having to load the entire environment which takes at least 10-20 seconds just to load up (I don't keep a running worker at all times as I'm using Heroku and pay for the time the worker runs). I'm using a resque worker to grab & parse data from an external website and then dumping the data into my database.
My question is whether I should rewrite the method to not use Rails and instead use DataMapper? Or something else which would load faster than activerecord.
Or If I should extract the code (using ActiveRecord) which figures out what to do with the external data and move it out of the worker and back into the app?
Hope that makes sense.
I have the same problem.
you could setup your environment on the rake resque:setup rake task
I tried this. assuming my rake resque task is on lib/tasks/resque.rake
require "resque/tasks"
task "resque:setup" do
root_path = "#{File.dirname(__FILE__)}/../.."
db_config = YAML::load(File.open(File.join(root_path,'config','database.yml')))["development"]
ActiveRecord::Base.establish_connection(db_config)
require "#{root_path}/app/workers/photo_downloader.rb" #workers
#Dir.glob("#{root_path}/app/models/*").each { |r| puts r; require r } #require all model
require "#{root_path}/app/models/photo.rb" # require model individually
end
I haven't completely success beacuse I use the Paperclip gem which require rails environment
Rails’ bootstrap is really slow; it is intended to be kept running, until certain time for restart (to eliminate some memory leaks there most likely is, any software is not bug-free), and is not intended to be used as a site that is launched for one request and then shut down.
That kind of usage more resembles a script. If you need to launch it with browser, you can easily use something like Erubis to write the page and use ActiveRecord in the script (I think it was useable outside of rails) or similar abstraction layer. Myself, for small tasks, I just use Mysql2.
Use bundler to get active_record and other gem to you without rails application .
require 'rubygems'
require 'logger'
require 'active_record'
require 'bundler'
require "active_support"
require "spreadsheet"
require 'net/ping'
require 'net/http'
Bundler.setup
Bundler.require(:default) if defined?(Bundler)
$config_logger = Logger.new("./log/dev.log")
class Dbconnect
def initialize
#settings = YAML.load_file('./config/database.yml')["development"]
#adapter = #settings["adapter"] if #settings["adapter"]
#database = #settings["database"] if #settings["database"]
#pool = #settings["pool"] if #settings["pool"]
#timeout = #settings["timeout"] if #settings["timeout"]
end
def connect_to_db
ActiveRecord::Base.establish_connection(
:adapter => #adapter,
:database => #database,
:reconnect => #reconnect,
:pool => #pool,
:timeout => #timeout)
$config_logger.info "\n db Connected: to => #{#database} "
end
end
end
}
Example Gemfile :
source "http://rubygems.org"
gem 'mail'
gem "escape_utils"
gem 'json',:require => "json"
gem 'json_pure'
gem 'resque'
gem 'resque-scheduler'
gem 'redis-namespace'
gem 'resque-status'
gem 'rake'
gem 'em-udns'
gem 'sqlite3'
gem 'spreadsheet'
gem 'activerecord', '3.2.1', :require => "active_record"
gem 'net-scp', :require => 'net/scp'
gem 'net-sftp', :require => 'net/sftp'
gem 'net-ssh', :require => 'net/ssh'
gem 'dir'
gem 'amatch'
gem 'haml'
gem 'net-ping'
gem install bundler
rest of the thing : bundle install .

Rails 3 + FactoryGirl: NameError: uninitialized constant Factory

ruby-1.9.2-p180 :007 > Factory.define :user do |user|
ruby-1.9.2-p180 :008 > user.email "user#example.com"
ruby-1.9.2-p180 :009?> user.password "foobar"
ruby-1.9.2-p180 :010?> user.password_confirmation "foobar"
ruby-1.9.2-p180 :011?> end
NameError: uninitialized constant Factory
My Gemfile:
group :test do
gem "rspec-rails"
gem 'webrat', '0.7.1'
gem 'spork', '0.9.0.rc4'
gem 'factory_girl_rails'
end
Even tough it seems I have everything as it should, I keep getting that error. I also have factories.rb created.
Thanks
I suppose you try in console in development environment. But you add the Factory gem only in test environment.
If you want access to Factory_girl in development use in your Gemfile :
group :test, :development do
gem 'factory_girl_rails'
end
Or if you want test your Factory launch your console in test environment :
rails c test
We had a similar problem on our end, rake spec seemed to be randomly failing with the error uninitialized constant FactoryGirl. The error was random -> coming and going. We went back half a dozen git commits to try and resolve it. In the end it was a silly mistake.
The fundamental problem is RAILS_ENV is set to development. It needs to be set to test when you run rake spec.
Address by:
Making certain that we are running rake spec in the RAILS_ENV test environment and its exported/sourced properly. To never confuse our environemnts, we modified zsh $RPROMPT env variable to show the current env.
export RPROMPT="[%{$fg_no_bold[yellow]%}$RAILS_ENV%{$reset_color%}]"
Require FactoryGirl in the spec ruby files gave a much better error message. At least rspec would run vs just fail outright this way when the environment was wrong. We also updated our gemfile to make sure factory_girl_rails and factory_girl were loaded both for development and testing.
Now we just run rpsec using the gem guard in a dedicated terminal with the proper RAILS_ENV set.
Its one of those gotchas.
We're running Ruby 1.9.3, Rails 3.2.9, Rspec 2.12, factory_girl 4.1.
I also ran into this while I was doing the Hartl tutorial.
If you are using "Spork" to speed up your tests and it is running in the background when you add the FactoryGirl code, you will need to stop it and restart it.
You should also change Factory.define to FactoryGirl.define, like in this example
FactoryGirl.define do
factory :user do
name 'John Doe'
date_of_birth { 21.years.ago }
end
end
from the factory_girl documentation

How do I properly arrange my Gemfile for Rails 3?

I am trying to install vote-fu into my rails 3 project.
The documentation ( for rails 2 ) says to install it into my environments.rb file as so..
config.gem "peteonrails-vote_fu", :lib => 'vote_fu', :source => 'http://gems.github.com'
How could I convert that to rails 3 for the Gemfile?
In Gemfile:
gem 'vote_fu'
http://gems.github.com is no more. The standard gem host is http://gemcutter.org which Bundler is setup to use by default, so you don't need to specify source. Vote_fu is hosted on gemcutter (see: http://rubygems.org/gems/vote_fu).
For reference, you would convert the options in config.gem like this:
:source => 'example.com' changes to (on its own line in Gemfile):
source 'example.com'
:lib => 'mylib' changes to:
gem 'libkey_mylib', :require => 'mylib'