Linux/Ruby/Rails tips @def-end - Tumblr Blog

How to control Unicorn’s middleware stack

Say your client app consumes only not chunked content (”Transfer-Encoding: chunked” header should not be present) and it requires Content-Length header strictly. How could you achieve that?

Firstly you should set RACK_ENV variable to ‘none’ whether in your deploy.rb file (if you use Capistrano for running Unicorn on staging/production) or just run unicorn_rails prepending with RACK_ENV=none. Here’s the line for the Capistrano’s deploy.rb

set :rack_env, :none

By specifying that Unicorn’s Rack::Builder will use only your app’s middlewares and wouldn’t add Rack::Chunked and Rack::ContentLength ( please see https://github.com/defunkt/unicorn/blob/master/lib/unicorn.rb#L60)

Ok, so how to add Content-Length header to server’s response then?

That’s pretty simple: just add

config.middleware.use Rack::ContentLength

to config/environments/environment_file.rb

That’s it! Thanks!

#unicorn #rails #tip #middleware

RSpec doesn't load helpers for view specs

If your view spec breaks with the following message:

Failure/Error: render

ActionView::Template::Error:

undefined method `tour_pack_url' for #<#<Class:0x00000002a13c70>:0x0000000280ff78>

Then chances are you hit the already known issue https://github.com/rspec/rspec-rails/issues/476

To fix it just monkeypatch ActionView::TestCase::TestController class like this (you could do it in spec/spec_helper.rb for example):

ActionView::TestCase::TestController.instance_eval do

helper NeededHelper

end

That's it. Now all tests should pass without failing.

Affected RSpec versions: 2.8.x - 2.11.x

#tip #BDD #rspec

ruby-debug in ruby 1.9.3

If you work with ruby 1.9 and want to debug your application usual 'require "ruby-debug"' just won't work. It'll keep throwing weird errors and you'll just spend a couple hours trying to figure out what's wrong.

To start using ruby-debug within your application or library you need to add these lines to your Gemfile:

gem 'linecache19',:git => 'git://github.com/mark-moseley/linecache' gem 'ruby-debug-base19x', '~> 0.11.30.pre10' gem 'ruby-debug19'

Then you could invoke your scripts like this (after running 'bundle install' of course):

bundle exec ruby script.rb

# or

bundle exec rspec script.rb

#ruby19 #tip

the very minimal mruby program

Recently I tried to understand how mruby works. So I tried to run minimal embedded program in mruby. Despite the high activity on that theme I found only a few examples of working programs.

So here's my first mruby "Hello, World" program:)

In order to compile it you should run (in mruby's folder):

gcc -o test_compile test_compile.c -Iinclude -Isrc lib/libmruby.a -lm

[mruby] ./test_compile

Hello, World!

Resources:

http://www.arangodb.org/2012/07/02/mruby-links

#ruby #mruby #tip

ActiveRecord::Base methods chain introspection

Recently I have an issues with some derived from ActiveRecord::Base class. I couldn't get how the request was made through the chain of the method calls so I made a little hack to show the name of the method once it gets invoked.

So here's the code:

https://gist.github.com/2828416

You could either include it in some kind of init.rb/environment.rb file to load it every time you run your project or you could enter it manually in the irb session.

Here's the my example of usage (inside of irb): https://gist.github.com/2828602

NB: You could apply this technique not just to ActiveRecord::Base but also to all modules/classes you want (e.g. complicated classes derived from ActiveRecord::Base).

#ruby #activerecord #tip

how to test links with :delete method in Cucumber/Capybara

By default for :delete method in #link_to helper function:

<%= link_to "Delete User", admin_user_path(@user.id), :method => :delete, :class => "btn danger" %>

Rails generates following HTML:

<a href="/admin/users/3" class="btn danger" data-method="delete" rel="nofollow">Delete User</a>

All's good but when you'll start testing this code with Cucumber/Capybara you may notice errors, i.e. sending DELETE request as regular GET (as it's written in href attribute).

So this step will not work:

Then /^I click on "Delete User" link for "([^"]*)" user$/ do |user_name| click_link user_name end

To achieve desired behavior you need to implicitly invoke #delete method, so the previous step should be rewritten as follows:

Then /^I click on "Delete User" link for "([^"]*)" user$/ do |user_name| user = User.find_by_name(user_name) user.should be Capybara.current_session.driver.delete admin_user_path(user.id) end

P.S.: You may also want to check whether the link has data-method attribute set to 'delete' before sending delete request.

#cucumber #BDD #tip #ruby #capybara

#click_link not working in Cucumber with Capybara

The reason why the simple step definition with only one command could throw an error:

And I click on the "Add Client" link # features/step_definitions/create_client_by_admin_steps.rb:26

You have a nil object when you didn't expect it!

You might have expected an instance of Array.

The error occurred while evaluating nil.[] (NoMethodError)

(eval):2:in `click_link'

./features/step_definitions/create_client_by_admin_steps.rb:31:in `/^I click on the "([^"]*)" link$/'

features/create_client_by_admin.feature:10:in `And I click on the "Add Client" link'

is that you didn't specified "href" attribute in <a> tag or specified it wrong:

<a href="javascript:void(0)" class="btn danger" data-controls-modal="add_client_modal_dialog">Add Client</a>

while the right way to specify "href" attribute is this:

<a href="#" class="btn danger" data-controls-modal="add_client_modal_dialog">Add Client</a>

#tip #BDD #cucumber #capybara

how to to test you code against Authorize.net

If you want to test your integration with Authorize.net you could do it with a Visa test credit card number. The payment gateway has been designed so that this special test credit card number can be used to generate specific response codes or errors from Authorize.Net so that you may test your working code.

To cause the payment gateway to respond with a specific response code, send a transaction with the Visa error test credit card number (x_card_num) 4222222222222 and an amount (x_amount) equal to the response code you want the payment gateway to return. For example, if a transaction is sent in Test Mode with the credit card number 4222222222222 and an amount of 12 dollars, the payment gateway will respond with response code 12, "Authorization Code is required but is not present."

This could be very helpful in unit/functional tests (not only in Rails)

Source: Authorize.net documentation

#authorize.net #tip #TDD

god config for delayed_job

I've been using delayed_job for a while now and I must say it's pretty unstable, plus it eats a lot of memory (especially in ruby 1.8). In order to add monitoring of dj daemon (e.g. restart it if it's memory use is greater then some threshold) I tried monit and god utilities.

Monit (for whatever reason) could not execute inline bash script in "start" clause to enter application folder (which should be current folder to properly load rails environment and maybe some files from ./lib folder). So the only tool that works for me is god with config file as shown below.

Here's simple god config for one delayed_job (could be easily scaled to multiple processes):

# run with: god -c /path/to/config.god [add -D if you want to not-deamonize god] # This is the actual config file used to keep the delayed_job running APPLICATION_ROOT = "/var/www/application" RAILS_ENV = "production" God.watch do |w| w.name = "delayed_job_production" w.interval = 15.seconds w.start = "/bin/bash -c 'cd #{APPLICATION_ROOT}/current; /usr/bin/env RAILS_ENV=#{RAILS_ENV} #{APPLICATION_ROOT}/current/script/delayed_job start > /tmp/delay_job.out'" w.stop = "/bin/bash -c 'cd #{APPLICATION_ROOT}/current; /usr/bin/env RAILS_ENV=#{RAILS_ENV} #{APPLICATION_ROOT}/current/script/delayed_job stop'" w.log = "#{APPLICATION_ROOT}/shared/log/god_delayed_job.log" w.start_grace = 30.seconds w.restart_grace = 30.seconds w.pid_file = "#{APPLICATION_ROOT}/shared/pids/delayed_job.pid" w.behavior(:clean_pid_file) w.start_if do |start| start.condition(:process_running) do |c| c.interval = 5.seconds c.running = false end end w.restart_if do |restart| restart.condition(:memory_usage) do |c| c.above = 300.megabytes c.times = [3, 5] # 3 out of 5 intervals end restart.condition(:cpu_usage) do |c| c.above = 50.percent c.times = 5 end end # lifecycle w.lifecycle do |on| on.condition(:flapping) do |c| c.to_state = [:start, :restart] c.times = 5 c.within = 5.minute c.transition = :unmonitored c.retry_in = 10.minutes c.retry_times = 5 c.retry_within = 2.hours end end end

The same code in gist - https://gist.github.com/1237902

The key line here is:

w.start = "/bin/bash -c 'cd #{APPLICATION_ROOT}/current; /usr/bin/env RAILS_ENV=#{RAILS_ENV} #{APPLICATION_ROOT}/current/script/delayed_job start > /tmp/delay_job.out'"

it starts delayed_job within correct environment (all as inline bash script to ensure 'start' is the one command) and with logging errors while starting to /tmp/delay_job.out.

You could store your config.god file in shared folder (if capistrano folder layout is used) and start god from command line like this:

god -c config.god

Then monitoring of all processes could be done with:

god status

The god gem lacks only convenient web-interface to check running processes (which monit have).

Resources:

God - a process monitoring framework in ruby DelayedJob

#god #delayed_job #gem #ruby #rails

redis_orm 0.6

I finally released redis_orm 0.6 gem. In this version:

added equality operator for object, #to_s method for inspecting objects, #find! which could throw RecordNotFound error

added self.descendants class method which returns all inherited from RedisOrm::Base classes

introduced :sortable option (in property declaration and #find conditions hash) - rudimentary ability to sort records by any property (not just by default 'created_at')

now handling models withing modules definitions (test for this in associations_test.rb)

properly handling :as parameter in options for :has_many/:belongs_to self-references

binding related models while creating model instance (like this: Article.create(:comment => comment))

bunch of small fixes, updated tests and README.md

Source code: https://github.com/german/redis_orm

Gem: http://rubygems.org/gems/redis_orm

#ruby #gem #redis #redis_orm

URI.parse error

If url you want to receive via HTTParty or Net::HTTP contains special characters (like '|' or vowels with umlauts) an exception will be thrown:

>> URI.parse "http://staticmaps.cloudmade.com/#{CLOUDMADE_API_KEY}/staticmap?size=600x500&styleid=4&bbox=51.5159,-0.0868016,51.5047,-0.0807667&marker=label:0|51.5159,-0.0807667&marker=51.5047,-0.0868016"

=> URI::InvalidURIError: bad URI(is not URI?):

To make URI.parse work you should invoke URI.encode on the url string first:

>> URI.parse(URI.encode("http://staticmaps.cloudmade.com/#{CLOUDMADE_API_KEY}/staticmap?size=600x500&styleid=4&bbox=51.5159,-0.0868016,51.5047,-0.0807667&marker=label:0|51.5159,-0.0807667&marker=51.5047,-0.0868016"))

=> #<URI::HTTP:0x7f0b095136d0 URL:http://staticmaps.cloudmade.com/#{CLOUDMADE_API_KEY}/staticmap?size=600x500&styleid=4&bbox=51.5159,-0.0868016,51.5047,-0.0807667&marker=label:0%7C51.5159,-0.0807667&marker=51.5047,-0.0868016>

#gem #tip

File attachment management with paperclip and redis

If you recently migrated from the well known SQL-land to the terra incognita of key-value storages (in particular to Redis) and are now wondering how the hell you'll be handling all those avatars and photos users keep uploading to the servers, this small article is for you.

Things are much easier in case you are already using paperclip. Here are 3 steps you should follow to manage your file attachments with redis and paperclip:

1. Add "redis_orm" gem to your Gemfile:

gem "redis_orm", "~> 0.5"

2. Inherit your model with attachments from RedisOrm::Base class like this:

class User < RedisOrm::Base

# here goes 4 specific to paperclip properties:

property :avatar_file_name, String property :avatar_content_type, String property :avatar_file_size, Integer property :avatar_updated_at, Time

include Paperclip::Glue # here's the most tricky part (:

has_attached_file :avatar, :styles => {:big => "900x900>", :medium => "200x200>", :chat => "42x42>"} end

3. my pull request to paperclip has been recently merged (w00t!), so before new gem is released you should pull it right from github:

gem "paperclip", :git => 'git://github.com/thoughtbot/paperclip.git'

And that's it! It just works.

All other code (e.g. on view and controller level) is specific to paperclip and could be found on their github page.

#Redis #paperclip #tip #redis_orm

redis_orm

I recently wrote an ORM for Redis key-value store. It tries to mimic ActiveRecord 2.x API.

It supports associations (belongs_to/has_many/has_one, self-referencing, polymorphic), ActiveModel's validations, create/destroy/change callbacks, indices (in order to find records but not to increase the access speed) and dynamic finders.

Here's an example of defining a model:

class User < RedisOrm::Base property :first_name, String property :last_name, String timestamps # OR # property :created_at, Time # property :modified_at, Time index :last_name index [:first_name, :last_name] has_many :photos has_one :profile after_create :create_mailboxes def create_mailboxes # ... end end

More documentation could be found on https://github.com/german/redis_orm Gem is available too https://rubygems.org/gems/redis_orm

#gem #ruby #redis

how to delete yourself from shared repo on github

If you want to delete yourself from the shared repo on github (you'll loose push access to this repo) you could do this on https://github.com/account/repositories page.

#tip #github

Vkontakte OAuth2 provider for OmniAuth

Vkontakte.ru is a major social network in Russia and CIS and they just recently added OAuth2 support. So I added Vkontakte as a standard provider to OmniAuth and Michael Bleigh has merged my pull request into a master (w00t!).

So if you already using OmniAuth and/or an old OpenAPI Vkontakte authentication please consider upgrading the gem to 0.2.3. After upgrading you'll need to add a single line with new provider to your config/initializers/omniauth.rb:

Rails.application.config.middleware.use OmniAuth::Builder do provider :vkontakte, VKONTAKTE_API_ID, VKONTAKTE_SECRET_KEY end

Now you could use it in your views like an ordinary link to provider:

<%= link_to t('sign_in_through.vkontakte'), "/auth/vkontakte" %>

I also encourage you to check an excellent screencast by Ryan Bates covering an OmniAuth authentication.

#ruby #gem #omniauth #oauth2 #authentication

2 mocking techniques for Net::HTTP

I assume that for both cases you're storing HTML files you want to mock in 'resources' folder which is relative to the file with tests.

1st method: if you're using Net::HTTP::GET/POST classes you'll need the 'mocha' gem. For example if you wrote something like this:

url = URI.parse("http://your.site.com") http = Net::HTTP.start(url.host, url.port) http.request(Net::HTTP::GET.new(url.path) # or something similair

to feed your html to the above code you could use following simple technique:

require 'mocha' class TestResponse < Struct.new(:head, :body, :response_code) end class OpenSession def request(url) body = File.read(File.join(File.dirname(File.expand_path(__FILE__)), 'resources', url.path)) TestResponse.new("head", body, 200) end

and then somwhere before tests (for example in def setup):

Net::HTTP.any_instance.stubs(:start).returns(OpenSession.new)

2nd technique: if you're using Net::HTTP.get_response class method you could override it and use specifically constructed MySocketStub class as a mock for TCPSocket:

class MySocketStub def initialize(body) @body = body end def closed? false end def read_all(from) @body end end def setup # some setup of your code (if any) Net::HTTP.instance_eval do def self.get_response(url) body = File.read(File.join(File.dirname(File.expand_path(__FILE__)), 'resources', url.path)) resp = Net::HTTPOK.new(Net::HTTP.version_1_2, '200', '') resp.reading_body(MySocketStub.new(body), true){} resp.body = body resp end end end

I'm now using second approach in my nethttp_ab gem although I'd tried a former one earlier.

Resources:

https://github.com/german/nethttp_ab/blob/master/test/nethttp_ab_test.rb

#ruby #TDD #mocks

nethttp_ab - simple tool to test and benchmark sites

Sometimes you need to quickly benchmark your site without installing apache2-utils package with the bunch of dependencies, that's why I wrote a ruby gem to benchmark sites - nethttp_ab.

It's very simple and mimics ab in some command-line options. For example to make 100 requests with 3 concurrent threads:

nethttp_ab -n100 -c3 http://www.yoursite.com

But there's also a difference: you could simulate a user who follows all local links on the page with:

nethttp_ab -f http://localhost:3000

-f stands for 'follow links'. To simulate 3 users (all local links on the page will be visited once) you could write:

nethttp_ab --follow_links -c3 http://localhost:3000

Resources:

Gem: https://rubygems.org/gems/nethttp_ab Source code: https://github.com/german/nethttp_ab Issue tracker: https://github.com/german/nethttp_ab/issues

#gem #ruby #benchmarking #utility

Trending Blogs

Recently Viewed Blogs

Linux/Ruby/Rails tips