RSpec around filter for setting subdomain from example metadata

gistfile1.rb

RSpec is absolutely flexible tool. One of its multitarget features is metadata. The main purpose of this feature is filtering and groupping examples by given attributes. You can found a lot of examples in article “Filtering examples in rspec-2”.

Recently I was faced with the need to test application behaviour depending on current subdomain. It was a group of examples so I need to use filters. Spec was similar to script below:

```ruby
feature "On custom subdomain", :subdomain => 'custom' do
  scenario "homepage should change colorscheme to blue and use custom logo" do
    ...
  end

  scenario "about us page should show custom contact info" do
    ...
  end
  ...
end
```

It also need to configure that filter in spec_helper.rb:

```ruby
RSpec.configure do |config|
  config.around(:each, :subdomain => 'custom') do |example|
    Capybara.default_host = "custom.example.com"
    Capybara.current_session.reset!
    example.run
    Capybara.default_host = nil
    Capybara.current_session.reset!
  end
end
```

But what if we need to take into account particular subdomain name wich could be changed for another group. Here is that litle trick: RSpec hook conditions accept not only plain values. To solve our problem we need to change only two lines:

```ruby
  config.around(:each, :subdomain => /[\w-]+/) do |example|
    Capybara.default_host = "#{example.metadata[:subdomain]}.example.com"
```

This code means that we looking for examples or example groups with key :subdomain and value matching /[\w-]+/. To see how it works let see RSpec::Core::Metadata#apply_conditions. It has special handling for Hash, Regexp, Proc and Fixnum. So you can even use :dup.to_proc as a filter condition.

Shamelessly copied from https://avsej.net/2011/rspec-metadata-trick/