Question

I have a many-to-many association between a Post and a Tag model.
Then, I have the following in my post view:

posts/show.html.erb:

<p><%= raw @post.tags.map { |t| link_to t.name, tag_path(t.name) }.join(', ') %></p>

This is the spec I wrote:

post_pages_spec.rb:

  describe "show page" do
    let!(:post) { FactoryGirl.create(:post, user:        user,
                                            title:       "Lorem",
                                            content:     "Lorem ipsum",
                                            category_id: category.id,
                                            tag_list:     "#{tag.id}") }

    before do 
      sign_in user
      visit post_path(post)
    end

    it { should have_selector('h1',       text: post.title) }
    it { should have_selector('title',    text: post.title) }
    it { should have_link(post.user.name, href: user_path(user)) }
    it { should have_selector('p',        text: post.content) }
    it { should have_selector('p',        text: post.category.name) }
    it { find("p").should have_content(post.tags.map { |t| t.name }.join(", ")) }
    .
    .
    .

I’m getting this error:

Failures:

1) Post pages show page
Failure/Error: it { find(“p”).should have_content(post.tags.map { |t| t.name }.join(“, “)) }
expected there to be content “1, tag28” in “Lorem ipsum”
# ./spec/requests/post_pages_spec.rb:28:in `block (3 levels) in ‘

The code works in the real site, but as you can see, the spec is failing. What’s the right way of writing this spec?

EDIT:

An example of the post model (just in case):

class Post < ActiveRecord::Base
  include ActionView::Helpers

  attr_accessible :title, :content, :category_id, :tag_list

  has_many :taggings, :dependent => :destroy  
  has_many :tags, :through => :taggings
  .
  .
  .

Output from live site:

  <p><a href="/tags/inkscape">inkscape</a>, <a href="/tags/gimp">gimp</a></p>

Answer 1

Update

From the comment thread, we’ve figured out that replacing the find with a simple have_selector passes:

it { should have_selector('p', text: post.tags.map { |t| t.name }.join(", ")) }

It’s not clear why find('p').should have_content(...) does not work. I’ve tested the same type of view (with <p> tags wrapping a list of links) and found that the find('p').should ... pattern works fine, so something funny is happening in your code. It may or may not be related to the issue with have_content in Capybara 2.0, which I discuss below.

If anyone has any ideas please share them! This is the best I can offer.

Original answer

The new text matcher in Capybara 2.0 excludes content that is not visible, producing somewhat unintuitive results:

it { should have_selector('title', text: 'Some Title') } # <= fails
it { should have_selector('title') }                     # <= passes
it { should have_text('Some Title') }                    # <= passes

And, relevant to the case in this question, this:

find("title").should have_content("some text") # <= fails

See also this post: How can I test the page title with Capybara 2.0?

If page.body.should ... is working but page.should ... (or in another form, subject { page} ; it { should ... }) isn’t, then perhaps this is the issue. Note that page.body is an HTML string, whereas page is a Capybara::Session, they’re totally different, although it seems in Capybara 2.0 at least you can run your expectations against either. I think page.body includes all HTML, hence bypassing the visibility issue.