Test PDF within Cucumber and Capybara

Gemfile

# normal Gem dependancy declarations
# ...

group :test, :cucumber do
  gem 'pdf-reader'
end

pdf_steps.rb

When /^I follow the PDF link "([^"]+)"$/ do |label|
  click_link(label)
  temp_pdf = Tempfile.new('pdf')
  temp_pdf << page.source.force_encoding('UTF-8')
  temp_pdf.close
  pdf_text = PDF::PdfToText.new(temp_pdf.path)
  page.driver.response.instance_variable_set('@body', pdf_text.get_text)
end

pdf_test.feature

  Scenario: Check printable PDF
    When I follow the PDF link "Print"
    Then I should see "My name"
      And I should see "Something else"

pdf_to_text.rb

module PDF
  class PdfToText
    def initialize(pdf_file)
      @receiver = PDF::SimplePageTextReceiver.new
      pdf = PDF::Reader.file(pdf_file, @receiver)
    end

    def get_text
      @receiver.content.inspect
    end
  end
end

simple_page_text_receiver.rb

module PDF
  class SimplePageTextReceiver
    attr_accessor :content

    def initialize
      @content = []
    end

    # Called when page parsing starts
    def begin_page(arg = nil)
      @content << ""
    end

    # record text that is drawn on the page
    def show_text(string, *params)
      @content.last << string.strip
    end

    # there's a few text callbacks, so make sure we process them all
    alias :super_show_text :show_text
    alias :move_to_next_line_and_show_text :show_text
    alias :set_spacing_next_line_show_text :show_text

    # this final text callback takes slightly different arguments
    def show_text_with_positioning(*params)
      params = params.first
      params.each { |str| show_text(str) if str.kind_of?(String)}
    end
  end
end

Thank you for the page.driver.response.instance_variable_set bit, I was trying to figure that out for the last 2 hours. I was able to avoid using any temp files or extra modules by using pdf-inspector instead of pdf-reader:

require "pdf/inspector"

Then /^I should be served the referral document as a PDF$/ do
  page.response_headers['Content-Type'].should == "application/pdf"
  pdf = PDF::Inspector::Text.analyze(page.source).strings.join(" ")
  page.driver.response.instance_variable_set('@body', pdf)
end

Then /^I should see the referral document details$/ do
  page.should have_content("#{@document.customer.name}")
  page.should have_content("#{@document.resources.first.resource.name}")
  page.should have_content("Document opened at #{@document.created_at.strftime("%e-%b-%4Y %r")}")
end

If you happen to get this error:

undefined method `join' for #<String:...> (NoMethodError)

you can fix it by converting the text to an array like this:

page.driver.response.instance_variable_set('@body', [pdf_text.get_text])

Thanks for the inspiration. I used StringIO to grab the contents of my pdf response and turn it into an io stream for PDF::Reader

 And the pdf file should contain:
          | Subtotal      | $10.00     |
          | Shipping      | $20.00     |
          | IN Sales Tax  | $2.00      |
          | Total         | $32.00     |

Then(/^the pdf file should contain:$/) do |table|
  pdf_io = StringIO.new(page.source.force_encoding('UTF-8'))
  reader = PDF::Reader.new(pdf_io)
  contents = reader.pages.map(&:to_s).join("\n")

  table.rows.each do |cells|
    cells.each do |cell|
      expect(contents).to include cell
    end
  end
end