absolute.rb

require 'uri'

module Julik
  module AbsolutizeHelper
    def canonicalize_hyperlinks(htmlcontent, root, host = nil, prot = nil)
      htmlcontent.gsub(/(<(img|a)\b[^>]*(src|href)=)(["'])(.*?)\4/) do
        md = $~

        prefix = (host && prot) ? (prot + host) : ''
        prefix.gsub!(/\/$/, '')
        begin
          url = URI.parse CGI.unescapeHTML(md[5])
          prefix = '' if (md[5].include?(prefix) && prefix.length > 0)
          if url.relative? && url.path !~ /^\//
            absolutized = File.expand_path(File.join(root, md[5], md[4]))
            # Clean the trailing slash but leave the quote
            absolutized.gsub!(/(\/)(["'])$/) { $2 }
            md[1] + md[4] + prefix + absolutized
          else
            md[1] + md[4] + prefix + md[5] + md[4]
          end
        rescue URI::InvalidURIError
          md[0]
        end
      end
    end
    
    def canonicalize_hyperlinks_in_context(html)
      host =  request.host
      proto = request.protocol
      # honor substandard port
      host += ":#{request.port}" unless request.port == '80'
      canonicalize_hyperlinks(html, '/', request.host, request.protocol)
    end
    
    def markdown_with_canonic_links(text)
      canonicalize_hyperlinks_in_context(markdown(text))
    end
    
    def textilize_with_canonic_links(text)
      canonicalize_hyperlinks_in_context(textilize(text))
    end
  end
end


require 'rubygems'
require 'action_pack'
require 'action_controller'
require 'action_controller/test_process'
require File.dirname(__FILE__) + '/../init'

# Re-raise errors caught by the controller.
class SomeController < ActionController::Base

  def absolute_action_with_markdown
    render :inline => "<%= markdown_with_canonic_links('And [a link](../some/stuff)') %>"
  end

  def absolute_action_with_textile
    render :inline => "<%= textilize_with_canonic_links('And \"a link\":../some/stuff') %>"
  end

  def rescue_action(e)
    raise e
  end
end

ActionController::Routing::Routes.draw do |map|
  map.connect ':controller/:action/:id'
end

class AbsolutizerTest < Test::Unit::TestCase
  include Julik::AbsolutizeHelper
  
  def setup
    @request = ActionController::TestRequest.new
    @response = ActionController::TestResponse.new
    @controller = SomeController.new
  end

  def test_basisc
    stuff = "Here is <a href='http://site.net/link'>stuffed</a> and \
    <a href='/some/interesting url to check out'>some ad uri</a> and \
    <a href='/some_link'>link</a> and <a href='../things'>another</a>"

    result =  "Here is <a href='http://site.net/link'>stuffed</a> and \
    <a href='/some/interesting url to check out'>some ad uri</a> and \
    <a href='http://site.net/some_link'>link</a> and <a href='http://site.net/stuff/bla/things'>another</a>"

    assert_equal result, canonicalize_hyperlinks(stuff, '/stuff/bla/badaboum', 'site.net', 'http://')
    assert_equal result, canonicalize_hyperlinks(stuff, '/stuff/bla/badaboum/', 'site.net', 'http://')
    
    stuff = 'And <a href="/this/link">is a root link</a> as we know'
    result_without_host = 'And <a href="/this/link">is a root link</a> as we know'
    result_with_host = 'And <a href="https://host.com:3030/this/link">is a root link</a> as we know'

    assert_equal result_without_host, canonicalize_hyperlinks(stuff, '/')    
    assert_equal result_with_host, canonicalize_hyperlinks(stuff, '/', 'host.com:3030', 'https://')

    expectation = '<p>And <a href="http://test.host/some/stuff">a link</a></p>'

    get :absolute_action_with_markdown
    assert_equal expectation, @response.body

    get :absolute_action_with_textile
    assert_equal expectation, @response.body
  end
end