external links to tabs

action_view.rb

module ActionView
  module Helpers

    module TextHelper
      # Modify all html <a> tags in text to show them in a new browser's tab if they are externals to
      # the domain of the url string opts[:except_site_host]
      def external_links_to_tabs(text, opts = {})
        text.gsub(A_LINK_RE) do
          link, html_attrs = $&, $1
          if href_to_this_site?(link, opts[:except_site_host].to_s)
            link
          else
            new_html_attrs = [" target='_blank'", html_attrs.gsub(TARGET_RE, '')].join(' ')
            link.sub!(html_attrs, new_html_attrs)
          end
        end
      end


      private

        A_LINK_RE = /<a(\b.*?)>\w*<\/a>/i                 # Regexp to match an html <a...>...</a> tag
        TARGET_RE = /target(\b)*\=(\b)*['"](\b.*?)['"]/i  # Regexp to match an html target="something" tag's attribute

        # Wether text include href='protocol://anysubdomains.domain' where:
        #   protocol:// is http://, https:// or nothing
        #   domain      is the same main domain of the url string arg
        def href_to_this_site?(text, url)
          text =~ href_rexp_of_site(url)
        end

        # Returns a regexp to match the criteria above
        def href_rexp_of_site(str_url)
          href_prefix = "href(\b)*\=(\b)*"
          protocol    = "(\b)*(https?://)*"
          subdomain   = "((\\\w)+\.)*"
          quote       = "(\\\'|\\\")"
          domain      = extract_domain(str_url)
          Regexp.new [href_prefix, quote, protocol, subdomain, domain, quote].join
        end

        # Extracts the domain.first_level_domain (e.g: ideas4all.com) part of a url given as string (ex: www1.ideas4all.com).
        def extract_domain(str_url)
          str_url.split('.').last(2).join('.')
        end
    end

  end
end