require 'capybara'
require 'capybara/poltergeist'

module Spieker
  class LinkScraper
    LOCAL_LINK_REGEX = /^(?!(http(s)?\:|\/\/)|data\:).*/

    include Capybara::DSL
    attr_writer :links

    def initialize(url)
      @url = URI.parse(url)
      Capybara.app_host = app_host

      Capybara.register_driver :poltergeist do |app|
          Capybara::Poltergeist::Driver.new(app, phantomjs_logger: NullStream.new )
      end

      Capybara.current_driver = :selenium
    end

    def result
      cleaned_up_links(found_links)
    end

    def app_host
      "#{@url.scheme}://#{@url.hostname}"
    end

    private

    def found_links
      @links ||= drive_page_for_links 
    end

    def drive_page_for_links
      begin
        visit @url.path
        page.all('a').map { |el| el[:href]}
      rescue
        puts "Error parsing #{@url.to_s}"
        []
      end
    end

    def cleaned_up_links(links)
      links.select { |link|
        is_local?(link) && !is_email?(link)
      }.map(&method(:filter_hash)).compact.uniq
    end

    def is_local?(link)
      link =~ LOCAL_LINK_REGEX ||
        begin
          URI.parse(link).hostname == @url.hostname
      rescue
        false
      end
    end

    def filter_hash(link)
      if match = link.match(/(.*)#(.*)$/)
        match[1]
      else
        link
      end
    end

    def is_email? link
      link =~ /mailto/
    end
  end
end

class NullStream
  def puts; end
end