404

#!/usr/bin/env ruby

require 'rubygems'
require 'anemone'
require 'optparse'
require 'ostruct'

Anemone.crawl(ARGV[0], {:discard_page_bodies => true}) do |anemone|

  anemone.after_crawl do |pages|

    # print a list of 404's
    not_found = []
    pages.each_value do |page|
      url = page.url.to_s
      not_found << url if page.not_found?
    end

    unless not_found.empty?
      puts "\n404's:"

      missing_links = pages.urls_linking_to(not_found)
      missing_links.each do |url, links|
        puts URI(url).path.to_s
        links.each do |u|
          u = u.path if options.relative
          puts "  linked from #{u}"
        end

      end

    end

  end

end