complete links function

gistfile1.txt

def complete_links(opts = {}) 
      return @complete_links unless @complete_links.blank?
      link_list = if opts[:limit]
        links.slice(0..opts[:limit])
      else
        links
      end

      @complete_links = []

      index = []   

      return @complete_links unless link_list.is_a?(Nokogiri::XML::NodeSet) # to try to solve bug iwth bad encoding on a nodeset
     
      link_list.each do |link| 
        begin
          unless link.class.to_s.include?("Nokogiri::XML::Element") 
            log "TYPHOEUSPATCH: ERROR: NonXMLElement #{link.inspect}"
            next
          end
        rescue => e    
          puts "TYPHOEUS: ERROR: #{e.inspect} #{e.backtrace.first}" 
        end
        
        begin  
          next if link["href"].nil?      
          next if link["href"].length == 0    
          next if index.include?(link["href"]) 
          next if link["href"].downcase.scan(/^javascript\:/).length > 0 
          next if link["href"].downcase.scan(/^mailto\:/).length > 0
        rescue => e
          puts "TYPHOEUSPATCH: ERROR: #{link["href"]} #{e.inspect} #{e.backtrace.first}"   
          next
        end

        link_info = {}

        begin
          begin  
            # gsubbing the link hrefs/urls to get rid of escaped characters that impact our ability to do our job
            parameter_cleaned_href = link["href"].gsub("%25", "%").gsub("%3D", "=").gsub("%26", "&").gsub("%3A", ":").gsub("%2F", "/").gsub("%3F", "?").gsub("%3D", "=").gsub("%25", "%").gsub("%3B", ";") unless link["href"].nil?
            parameter_cleaned_href = parameter_cleaned_href.gsub("%3A", ":").gsub("%2F", "/") unless parameter_cleaned_href.nil?
            x = Addressable::URI.parse(url).join(parameter_cleaned_href)
          rescue => e
            puts "TYPHOEUSPATCH: Error: ParseError: #{link["href"]}"
            next
          end
               
          x.path = x.path.squeeze('/')
          x.query = nil if x.query && x.query.empty?
          x.fragment = nil
          x.path = "/" if x.path.nil?
          #          x = PostRank::URI.c18n(PostRank::URI.unescape(x.to_s))

          link_info['href'] = link['href']
          link_info['ref'] = link['ref']
          link_info['rel'] = link['rel']
          link_info['title'] = link['title']
          link_info['alt'] = link['alt']
          imgz = link.css('img')
          img_array = []
          imgz.each do |i|
            img_array << {:src => i["src"], :alt => i["alt"], :title => i["title"]}
          end

          link_info['images'] = img_array
          link_info['text'] = link.text.to_s
          link_info['url'] = x.to_s
          link_info['absolute_url'] = x.to_s

          info = link_info_for_pagelink(link_info).stringify_keys

          @complete_links << info
          
          index << link["href"]     
        rescue => e
          puts "TYPHOEUSPATCH: ERROR: ParseError: #{link["href"]}"
          next
        end

      end

      @complete_links
    end