gistfile1.rb

class Nokogiri::HTML::Document
  def meta_encoding
    content_type = css("meta[http-equiv=content-type]").each do |meta|
      break meta.attribute("content").value
    end
    return unless content_type

    content_type.split("; ").each do |part|
      next unless part =~ /^charset=(.*)/
      return $1
    end
    
    nil
  end
end

module Nokogiri::HTML
  def self.with_meta_encoding(data)
    doc = Nokogiri.HTML(data)
    
    meta_encoding = doc.meta_encoding
    return doc unless meta_encoding && doc.encoding != meta_encoding
    
    # try to reread with meta_encoding
    doc2 = Nokogiri.HTML(data, nil, meta_encoding)
    return doc2 if doc2.encoding == meta_encoding

    # rereading failed, return original document
    doc
  end
end