#!/home/reeses/bin/ruby

# Requires ruby-htmltools 
# (http://www.bike-nomad.com/ruby/ruby-htmltools-1.01.tar.gz) 
# and all that implies.
# (c) Art Taylor, 2002, etc. etc.  There's so much new here that the
# copyright claim is so very, very necessary. ;>

require 'html/tree'
require 'net/http'

module TreeElement
  def ancestor(n) 
    if n == 1 
      parent
    else
      parent.ancestor(n - 1)
    end
  end

  def to_html()
    if data?
      ret = "#{self}\n"
    else
      ret = "<#{tag}"
      self.attributes.each { |attr,v| 
        if nil != v
          s = v.to_s
          if attr.to_s == "src" and s =~ /^.images/i
            s.gsub!(/^"/, "\"http\://www.compgeeks.com/")
          end
          s = "\"100%\"" if tag == "table" and attr.to_s == "width"
          ret << " #{attr.to_s}=#{s}" unless attr.to_s =~ /color/i
        else
          ret << " #{attr.to_s}"
        end
      }
      ret << ">\n"
      self.children.each { |child| ret << child.to_html }
      ret << "</#{tag}>\n"
    end
    ret
  end
end

def parseUrl(host, path)
  parser = HTMLTreeParser.new()
  Net::HTTP.new(host).get(path) { |str| parser.feed(str) }
  parser
end

class Special
  def getSpecial
    parser = parseUrl(@host, @path)
    parser.tree.each { |e|
      if e.to_s =~ @regexp
        print e.ancestor(@depth).to_html
      end
    }
  end
end

class CompGeeksSpecial < Special
  def initialize
    @regexp = /SPECIAL/
    @depth = 9
    @host = "www.compgeeks.com"
    @path = "/"
  end
end

class BuyDotComSpecial < Special
  def initialize
    @regexp = /mistake/
    @depth = 7
    @host = "www.buy.com"
    @path = "/default.asp"
  end
end

CompGeeksSpecial.new.getSpecial
BuyDotComSpecial.new.getSpecial