Ruby HTML Diff

gistfile1.rb

require 'hpricot'
require 'diff/lcs'

class HTMLDiff
  def self.diff a, b
    self.new(a, b).compare
  end
  
  def initialize a, b
    @ah = Hpricot.parse a
    @bh = Hpricot.parse b
    
    @diff_source = ""
  end
  
  def compare
    @diff_source = compare_children(children(@ah), children(@bh))
  end
  
  def compare_children ac, bc
    src = ''
    Diff::LCS.sdiff(ac, bc).each do |diff|
      case diff.action
      when '-'
        src << "<del class=\"diff\">#{diff.old_element.to_s}</del>"
      when '+'
        src << "<ins class=\"diff\">#{diff.new_element.to_s}</ins>"
      when '!'
        kidcomp = []
        [diff.old_element, diff.new_element].each do |orig|
          if orig.is_a? Tag
            kidcomp << children(orig.elem)
          else
            kidcomp << orig.split(/\b/)
          end
        end
        if kidcomp.all? {|c| c.length == 1 and c.first.is_a? String }
          src << "<del class=\"diff modified\">#{diff.old_element.to_s}</del>"
          src << "<ins class=\"diff modified\">#{diff.new_element.to_s}</ins>"
        else
          src << compare_children(*kidcomp)
        end
      else
        src << diff.new_element.to_s
      end
    end
    src
  end
  
  def children h
    kids = []
    h.children.each do |child|
      case child
      when Hpricot::Text
        kids.push *child.to_s.split(/\b/)
      when Hpricot::Elem
        kids << Tag.new(child)
      else
        raise 'unexpected element type'
      end
    end
    kids
  end
  
  class Tag
    attr_accessor :elem, :name, :attrs, :contents
    def initialize hp
      @elem = hp
      @name = hp.name
      @attrs = hp.attributes
      @contents = hp.inner_text
    end
    
    def to_s
      e = Hpricot::Elem.new @name
      @attrs.to_hash.each {|k, v| e.set_attribute k, v }
      e.inner_html = @contents
      e.to_s
    end
    
    def == b
      (@name == b.name) and (@contents == b.contents)
    end
  end
end