hadoop_xml_diff.rb

#!/usr/bin/env ruby
require 'rexml/document'
require 'set'

Property = Struct.new(:name, :value, :description, :final)

$PROP_NAMES = SortedSet.new

def read_document(path)
  props = REXML::XPath.match(REXML::Document.new(File.open(path){|f| f.read}), "/configuration/property").map {|prop|
    Property.new(
      prop.elements["name"].text,
      prop.elements["value"].text,
      prop.elements["description"] ? prop.elements["description"].text : nil,
      prop.elements["final"] ? prop.elements["final"].text : nil
    )
  }
  props_map = {}
  props.each do |p|
    $PROP_NAMES << p.name
    props_map[p.name] = p
  end
  props_map
end

def printlines(prefix, str)
  str.each_line do |line|
    puts "#{prefix}#{line.chomp}"
  end
end

def element(p, element_sym)
  if p.send(element_sym).nil?
    (" " * 4) + "<#{element_sym}/>"
  else
    (" " * 4) + "<#{element_sym}>#{p.send(element_sym)}</#{element_sym}>"
  end
end

def write_property(p)
  str = ""
  str << "  <property>\n"
  str << element(p, :name) + "\n"
  str << element(p, :value) + "\n"
  str << element(p, :description) + "\n" unless p.description.nil?
  str << element(p, :final) + "\n" unless p.final.nil?
  str << "  </property>\n\n"
  str
end

def description_equal(desc1, desc2)
  if desc1.nil? || desc2.nil?
    return desc1 == desc2
  end
  desc1 == desc2 || desc1.lines.map(&:strip).join('') == desc2.lines.map(&:strip).join('')
end

def write_property_diff(p1, p2)
  printlines(" ", "  <property>")
  printlines(" ", element(p1, :name))
  if p1.value == p2.value
    printlines(" ", element(p1, :value))
  else
    printlines("-", element(p1, :value))
    printlines("+", element(p2, :value))
  end

  if p1.final != p2.final
    printlines("-", element(p1, :final))
    printlines("+", element(p2, :final))
  end

  unless description_equal(p1.description, p2.description)
    printlines("-", element(p1, :description))
    printlines("+", element(p2, :description))
  end
  printlines(" ", "  </property>")
  puts " "
end

def diff(p1, p2)
  if p1.nil? && p2.nil?
    raise "BUG: both are nil"
  elsif p1.nil?
    printlines("+", write_property(p2))
  elsif p2.nil?
    printlines("-", write_property(p1))
  elsif p1.name == p2.name && p1.value == p2.value && description_equal(p1.description, p2.description) && p1.final == p2.final
    return
  else
    raise "BUG: property name mismatch, #{p1.name}, #{p2.name}" unless p1.name == p2.name
    write_property_diff(p1, p2)
  end
end

doc1 = read_document(ARGV[0])
doc2 = read_document(ARGV[1])

puts "--- #{File.basename(ARGV[0])}"
puts "+++ #{File.basename(ARGV[1])}"

puts "<configuration>"
puts ""

$PROP_NAMES.each do |name|
  diff doc1[name], doc2[name]
end

puts "</configuration>"