diff yaml trees semantically

ydiff

#!/usr/bin/env ruby

# diffs the contents of yaml files semantically. that is, ydiff will parse each
# yaml file and then diff their resulting trees, instead of attempting to diff
# the text. it's specifically for comparing directories of translations from
# Crowdin, so it's effectively only concerned with strings.

require 'pathname'
require 'rubygems'
require 'yaml'
require 'pp'

def diff_dir(left_path, right_path)
  left_path.each_child do |left_child|
    if left_child.directory?
      right_candidate = right_path.join(left_child)
      next unless right_candidate.exist?
      diff_dir(left_child, right_candidate)
    else
      right_candidate = right_path.join(left_child)
      next unless right_candidate.exist?
      diff_file(left_child, right_candidate)
    end
  end
end

def diff_file(left_path, right_path)
  left_tree = YAML.load_file(left_path)
  right_tree = YAML.load_file(right_path)
  puts "# ================================================================================"
  puts "# < #{left_path}"
  puts "# > #{right_path}"
  puts "# ================================================================================"
  puts ""
  diff_tree(left_tree, right_tree)
end

def diff_tree(left, right, path = '')
  case left
  when Hash
    diff_hash(left, right, path)
  when String
    diff_string(left, right, path)
  when Array
    diff_array(left, right, path)
  when Symbol
    diff_symbol(left, right, path)
  else
    STDERR.puts "SKIP #{path}: left type of #{left.class} is unhandled"
  end
end

def added(item, path)
  case item
  when Hash
    item.each_key do |key|
      added(item[key], "#{path}/#{key}")
    end
  when String
    puts "+ #{path}: #{item}"
  when Array
    item.each_with_index do |v, i|
      added(v, "#{path}[#{i}]")
    end
  when Symbol
    puts "+ #{path}: #{item}"
  else
    STDERR.puts "SKIP ADDED #{path}: dunno how to report #{item.class}: #{item}"
  end
end

def diff_symbol(left, right, path = '')
  if left != right
    puts "- #{path} #{left}"
    puts "+ #{path} #{right}"
  end
end

def diff_array(left, right, path = '')
  left.each_with_index do |v, i|
    diff_tree(v, right[i], "#{path}[#{i}]")
  end
end

def diff_string(left, right, path = '')
  if left != right
    puts "- #{path} #{left}"
    puts "+ #{path} #{right}"
  end
end

def diff_hash(left, right, path = '')
  left.each_key do |key|
    if right[key]
      diff_tree(left[key], right[key], "#{path}/#{key}")
    else
      puts "- #{path}/#{key}"
    end
  end

  right.each_key do |key|
    unless left[key]
      added(right[key], "#{path}/#{key}")
    end
  end
end

left_path = ARGV[0]
right_path = ARGV[1]

if File.directory? left_path
  diff_dir(Pathname.new(left_path), Pathname.new(right_path))
else
  diff_file(left_path, right_path)
end