Converts a number into a UTF16 surrogate pair.

surrogate.rb

#!/usr/bin/env ruby

require 'fileutils'
require 'debugger'

class Fixnum
  def to_surrogate_pair
    if self >= 0x10000 && self <= 0x10FFFF
      high = ((self - 0x10000) / 0x400).floor + 0xD800
      low = ((self - 0x10000) % 0x400) + 0xDC00
    end
    '\U' + [high, low].map { |x| x.to_s(16) }.join('\U').upcase
  end
end

class String
  def to_hex
    self.gsub('\U000', '0x').to_i(16)
  end
end

file = File.open(ARGV[0], 'rb') { |f| f.read }

begin
  encoded = file.encode('UTF-8')
rescue Encoding::UndefinedConversionError => e
  file = File.open(ARGV[0], 'rb:UTF-16LE') { |f| f.read } 
  encoded = file.encode('UTF-8')
end

regex = %r{(\\U\w{8})}i
fixed = 0
revised = encoded.gsub(regex) do |mbc|
  fixed += 1
  mbc.to_hex.to_surrogate_pair
end

FileUtils.mv(ARGV[0], "#{ARGV[0]}.old")
File.open(ARGV[0], 'w:UTF-16LE') { |f| f.write revised }
puts "File written with #{fixed} corrections."