#!/usr/bin/env ruby

# pure Ruby version of Uconv.
#
module RUconv
  
  # convert UTF-16 string into UTF-8
  #
  def u16_to_u8(str)
    case str
    when /\A\xFF\xFE/
      u16le_to_u8(str[2,-1])
    when /\A\xFF\xFE/
      u16be_to_u8(str[2,-1])
    when /\A[\x01-\x7f]\x00[\x01-\x7f]\x00[\x01-\x7f]\x00/  ## incorrect guess
      u16le_to_u8(str)
    when /\A\x00[\x01-\x7f]\x00[\x01-\x7f]\x00[\x01-\x7f]/  ## incorrect guess
      u16be_to_u8(str)
    else
      raise ArgumentError, "cannot detect encoding"
    end
    module_function :u16_to_u8
  end

  # convert UTF-16BE string into UTF-8
  #
  def u16be_to_u8(str)
    buf = []
    str.scan(/((?:\x00[\x00-\x7f])+)|((?:[\x00-\x07][\x00-\xff])+)|((?:[\x08-\xd7][\x00-\xff])+)|((?:[\xd8-\xdb][\x00-\xff][\xdc-\xdf][\x00-\xff])+)|((?:[\xe0-\xff][\x00-\xff])+)|(.)/){
      |u1,u2,u3,u4,u5,err|
      if u1
	(u1.length/2).times{|i|
	  buf << u1[i*2+1]
	}
      elsif u2
	(u2.length/2).times{|i|
	  buf << (u2[i*2]<<8)+u2[i*2+1]
	}
      elsif u3
	(u3.length/2).times{|i|
	  buf << (u3[i*2]<<8)+u3[i*2+1]
	}
      elsif u4
	(u4.length/4).times{|i|
	  u_buf_f = (u4[0]<<8)+u4[1]
	  u_buf_b = (u4[2]<<8)+u4[3]
	  buf << ((u_buf_f-0xd800)<<10)+(u_buf_b-0xdc00)
	}
      elsif u5
	(u5.length/2).times{|i|
	  buf << (u5[i*2]<<8)+u5[i*2+1]
	}
      elsif err
	raise RangeError, 'illegal UTF-16 sequence'
      end
    }
    buf.pack("U*")
  end
  module_function :u16be_to_u8

  # convert UTF-16LE string into UTF-8
  #
  def u16le_to_u8(str)
    buf = Array.new(str.length/2)
    pos = 0
    str.scan(/((?:[\x00-\x7f]\x00)+)|((?:[\x00-\xff][\x00-\x07])+)|((?:[\x00-\xff][\x08-\xd7])+)|((?:[\x00-\xff][\xdc-\xdf][\x00-\xff][\xd8-\xdb])+)|((?:[\x00-\xff][\xe0-\xff])+)|(.)/){
      |u1,u2,u3,u4,u5,err|
      if u1
	(u1.length/2).times{|i|
	  buf[i+pos] = u1[i*2]
	}
	pos += (u1.length/2)
      elsif u2
	(u2.length/2).times{|i|
	  buf[i+pos] = (u2[i*2+1]<<8)+u2[i*2]
	}
	pos += (u2.length/2)
      elsif u3
	(u3.length/2).times{|i|
	  buf[i+pos] = (u3[i*2+1]<<8)+u3[i*2]
	}
	pos += (u3.length/2)
      elsif u4
	(u4.length/4).times{|i|
	  u_buf_f = (u4[3]<<8)+u4[2]
	  u_buf_b = (u4[1]<<8)+u4[0]
	  buf[i+pos] = ((u_buf_f-0xd800)<<10)+(u_buf_b-0xdc00)
	}
	pos += (u4.length/2)
      elsif u5
	(u5.length/2).times{|i|
	  buf[i+pos] = (u5[i*2+1]<<8)+u5[i*2]
	}
	pos += (u5.length/2)
      elsif err
	raise RangeError, 'illegal UTF-16 sequence'
      end
    }
    buf.pack("U*")
  end
  module_function :u16le_to_u8

  # Byte-swaps UTF-16 string.
  #
  def u16swap(str)
    str.unpack("n*").pack("v*")
  end
  module_function :u16swap


  # convert UCS-4BE string into UTF-8
  #
  def u4be_to_u8(str)
    str.unpack("N*").pack("U*")
  end
  module_function :u4be_to_u8

  # convert UCS-4LE string into UTF-8
  #
  def u4le_to_u8(str)
    str.unpack("V*").pack("U*")
  end
  module_function :u4le_to_u8

  # Byte-swaps UCS-4 string.
  #
  def u4swap(str)
    str.unpack("N*").pack("V*")
  end
  module_function :u4swap

  # convert UCS-4(2143 order) string into UTF-8
  #
  def u42143_to_u8(str)
    u4be_to_u8(u16swap(str))
  end
  module_function :u42143_to_u8

  # convert UCS-4(3412 order) string into UTF-8
  #
  def u43412_to_u8(str)
    u4le_to_u8(u16swap(str))
  end
  module_function :u43412_to_u8

  
  # delete BOM in UTF-8
  #
  def u8_delete_bom(str)
    return str[3..-1]
  end
  module_function :u8_delete_bom

  def iso88591_to_u8(str)
    str.unpack("C*").pack("U*")
  end
  module_function :iso88591_to_u8

end

if __FILE__ == $0

  include RUconv
#  print u16_to_u8(ARGF.read)
  str = ARGF.read
  10.times{
    s = u16le_to_u8(str)
  }
#  print u16le_to_u8(str)

end