gistfile1.py

text = open('./no-network.txt').read().replace('\n', '')

utf9_array = []
i = 0
print len(text)
while i < len(text) :
  utf9 = (int(text[i])<<6) + (int(text[i+1])<<3) + int(text[i+2])
  utf9_array.append(utf9)
  i += 3

i = 0
print len(utf9_array)
ucs4_array = []
while i < len(utf9_array) :
  nonet = utf9_array[i]; i+=1
  ucs4 = nonet & 0xff
  while nonet & 0x100 :
      ucs4 <<= 8;
      nonet = utf9_array[i]; i+=1
      ucs4 |= nonet & 0xff
  ucs4_array.append(ucs4)

i = 0
byte_array = ''
print len(ucs4_array)
while i < len(ucs4_array) :
  byte_array += unichr(ucs4_array[i]).encode('utf-8')
  i += 1

print len(byte_array)
print byte_array