# udump.py -- display the characters in a Unicode-encoded text file
# Jere Kapyaho (cone at iki dot fi) 2006-03-24

import sys
import codecs
import unicodedata

inputFile = codecs.open(sys.argv[1], "r", "utf-8")
fileData = inputFile.read()
inputFile.close()

print "Using Unicode %s data" % unicodedata.unidata_version
print "Read %d characters" % len(fileData)

offset = 0
for ch in fileData:
    utf8ch = ch.encode("utf-8")
    print "%08X: U+%06X %s" % (offset, ord(ch), unicodedata.name(ch, "(unnamed character)"))
    offset = offset + len(utf8ch)