#!/usr/bin/env python
# encoding: utf-8

combining_chars    = {}
decomposable_chars = {}

data = open("UnicodeData.txt")

for record in data:
    record = record.split(";")
    
    codepoint       = int(record[0], 16)
    combining_class = int(record[3])
    decomposition   = record[5]
    
    if combining_class:
        combining_class = int(combining_class)
        if combining_class != 0:
            combining_chars[codepoint] = combining_class
            continue
    
    if decomposition:
        decomposition = decomposition.split()
        # Only canonical decompositions
        if not decomposition[0].startswith("<"):
            decomposable_chars[codepoint] = [int(cp, 16) for cp in decomposition]

data.close()


# Prettyprinting as JS data structures

out = open("normalization_tables.js", "w")

combining_chars_out = (u'    "\\u%04X" : %i,\n' % (codepoint, combining_class) for 
                   codepoint, combining_class in combining_chars.iteritems())

def prettyprint_decomposition(list):
    return u'"%s"' % u''.join(u'\\u%04X' % item for item in list)

decomposable_chars_out = (u'    "\\u%04X" : %s,\n' % 
                      (codepoint, prettyprint_decomposition(decompostion)) for 
                      codepoint, decompostion in decomposable_chars.iteritems())

out.write(u"var Normalization = Normalization || {};\n\n")
out.write(u"Normalization.COMBINING_CHARS = {\n%s};\n\n" % "".join(combining_chars_out))
out.write(u"Normalization.DECOMPOSABLE_CHARS = {\n%s};\n\n" % "".join(decomposable_chars_out))