# -*- coding: utf-8 -*-

import codecs, types

# from utf-8 to iso6937
def iso6937_encode(input,errors,encoding_map):
    result = ''
    num_bytes = 0
    for c in input:
	try:
	    if c in encoding_map:
		num_bytes += 2
		result += chr(ord(encoding_map[c][0].decode('unicode-escape'))) + encoding_map[c][1]
	    else:
		code = ord(c)
		if code > 0xFF:
		    num_bytes += 1
		    result += chr(code >> 8)
		num_bytes += 1
		result += chr(code & 0xFF)
	except KeyError:
	    raise UnicodeError, "internal conversion algoritm error"
    return (result, num_bytes)


# from iso6937 to utf-8
def iso6937_decode(input,errors,decoding_map):
    result = u''
    num_bytes = 0
    i = 0
    while i <= len(input) - 1:
	c = input[i]
	try:
	    if c in nonspacing_diacritical_marks:
		i += 1
		if i > len(input) - 1:
		    raise KeyError
		str = c + input[i]
		if str in decoding_map:
		    num_bytes += 2
		    result += decoding_map[str]
		else:
		    raise KeyError
	    else:
		num_bytes += 1
		result += c
	except KeyError:
	    if errors == 'strict': 
		raise UnicodeError, "invalid iso6937 character"
	    elif errors == 'replace': 
		num_bytes += 1
		result += chr(0x3f) #question mark
	    elif errors == 'ignore': 
		pass
	    else: raise UnicodeError, "unknown error handling"
	i += 1
    return (result, num_bytes)

### Codec APIs
class Codec(codecs.Codec):

    def encode(self,input,errors='strict'):
	return iso6937_encode(input,errors,encoding_map)

    def decode(self,input,errors='strict'):
	return iso6937_decode(input,errors,decoding_map)

class IncrementalEncoder(codecs.IncrementalEncoder):
    def encode(self, input, final=False):
	return iso6937_encode(input,self.errors,encoding_map)[0]

class IncrementalDecoder(codecs.IncrementalDecoder):
    def decode(self, input, final=False):
	return iso6937_decode(input,self.errors,decoding_map)[0]

class StreamWriter(Codec,codecs.StreamWriter):
    pass

class StreamReader(Codec,codecs.StreamReader):
    pass

### encodings module API
def getregentry():
    return codecs.CodecInfo(
        name='iso6937',
        encode=Codec().encode,
        decode=Codec().decode,
        incrementalencoder=IncrementalEncoder,
        incrementaldecoder=IncrementalDecoder,
        streamreader=StreamReader,
        streamwriter=StreamWriter,
    )

nonspacing_diacritical_marks = ['\xc1', '\xc2', '\xc3', '\xc4', '\xc5', '\xc6', '\xc7', '\xc8', '\xca', '\xcb', '\xcd', '\xce', '\xcf']

### Decoding Map
decoding_map = {
    '\xc1'+'A': u'À',
    '\xc1'+'E': u'È',
    '\xc1'+'I': u'Ì',
    '\xc1'+'O': u'Ò',
    '\xc1'+'U': u'Ù',
    '\xc1'+'a': u'à',
    '\xc1'+'e': u'è',
    '\xc1'+'i': u'ì',
    '\xc1'+'o': u'ò',
    '\xc1'+'u': u'ù',

    '\xc2'+'A': u'Á',
    '\xc2'+'C': u'Ć',
    '\xc2'+'E': u'É',
    '\xc2'+'I': u'Í',
    '\xc2'+'L': u'Ĺ',
    '\xc2'+'N': u'Ń',
    '\xc2'+'O': u'Ó',
    '\xc2'+'R': u'Ŕ',
    '\xc2'+'S': u'Ś',
    '\xc2'+'U': u'Ú',
    '\xc2'+'Y': u'Ý',
    '\xc2'+'Z': u'Ź',
    '\xc2'+'a': u'á',
    '\xc2'+'c': u'ć',
    '\xc2'+'e': u'é',
    '\xc2'+'g': u'ģ',
    '\xc2'+'i': u'í',
    '\xc2'+'l': u'ĺ',
    '\xc2'+'n': u'ń',
    '\xc2'+'o': u'ó',
    '\xc2'+'r': u'ŕ',
    '\xc2'+'s': u'ś',
    '\xc2'+'u': u'ú',
    '\xc2'+'y': u'ý',
    '\xc2'+'z': u'ź',

    '\xc3'+'A': u'Â',
    '\xc3'+'C': u'Ĉ',
    '\xc3'+'E': u'Ê',
    '\xc3'+'G': u'Ĝ',
    '\xc3'+'H': u'Ĥ',
    '\xc3'+'I': u'Î',
    '\xc3'+'J': u'Ĵ',
    '\xc3'+'O': u'Ô',
    '\xc3'+'S': u'Ŝ',
    '\xc3'+'U': u'Û',
    '\xc3'+'W': u'Ŵ',
    '\xc3'+'Y': u'Ŷ',
    '\xc3'+'a': u'â',
    '\xc3'+'c': u'ĉ',
    '\xc3'+'e': u'ê',
    '\xc3'+'g': u'ĝ',
    '\xc3'+'h': u'ĥ',
    '\xc3'+'i': u'î',
    '\xc3'+'j': u'ĵ',
    '\xc3'+'o': u'ô',
    '\xc3'+'s': u'ŝ',
    '\xc3'+'u': u'û',
    '\xc3'+'w': u'ŵ',
    '\xc3'+'y': u'ŷ',

    '\xc4'+'A': u'Ã',
    '\xc4'+'I': u'Ĩ',
    '\xc4'+'N': u'Ñ',
    '\xc4'+'O': u'Õ',
    '\xc4'+'U': u'Ũ',
    '\xc4'+'a': u'ã',
    '\xc4'+'i': u'ĩ',
    '\xc4'+'n': u'ñ',
    '\xc4'+'o': u'õ',
    '\xc4'+'u': u'ũ',

    '\xc5'+'A': u'Ā',
    '\xc5'+'E': u'Ē',
    '\xc5'+'I': u'Ī',
    '\xc5'+'O': u'Ō',
    '\xc5'+'U': u'Ū',
    '\xc5'+'a': u'ā',
    '\xc5'+'e': u'ē',
    '\xc5'+'i': u'ī',
    '\xc5'+'o': u'ō',
    '\xc5'+'u': u'ū',

    '\xc6'+'A': u'Ă',
    '\xc6'+'G': u'Ğ',
    '\xc6'+'U': u'Ŭ',
    '\xc6'+'a': u'ă',
    '\xc6'+'g': u'ğ',
    '\xc6'+'u': u'ŭ',

    '\xc7'+'C': u'Ċ',
    '\xc7'+'E': u'Ė',
    '\xc7'+'G': u'Ġ',
    '\xc7'+'I': u'İ',
    '\xc7'+'Z': u'Ż',
    '\xc7'+'c': u'ċ',
    '\xc7'+'e': u'ė',
    '\xc7'+'g': u'ġ',
    '\xc7'+'z': u'ż',

    '\xc8'+'A': u'Ä',
    '\xc8'+'E': u'Ë',
    '\xc8'+'I': u'Ï',
    '\xc8'+'O': u'Ö',
    '\xc8'+'U': u'Ü',
    '\xc8'+'Y': u'Ÿ',
    '\xc8'+'a': u'ä',
    '\xc8'+'e': u'ë',
    '\xc8'+'i': u'ï',
    '\xc8'+'o': u'ö',
    '\xc8'+'u': u'ü',
    '\xc8'+'y': u'ÿ',

    '\xca'+'A': u'Å',
    '\xca'+'U': u'Ů',
    '\xca'+'a': u'å',
    '\xca'+'u': u'ů',

    '\xcb'+'C': u'Ç',
    '\xcb'+'G': u'Ģ',
    '\xcb'+'K': u'Ķ',
    '\xcb'+'L': u'Ļ',
    '\xcb'+'N': u'Ņ',
    '\xcb'+'R': u'Ŗ',
    '\xcb'+'S': u'Ş',
    '\xcb'+'T': u'Ţ',
    '\xcb'+'c': u'ç',
    '\xcb'+'k': u'ķ',
    '\xcb'+'l': u'ļ',
    '\xcb'+'n': u'ņ',
    '\xcb'+'r': u'ŗ',
    '\xcb'+'s': u'ş',
    '\xcb'+'t': u'ţ',

    '\xcd'+'O': u'Ő',
    '\xcd'+'U': u'Ű',
    '\xcd'+'o': u'ő',
    '\xcd'+'u': u'ű',

    '\xce'+'A': u'Ą',
    '\xce'+'E': u'Ę',
    '\xce'+'I': u'Į',
    '\xce'+'U': u'Ų',
    '\xce'+'a': u'ą',
    '\xce'+'e': u'ę',
    '\xce'+'i': u'į',
    '\xce'+'u': u'ų',

    '\xcf'+'C': u'Č',
    '\xcf'+'D': u'Ď',
    '\xcf'+'E': u'Ě',
    '\xcf'+'L': u'Ľ',
    '\xcf'+'N': u'Ň',
    '\xcf'+'R': u'Ř',
    '\xcf'+'S': u'Š',
    '\xcf'+'T': u'Ť',
    '\xcf'+'Z': u'Ž',
    '\xcf'+'c': u'č',
    '\xcf'+'d': u'ď',
    '\xcf'+'e': u'ě',
    '\xcf'+'l': u'ľ',
    '\xcf'+'n': u'ň',
    '\xcf'+'r': u'ř',
    '\xcf'+'s': u'š',
    '\xcf'+'t': u'ť',
    '\xcf'+'z': u'ž',
}

### Encoding Map
encoding_map = {
    u'À': ['\xc1','A'],
    u'È': ['\xc1','E'],
    u'Ì': ['\xc1','I'],
    u'Ò': ['\xc1','O'],
    u'Ù': ['\xc1','U'],
    u'à': ['\xc1','a'],
    u'è': ['\xc1','e'],
    u'ì': ['\xc1','i'],
    u'ò': ['\xc1','o'],
    u'ù': ['\xc1','u'],

    u'Á': ['\xc2','A'],
    u'Ć': ['\xc2','C'],
    u'É': ['\xc2','E'],
    u'Í': ['\xc2','I'],
    u'Ĺ': ['\xc2','L'],
    u'Ń': ['\xc2','N'],
    u'Ó': ['\xc2','O'],
    u'Ŕ': ['\xc2','R'],
    u'Ś': ['\xc2','S'],
    u'Ú': ['\xc2','U'],
    u'Ý': ['\xc2','Y'],
    u'Ź': ['\xc2','Z'],
    u'á': ['\xc2','a'],
    u'ć': ['\xc2','c'],
    u'é': ['\xc2','e'],
    u'ģ': ['\xc2','g'],
    u'í': ['\xc2','i'],
    u'ĺ': ['\xc2','l'],
    u'ń': ['\xc2','n'],
    u'ó': ['\xc2','o'],
    u'ŕ': ['\xc2','r'],
    u'ś': ['\xc2','s'],
    u'ú': ['\xc2','u'],
    u'ý': ['\xc2','y'],
    u'ź': ['\xc2','z'],

    u'Â': ['\xc3','A'],
    u'Ĉ': ['\xc3','C'],
    u'Ê': ['\xc3','E'],
    u'Ĝ': ['\xc3','G'],
    u'Ĥ': ['\xc3','H'],
    u'Î': ['\xc3','I'],
    u'Ĵ': ['\xc3','J'],
    u'Ô': ['\xc3','O'],
    u'Ŝ': ['\xc3','S'],
    u'Û': ['\xc3','U'],
    u'Ŵ': ['\xc3','W'],
    u'Ŷ': ['\xc3','Y'],
    u'â': ['\xc3','a'],
    u'ĉ': ['\xc3','c'],
    u'ê': ['\xc3','e'],
    u'ĝ': ['\xc3','g'],
    u'ĥ': ['\xc3','h'],
    u'î': ['\xc3','i'],
    u'ĵ': ['\xc3','j'],
    u'ô': ['\xc3','o'],
    u'ŝ': ['\xc3','s'],
    u'û': ['\xc3','u'],
    u'ŵ': ['\xc3','w'],
    u'ŷ': ['\xc3','y'],

    u'Ã': ['\xc4','A'],
    u'Ĩ': ['\xc4','I'],
    u'Ñ': ['\xc4','N'],
    u'Õ': ['\xc4','O'],
    u'Ũ': ['\xc4','U'],
    u'ã': ['\xc4','a'],
    u'ĩ': ['\xc4','i'],
    u'ñ': ['\xc4','n'],
    u'õ': ['\xc4','o'],
    u'ũ': ['\xc4','u'],

    u'Ā': ['\xc5','A'],
    u'Ē': ['\xc5','E'],
    u'Ī': ['\xc5','I'],
    u'Ō': ['\xc5','O'],
    u'Ū': ['\xc5','U'],
    u'ā': ['\xc5','a'],
    u'ē': ['\xc5','e'],
    u'ī': ['\xc5','i'],
    u'ō': ['\xc5','o'],
    u'ū': ['\xc5','u'],

    u'Ă': ['\xc6','A'],
    u'Ğ': ['\xc6','G'],
    u'Ŭ': ['\xc6','U'],
    u'ă': ['\xc6','a'],
    u'ğ': ['\xc6','g'],
    u'ŭ': ['\xc6','u'],

    u'Ċ': ['\xc7','C'],
    u'Ė': ['\xc7','E'],
    u'Ġ': ['\xc7','G'],
    u'İ': ['\xc7','I'],
    u'Ż': ['\xc7','Z'],
    u'ċ': ['\xc7','c'],
    u'ė': ['\xc7','e'],
    u'ġ': ['\xc7','g'],
    u'ż': ['\xc7','z'],

    u'Ä': ['\xc8','A'],
    u'Ë': ['\xc8','E'],
    u'Ï': ['\xc8','I'],
    u'Ö': ['\xc8','O'],
    u'Ü': ['\xc8','U'],
    u'Ÿ': ['\xc8','Y'],
    u'ä': ['\xc8','a'],
    u'ë': ['\xc8','e'],
    u'ï': ['\xc8','i'],
    u'ö': ['\xc8','o'],
    u'ü': ['\xc8','u'],
    u'ÿ': ['\xc8','y'],

    u'Å': ['\xca','A'],
    u'Ů': ['\xca','U'],
    u'å': ['\xca','a'],
    u'ů': ['\xca','u'],

    u'Ç': ['\xcb','C'],
    u'Ģ': ['\xcb','G'],
    u'Ķ': ['\xcb','K'],
    u'Ļ': ['\xcb','L'],
    u'Ņ': ['\xcb','N'],
    u'Ŗ': ['\xcb','R'],
    u'Ş': ['\xcb','S'],
    u'Ţ': ['\xcb','T'],
    u'ç': ['\xcb','c'],
    u'ķ': ['\xcb','k'],
    u'ļ': ['\xcb','l'],
    u'ņ': ['\xcb','n'],
    u'ŗ': ['\xcb','r'],
    u'ş': ['\xcb','s'],
    u'ţ': ['\xcb','t'],

    u'Ő': ['\xcd','O'],
    u'Ű': ['\xcd','U'],
    u'ő': ['\xcd','o'],
    u'ű': ['\xcd','u'],

    u'Ą': ['\xce','A'],
    u'Ę': ['\xce','E'],
    u'Į': ['\xce','I'],
    u'Ų': ['\xce','U'],
    u'ą': ['\xce','a'],
    u'ę': ['\xce','e'],
    u'į': ['\xce','i'],
    u'ų': ['\xce','u'],

    u'Č': ['\xcf','C'],
    u'Ď': ['\xcf','D'],
    u'Ě': ['\xcf','E'],
    u'Ľ': ['\xcf','L'],
    u'Ň': ['\xcf','N'],
    u'Ř': ['\xcf','R'],
    u'Š': ['\xcf','S'],
    u'Ť': ['\xcf','T'],
    u'Ž': ['\xcf','Z'],
    u'č': ['\xcf','c'],
    u'ď': ['\xcf','d'],
    u'ě': ['\xcf','e'],
    u'ľ': ['\xcf','l'],
    u'ň': ['\xcf','n'],
    u'ř': ['\xcf','r'],
    u'š': ['\xcf','s'],
    u'ť': ['\xcf','t'],
    u'ž': ['\xcf','z'],
}