Cipher with all unicode characters

Question

I'm trying to create a cipher with a scrambled alphabet contanting all possible characters, including unicodes, chinease/japanease symbols and etc...

I found out that I can print up to 65536 characters. I'm building the regular alphabet with a dict and a list of the numbers of those characters.

alphabet = { }
numeral = []
for n in xrange(65536):
    alphabet[unichr(n)] = n
    numeral.append(n)

As for the cipher:

cipher_alphabet = { }
for char in alphabet:
    cipher_alphabet[char] = choice(numeral)
    numeral.remove(cipher_alphabet[char])

To make a key/password I'm using the random.seed(key).

The problem is, when I try to use the dict to compare inputs from a file which contains unicode characters it give me:

KeyError: '\xe0'
�

In the file this character is 'à'.

The crypting part is like this:

message = open(file+'.txt').read()
crypted_message = ""
for word in message:
    for char in word:
        letter = cipher_alphabet.keys()[cipher_alphabet.values().index(alphabet[char])]
        crypted_message += letter

I've manage to use commom printable characters using:

for n in xrange(32, 127):
    alphabet[chr(n)] = n

But if I change the chr() to unichr() it gives me those errors.

Any hint?

Also, I've read that seed() is not a good method for cryptography, any hint for that too?

EDIT:

Thanks to @Joran manage to make it work.

For those interesed ...I've changed a bit the code.

for the alphabet:

for n in xrange(0, 65536):
    alphabet[n] = unichr(n)
    numeral.append(n)

for the cipher:

for x in alphabet:
    num = choice(numeral)
    crypted_alphabet[num] = alphabet[x]
    numeral.remove(num)

the crypting part:

message = open(file+'.txt','rb').read()
for n in message:
    num = alphabet.values().index(crypted_alphabet[n])
    crypted_message.append(num)