Let usernames have funnier char than just alphanumeric

[josephernest]

Because of many malicious utf8 characters such as this one:
http://unicode-table.com/fr/200D/

or even things like this in username: 
 (see http://code.cside.com/3rdpage/us/newLine.html)

I had to ban everything else than alphanumeric characters https://github.com/josephernest/talktalktalk/blob/master/talktalktalk.py#L49

username = re.sub(r'\W+', '', username)

Would be cool to allow more UTF8 characters (even ❤ ☀ ☆ ☂ ☻ ♞ ☯ ).

Which ones to allow, which ones to ban?

[dangayle]

Here's a great list: http://apps.timwhitlock.info/emoji/tables/unicode

These are basically the ones built into iOS, which you know have been scrutinized for safety and appropriateness. Some sort of regex whitelist might be available somewhere.