emoji favicons depend on python string length in a weird way

[tewhalen]

Description

Trying to use "⚔️" as a favicon for my app, and finding that it gives me the error "ValueError: invalid favicon: ⚔️", coming out of line 61 of favicon.py.

The issue here is that the crossed-swords emoji is made out of two unicode code points: \u2694 ("Crossed Swords") and \uFE0F ("Variation Selector-16" i.e. "display as emoji"). Python's len() gives 2 for this when it's a string, which means favicon.py's _is_char() returns false for this sort of emoji.

Solving this in a general way, i.e. "how can we determine whether a given string is actually a single emoji" seems way more difficult than this problem deserves, so my suggestion would simply be to rewrite the test in _is_char() to something like
return len(favicon) == 1 or "\ufe0f" in favicon - this feels like it'll catch a bunch more cases.

[falkoschindler]

Thanks for reporting this issue, @tewhalen!

Solving this in a general way, i.e. "how can we determine whether a given string is actually a single emoji" seems way more difficult than this problem deserves

I totally agree. Adding "\ufe0f" in favicon would be a good quick fix - until someone wants to use something like 👍🏻 as a favicon. But I guess a general solution would require something like emoji. We can think about using it in the future.

Would you like to create a pull request handling \ufe0f? 🙂