Normalize unicode identifier

[ickc]

Is your feature request related to a problem? Please describe.

Unicode normalization in unicode identifier. c.f.

• 2. Lexical analysis — Python 3.9.6 documentation
• Issue 44981: wildcard imports should normalise names in __all__ - Python tracker
• python - Identifier normalization: Why is the micro sign converted into the Greek letter mu? - Stack Overflow

The issue is that while some characters can be used as identifier, internally it is normalized. So when looked up using __all__ or getattr, it would results in an error. (See the 2nd & 3rd link for respective example.)

Describe the solution you'd like

Black should normalize all identifiers accordingly, perhaps also those inside __all__ or getattr, so that this normalization is made explicit in the code. When someone copy and paste an identifier to __all__ or getattr, it would then just works.

Describe alternatives you've considered

Fix it in Python: Issue 44981: wildcard imports should normalise names in __all__ - Python tracker.

Additional context

Edit: keys of globals() & locals() also shares the same problem:

• Issue 42680: unicode identifiers not accessible or assignable through globals() - Python tracker