Repository navigation
Disagreement with CPython about which unicode characters are alnum #7518
Description
Summary
The regex module interprets \w as matching more characters than it seems like it should (certainly more than CPython does).
Expected
import re
assert not re.match(r"\w", "\u0345"), r"\w should not match U+0345 (category Mn)"
This assertion should pass, but instead it fails.
Python Documentation
https://docs.python.org/3/library/re.html
This says that it should match anything for which isalnum() is true. This turns out to be also an area where RustPython and CPython disagree. "\u0345".isalnum() returns False on CPython and True on RustPython.
Activity
joshuamegnauth54 commentedon Apr 14, 2026
I took a look into this issue. It turns out that Rust and Python disagree on alphanumeric characters. Rust strictly(?) uses the Unicode property for alphanumeric whereas Python considers visibility. So,
\u{0345}is supposed to be combined with another character in order to be visible and therefore isn't considered alphanumeric by Python. Interestingly, Rust actually prints it whereas Python does not.^These differ!
Full disclosure, I'm not entirely sure if my research and tests are correct. This is just what I found so far. Seems like I'm on the right track, though. 😁 I'll try to fix this issue soon!
fix: Python-Rust combining char diff in isalnum
fix: Python-Rust combining char diff in isalnum
fix: Ignore combining characters in SRE
fix: Python-Rust combining char diff in isalnum
fix: Ignore combining characters in SRE
fix: Python-Rust combining char diff in isalnum (RustPython#7612)
fix: Python-Rust combining char diff in isalnum (#7612)