why is unicode segmentation used?

[rnag]

Just curious. Basically, in the context of this line:

pyheck/Cargo.toml

Line 17 in 28fa3d9

heck = { version = "0.4.0", features = ["unicode"] }

I'm not too sure about what the unicode feature is myself, reading up on it in the heck docs now.

Currently i have some test code in rust at least:

assert_eq!("a.b.c".to_snake_case(), "a_b_c");
assert_eq!("a'b'c".to_snake_case(), "a_b_c");
assert_eq!("2this##@$@#$.23432isaTest.Thanks.To.BOB.Myf\"r'e\"'e''nd.".to_snake_case(),
                 "2this_23432isa_test_thanks_to_bob_myf_r_e_e_nd")

this appears to work without the unicode feature, i.e. with heck = "0.4.0", but when I add the unicode feature then the assert stmts seem to fail. I might be wrong, but for valid snake case it might be worth removing the . and the , in the results. Basically, in my use case I want to turn a bunch of words into valid identifier names in python -- so for ex. a_b_c = 2 would work out in my use case, but with special chars it would fail. Is it possible to just have a _ in the returned string?

Side note: also 2this is not a valid identifier in python, but I understand that's not a goal of this library for obvious reasons of course.

[kevinheavey]

Hey didn't see this until now!

I'll ask around if there are good ways to publish multiple versions of the library with different features activated in the underlying Rust library

[rnag]

Awesome, no worries and thanks. Also, just wanted to add I appreciate the great work you are putting into this and with integrating with some cool Rust libraries.