fix(generator): use single underscore in kebab to snake case conversion

[DavidLeeds]

Addresses the case where "Kebab-Case" was converted to "kebab__case". After the change, we get "kebab_case".

issue: #88

[6d7a]

Unfortunately, this can lead to problematic edge cases. Consider the hypothetical ASN.1 type:

TestType ::= SEQUENCE {
   doner-kebab BOOLEAN,
   doner_kebab INTEGER
}

If we do not have a name transformation that retains distinction between - and _, the bindings for the type above will not compile. However, given that this is such an outrageously exotic edge case, I'd be up for introducing a config parameter to the Rasn backend, so that the user can decide how to handle name transformation.

[DavidLeeds]

@6d7a this is a valid concern, but I noted that the snake case conversion already makes no guarantees that the output will be unique for similar inputs. The following four existing test cases output the same result:

assert_eq!(generator.to_rust_snake_case("HelloWorld"), "hello_world");
assert_eq!(generator.to_rust_snake_case("helloWorld"), "hello_world");
assert_eq!(generator.to_rust_snake_case("hello-world"), "hello_world");
assert_eq!(generator.to_rust_snake_case("HelloWORLD"), "hello_world");

My change simply makes the outlier case consistent with the established behavior.

If schemas are using minor whitespace and punctuation changes to differentiate definitions, perhaps the code generator will be the forcing function to drive more responsible naming decisions.

[6d7a]

My bad, a valid fix indeed. Thank you and sorry for the confusion.