feat(placeholders): allow regex valuelist transformation

[m4dh4t]

No description provided.

[thomaspatzke]

Is there a specific reason to use __init__ instead of __post_init__? A string regexp can also be replaced in __post_init__.

[thomaspatzke]

Would prefer to keep this as dataclass for the provided features, consistency and clean structure.

[thomaspatzke]

Unnecessary to implement in a dataclass.

[m4dh4t]

Regarding the transition of SigmaRegularExpression from the original dataclass, I made this change at the very end to prevent any confusion about the API and prevent typing issues should a typechecker be added to the project down the line.

If we use a dataclass and set regexp to be a SigmaString, the caller will expect the SigmaRegularExpression to be initialized directly with a SigmaString, rather than a str. Without directly looking at the __post_init__ hook, he will not be aware that he can pass a str that is then going to be converted.

On the other hand, using Union[SigmaString, str] makes it clear that a str can be passed at initialization, but does not reflect that once initialized only a SigmaString will be kept.

Moving away from a dataclass seemed like a clean alternative to make both the initialization API and the class API itself clear while still honoring the typing.

If we decide to keep it as a dataclass with a SigmaString only, I think we should remove the implicit conversion made in __post_init__ and update the various places where the constructor is called to reflect this change.

[thomaspatzke]

Moving away from a dataclass seemed like a clean alternative to make both the initialization API and the class API itself clear while still honoring the typing.

If we decide to keep it as a dataclass with a SigmaString only, I think we should remove the implicit conversion made in __post_init__ and update the various places where the constructor is called to reflect this change.

Init-only variables are the solution for such situations in dataclasses. Here the init-only variable would be the str or Union[str, SigmaString] and is passed to __post_init__ which converts it then into the SigmaString-only self.regexp value, that must be defined as dataclass field with init=False.