feat: add AnonymousExprError

[EdAbati]

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

Checklist

• Code follows style guide (ruff)
• Tests added
• Documented the changes

If you have comments or can explain your changes, please do so below

I propose to add a error for the (very) common error message about anonlymous expressions.

Ideally I would like to move this a level up (since we raise it for every backend) but I get all sorts of mypy error if I do that. e.g. mypy thinks root_names could still be None

[EdAbati]

ruff would still complain if we don't move the string to a separate variable

[MarcoGorelli]

awesome, thanks! i think it's fine to now move this up - in theory, I don't see why we should be raising this error at all, i think for now it's fine to just raise the error at the compliant level

[EdAbati]

I don't see why we should be raising this error at all

Do you mean that we should start supporting this or that we should let the backends raise their own errors?

i think for now it's fine to just raise the error at the compliant level

I tried to do something like this:

class ExprNameNamespace:
    ...
    def keep(self: Self) -> ExprT:
         def _func(plx):
            compliant_expr = self._expr._to_compliant_expr(plx)
            if compliant_expr._root_names is None:
                raise AnonymousExprError.from_expr_name(".name.keep")
            return compliant_expr.name.keep()
        
        return self._expr.__class__(_func)

but mypy was not very happy :D

[MarcoGorelli]

Do you mean that we should start supporting this or that we should let the backends raise their own errors?

I think df.group_by('a').agg(nw.all().min()) is well-defined, and there's no reason any current backend can't do it, and that we can aim to support it everywhere in the future. In theory, I hope that in the future it'll be possible for us to support it - so, it's OK for now if for some backends we raise AnonymousExprError

[MarcoGorelli]

thanks @EdAbati !