How to make Pylance take into account decorators?

[baruchli]

I’m working with the python-telegram-bot library and created a decorator to handle boilerplate checks for method parameters. The purpose of the decorator is to streamline parameter validation by moving it outside the core function logic.

Here’s the decorator:

`def callback_query_check(func):
"""Decorator to check callback query and user data"""

async def wrapper(update, context, *args, **kwargs):
    if any(
        x is None
        for x in [context.user_data, update.callback_query, update.effective_user]
    ):
        await context.bot.send_message(
            chat_id=update.effective_chat.id,
            text=(
                f"Unexpected termination due to invalid parameters of function {func.__name__}."
                " Please restart process by command '/start'"
            ),
            reply_markup=None,
        )
        logger.error("user_data is None or query is None")
        return ConversationHandler.END
    return await func(update, context, *args, **kwargs)

return wrapper`

I’ve applied the decorator to a function like this:

@callback_query_check async def cron_parameters_dialog(update: Update, context: CallbackContext): query = update.callback_query user_data = context.user_data if not input_strings[query.data]: ....

However, Pylance raises the following error for the query.data line:
Argument of type "str | None" cannot be assigned to parameter "key" of type "str" in function "getitem".
Type "str | None" is not assignable to type "str".
"None" is not assignable to "str".

It seems Pylance doesn’t recognize that the decorator ensures None values are handled. Is there a way to make Pylance account for the decorator logic or suppress this false positive effectively?

[erictraut]

If pyright encounters an unannotated decorator, it assumes that it doesn't modify the signature of the decorated function (which is the typical case). If your decorator modifies the signature of the decorated function, you need to provide type annotations for your decorator.

[baruchli]

In my case, the decorator does not modify the signature of the decorated function but performs routine validation of its parameters. Is there a way to inform Pyright that these parameter values are valid?

[erictraut]

It's difficult for me to understand the problem based on the code you've provided. The code is not properly formatted and contains a bunch of symbols that are undefined. For example, I don't know how the Update class is defined. I'm guessing that callback_query is an attribute or property whose type is declared as Callable | None or something like that. If my assumption is correct, then the answer is no, the Python type system doesn't provide a way for one function (the decorator) to tell another function (the decorated function) that "I've tested this attribute within this object, and it is not None". You would need to add an assert in the code within the decorated function to handle this.

assert update.callback_query is not None

Your question here is not specific to pyright but is more a general question about how to use type annotations within Python. I'm happy to answer such questions here in the pyright project, but you may find that the Python Typing forum is a better place to post such questions.

[baruchli]

Thank you for your comprehensive response.