Variable value verified with a tuple and not a set or a list

[adrien-lsh]

Hey there,

I don't think the title is self explanatory enough, so I'll just paste the code here.

from typing import Literal

def greet(string: Literal['Hello', 'Bonjour', 'Hola']):
    ...

def function(string: str):
    if string == 'Hello' or string == 'Bonjour' or string == 'Hola':
        greet(string)
    if string in ['Hello', 'Bonjour', 'Hola']:  # error: Argument of type "str" cannot be assigned to parameter "string" of type "Literal['Hello', 'Bonjour', 'Hola']" in function "greet"
        greet(string)
    if string in {'Hello', 'Bonjour', 'Hola'}:  # error: Argument of type "str" cannot be assigned to parameter "string" of type "Literal['Hello', 'Bonjour', 'Hola']" in function "greet"
        greet(string)
    if string in ('Hello', 'Bonjour', 'Hola'):  # no error
        greet(string)

Why does pyright understands that the variable is a Literal['Hello', ...] when comparing with a tuple, but not a list or a set ?
If it is the expected behavior, should I use tuple just to silence the warning, or is it better in some ways that make it the only possible option for pyright ?

Any help is appreciated, thank you :)

[erictraut]

This behavior is intended. Literal values are retained when inferring the type of a tuple expression but are not when inferring the type of a list, dict or set expression. This is because list, dict and set objects are mutable. If literals were retained when inferring their types, it would be very inconvenient. For example, if you had the statement x = [1], you probably wouldn't want the type of x to be inferred as list[Literal[1]]. For more details about pyright's inference behaviors for container expressions, refer to this documentation.

[adrien-lsh]

Oh I see, that makes a lot of sense thank you !