Narrow down right hand-side of an == expression in ifs

[elenakrittik]

Is your feature request related to a problem? Please describe.

if a == 1 narrows a to Literal[1], but if 1 == a does not

playground

Describe the solution you’d like

Narrow down the type of a in both cases

[erictraut]

Yes, this is intended behavior. Pyright supports the x == L and x != L type guard patterns for type narrowing, but it doesn't support the converse, which is not a typical usage pattern in Python.

For a full list of supported type guard patterns, refer to this documentation.

We do occasionally add support for new patterns, but each one requires special-case logic and has the potential for slowing down overall type analysis, so we do so only in cases where we have strong signal from pyright users that it's a common pattern. Adding this case would likely slow down code flow analysis in a measurable way, especially for complex functions in untyped code. This is important when inferring return types for untyped library functions, especially in scientific libraries, so I don't think it would be a good tradeoff for pyright users to add support for this type guard pattern.

I've rarely seen L == x used in code, and it's trivial to change this to the more common x == L pattern.

For those reasons, I'm going to reject this enhancement request for now. I'm willing to revisit the decision if there's sufficient interest from pyright users.

[elenakrittik]

Thank you for the explanation!