Fails with unicode

[elinscott]

xdoctest fails to parse files containing unicode (even though this is valid python). In my instance an "ᵣ" was the issue. Converting to ASCII ("_r") made the test pass.

Full test can be found here https://github.com/elinscott/wannier90-input/actions/runs/18189413456/job/51780883667?pr=4

[Erotemic]

Can you point me at the doctest that failed? Or produce a MWE that I can add as a test to this repo?

[elinscott]

Is the link I provided in the above description not enough? What more do you need?

[elinscott]

... a bit more poking around showed that it's not just any unicode that triggers the bug, but here's one example which fails:

from pydantic import BaseModel, model_validator

class Model(BaseModel):
    """Here is some documentation."""
    x: int
    y: int

    @model_validator(mode="after")
    def validation_with_subscript_ᵣ(self):
        assert self.x > self.y
        return self

[Erotemic]

Odd, this seems to be a combination of pydantic and the unicode character in the function name.

"""
File: check.py

CommandLine:

    # Fails
    xdoctest -m check.py list

    # Works
    xdoctest -m check.py list --analysis=dynamic

    # Works
    xdoctest -m check.py all --analysis=dynamic

    # Works
    xdoctest -m check.py foo:0 --analysis=dynamic

    # Fails
    xdoctest -m check.py --analysis=dynamic Model.validation_with_subscript_ᵣ

    # Fails
    xdoctest -m check.py --analysis=static Model.validation_with_subscript_ᵣ

"""
from pydantic import BaseModel, model_validator


def foo():
    """

    Example:
        >>> from pydantic import BaseModel, model_validator
        >>> #
        >>> class Model(BaseModel):
        >>>     x: int
        >>>     y: int
        >>>     @model_validator(mode="after")
        >>>     def validation_with_subscript_ᵣ(self):
        >>>         assert self.x > self.y
        >>>         return self
    """


class Model(BaseModel):
    """Here is some documentation."""
    x: int
    y: int

    @model_validator(mode="after")
    def validation_with_subscript_ᵣ(self):
        """
        Example:
            >>> from pydantic import BaseModel, model_validator
            >>> #
            >>> class Model(BaseModel):
            >>>     x: int
            >>>     y: int
            >>>     @model_validator(mode="after")
            >>>     def validation_with_subscript_ᵣ(self):
            >>>         assert self.x > self.y
            >>>         return self

        """
        assert self.x > self.y
        return self

Wondering if it can be reproduced without pydantic at all?