Skip signature check for functions in the standard library, or give users control over whether the check is performed

[danhje]

Initial Checks

• I have searched Google & GitHub for similar requests and couldn't find anything
• I have read and followed the docs and still think this feature is missing

Description

This fails:

from datetime import datetime

from pydantic import AfterValidator, TypeAdapter
from typing_extensions import Annotated

ISOTimestamp = Annotated[datetime, AfterValidator(datetime.isoformat)]
print(TypeAdapter(ISOTimestamp).validate_python(1651576501.0))
# ValueError: no signature found for builtin <method 'isoformat' of 'datetime.datetime' objects>

A simple workaround is to call datetime.isoformat from a lambda, but that's a bit ugly.

inspect_validator excludes builtins from the inspection that fails, but does not exclude methods like datetime.timestamp from the inspection. I think it should. Or give the user control over whether the inspection takes place.

Affected Components

• Compatibility between releases
• Data validation/parsing
• Data serialization - .model_dump() and .model_dump_json()
• JSON Schema
• Dataclasses
• Model Config
• Field Types - adding or changing a particular data type
• Function validation decorator
• Generic Models
• Other Model behaviour - model_construct(), pickling, private attributes, ORM mode
• Plugins and integration with other tools - mypy, FastAPI, python-devtools, Hypothesis, VS Code, PyCharm, etc.

Selected Assignee: @adriangb

[adriangb]

I agree that it'd be nice if this just worked, but it's going to be hard to special case everything.
Maybe if we can't get the signature we can assume one, e.g. a single argument that is the input value?

@dmontagu I'm curious what you think

[danhje]

Yeah, I think that's reasonable.