conftest.py

from __future__ import annotations as _annotations

import functools
import gc
import importlib.util
import json
import os
import re
from dataclasses import dataclass
from pathlib import Path
from time import sleep, time
from typing import Any, Callable, Literal

import hypothesis
import pytest

from pydantic_core import ArgsKwargs, CoreSchema, SchemaValidator, ValidationError, validate_core_schema
from pydantic_core.core_schema import CoreConfig

__all__ = 'Err', 'PyAndJson', 'plain_repr', 'infinite_generator'

hypothesis.settings.register_profile('fast', max_examples=2)
hypothesis.settings.register_profile('slow', max_examples=1_000)
hypothesis.settings.load_profile(os.getenv('HYPOTHESIS_PROFILE', 'fast'))


def plain_repr(obj):
    r = repr(obj)
    r = re.sub(r',\s*([)}])', r'\1', r)
    r = re.sub(r'\s+', '', r)
    return r


@dataclass
class Err:
    message: str
    errors: Any | None = None

    def __repr__(self):
        if self.errors:
            return f'Err({self.message!r}, errors={self.errors!r})'
        else:
            return f'Err({self.message!r})'


def json_default(obj):
    if isinstance(obj, ArgsKwargs):
        raise pytest.skip('JSON skipping ArgsKwargs')
    else:
        raise TypeError(f'Object of type {type(obj).__name__} is not JSON serializable')


class PyAndJsonValidator:
    def __init__(
        self,
        schema: CoreSchema,
        config: CoreConfig | None = None,
        *,
        validator_type: Literal['json', 'python'] | None = None,
    ):
        self.validator = SchemaValidator(validate_core_schema(schema), config)
        self.validator_type = validator_type

    def validate_python(self, py_input, strict: bool | None = None, context: Any = None):
        return self.validator.validate_python(py_input, strict=strict, context=context)

    def validate_json(self, json_str: str, strict: bool | None = None, context: Any = None):
        return self.validator.validate_json(json_str, strict=strict, context=context)

    def validate_test(self, py_input, strict: bool | None = None, context: Any = None):
        if self.validator_type == 'json':
            return self.validator.validate_json(
                json.dumps(py_input, default=json_default), strict=strict, context=context
            )
        else:
            assert self.validator_type == 'python', self.validator_type
            return self.validator.validate_python(py_input, strict=strict, context=context)

    def isinstance_test(self, py_input, strict: bool | None = None, context: Any = None):
        if self.validator_type == 'json':
            try:
                self.validator.validate_json(json.dumps(py_input), strict=strict, context=context)
                return True
            except ValidationError:
                return False
        else:
            assert self.validator_type == 'python', self.validator_type
            return self.validator.isinstance_python(py_input, strict=strict, context=context)


PyAndJson = type[PyAndJsonValidator]


@pytest.fixture(params=['python', 'json'])
def py_and_json(request) -> PyAndJson:
    class ChosenPyAndJsonValidator(PyAndJsonValidator):
        __init__ = functools.partialmethod(PyAndJsonValidator.__init__, validator_type=request.param)

    return ChosenPyAndJsonValidator


class StrictModeType:
    def __init__(self, schema: bool, extra: bool):
        assert schema or extra
        self.schema = schema
        self.validator_args = {'strict': True} if extra else {}


@pytest.fixture(
    params=[
        StrictModeType(schema=True, extra=False),
        StrictModeType(schema=False, extra=True),
        StrictModeType(schema=True, extra=True),
    ],
    ids=['strict-schema', 'strict-extra', 'strict-both'],
)
def strict_mode_type(request) -> StrictModeType:
    return request.param


@pytest.fixture
def tmp_work_path(tmp_path: Path):
    """
    Create a temporary working directory.
    """
    previous_cwd = Path.cwd()
    os.chdir(tmp_path)

    yield tmp_path

    os.chdir(previous_cwd)


@pytest.fixture
def import_execute(request, tmp_work_path: Path):
    def _import_execute(source: str, *, custom_module_name: str | None = None):
        module_name = custom_module_name or request.node.name

        module_path = tmp_work_path / f'{module_name}.py'
        module_path.write_text(source)
        spec = importlib.util.spec_from_file_location('__main__', str(module_path))
        module = importlib.util.module_from_spec(spec)
        try:
            spec.loader.exec_module(module)
        except KeyboardInterrupt:
            print('KeyboardInterrupt')
        else:
            return module

    return _import_execute


@pytest.fixture
def pydantic_version():
    try:
        import pydantic

        # include major and minor version only
        return '.'.join(pydantic.__version__.split('.')[:2])
    except ImportError:
        return 'latest'


def infinite_generator():
    i = 0
    while True:
        yield i
        i += 1


def assert_gc(test: Callable[[], bool], timeout: float = 10) -> None:
    """Helper to retry garbage collection until the test passes or timeout is
    reached.

    This is useful on free-threading where the GC collect call finishes before
    all cleanup is done.
    """
    start = now = time()
    while now - start < timeout:
        if test():
            return
        gc.collect()
        sleep(0.1)
        now = time()
    raise AssertionError('Timeout waiting for GC')