How to use _get_to_datetime_fn method in pandera

Best Python code snippet using pandera_python

pandas_engine.py

Source:pandas_engine.py

...560    to_datetime_kwargs: Dict[str, Any] = dataclasses.field(561        default_factory=dict, compare=False, repr=False562    )563    @staticmethod564    def _get_to_datetime_fn(obj: Any) -> Callable:565        # NOTE: this is a hack to support pyspark.pandas. This needs to be566        # thoroughly tested, right now pyspark.pandas returns NA when a567        # dtype value can't be coerced into the target dtype.568        to_datetime_fn = pd.to_datetime569        if type(obj).__module__.startswith(570            "pyspark.pandas"571        ):  # pragma: no cover572            # pylint: disable=import-outside-toplevel573            import pyspark.pandas as ps574            to_datetime_fn = ps.to_datetime575        if type(obj).__module__.startswith("modin.pandas"):576            # pylint: disable=import-outside-toplevel577            import modin.pandas as mpd578            to_datetime_fn = mpd.to_datetime579        return to_datetime_fn580@Engine.register_dtype(581    equivalents=[582        "time",583        "datetime",584        "datetime64",585        datetime.datetime,586        np.datetime64,587        dtypes.Timestamp,588        dtypes.Timestamp(),589        pd.Timestamp,590    ]591)592@immutable(init=True)593class DateTime(_BaseDateTime, dtypes.Timestamp):594    """Semantic representation of a potentially timezone-aware datetime.595    Uses ``np.dtype("datetime64[ns]")`` for non-timezone aware datetimes and596    :class:`pandas.DatetimeTZDtype` for timezone-aware datetimes.597    """598    type: Optional[_PandasDatetime] = dataclasses.field(599        default=None, init=False600    )601    unit: str = "ns"602    """The precision of the datetime data. Currently limited to "ns"."""603    tz: Optional[datetime.tzinfo] = None604    """The timezone."""605    to_datetime_kwargs: Dict[str, Any] = dataclasses.field(606        default_factory=dict, compare=False, repr=False607    )608    "Any additional kwargs passed to :func:`pandas.to_datetime` for coercion."609    tz_localize_kwargs: Dict[str, Any] = dataclasses.field(610        default_factory=dict, compare=False, repr=False611    )612    "Keyword arguments passed to :func:`pandas.Series.dt.tz_localize` for coercion."613    _default_tz_localize_kwargs = {614        "ambiguous": "infer",615    }616    def __post_init__(self):617        if self.tz is None:618            type_ = np.dtype("datetime64[ns]")619        else:620            type_ = pd.DatetimeTZDtype(self.unit, self.tz)621            # DatetimeTZDtype converted tz to tzinfo for us622            object.__setattr__(self, "tz", type_.tz)623        object.__setattr__(self, "type", type_)624    def _coerce(625        self, data_container: PandasObject, pandas_dtype: Any626    ) -> PandasObject:627        to_datetime_fn = self._get_to_datetime_fn(data_container)628        _tz_localize_kwargs = {629            **self._default_tz_localize_kwargs,630            **self.tz_localize_kwargs,631        }632        def _to_datetime(col: PandasObject) -> PandasObject:633            col = to_datetime_fn(col, **self.to_datetime_kwargs)634            if (635                hasattr(pandas_dtype, "tz")636                and pandas_dtype.tz is not None637                and col.dt.tz is None638            ):639                # localize datetime column so that it's timezone-aware640                col = col.dt.tz_localize(641                    pandas_dtype.tz,642                    **_tz_localize_kwargs,643                )644            return col.astype(pandas_dtype)645        if isinstance(data_container, pd.DataFrame):646            # pd.to_datetime transforms a df input into a series.647            # We actually want to coerce every columns.648            return data_container.transform(_to_datetime)649        return _to_datetime(data_container)650    @classmethod651    def from_parametrized_dtype(cls, pd_dtype: pd.DatetimeTZDtype):652        """Convert a :class:`pandas.DatetimeTZDtype` to653        a Pandera :class:`pandera.engines.pandas_engine.DateTime`."""654        return cls(unit=pd_dtype.unit, tz=pd_dtype.tz)  # type: ignore655    def coerce(self, data_container: PandasObject) -> PandasObject:656        return self._coerce(data_container, pandas_dtype=self.type)657    def coerce_value(self, value: Any) -> Any:658        """Coerce an value to specified datatime type."""659        return self._get_to_datetime_fn(value)(660            value, **self.to_datetime_kwargs661        )662    def __str__(self) -> str:663        if self.type == np.dtype("datetime64[ns]"):664            return "datetime64[ns]"665        return str(self.type)666@Engine.register_dtype(667    equivalents=[668        "date",669        datetime.date,670        dtypes.Date,671        dtypes.Date(),672    ]673)674@immutable(init=True)675class Date(_BaseDateTime, dtypes.Date):676    """Semantic representation of a date data type."""677    type = np.dtype("object")678    to_datetime_kwargs: Dict[str, Any] = dataclasses.field(679        default_factory=dict, compare=False, repr=False680    )681    "Any additional kwargs passed to :func:`pandas.to_datetime` for coercion."682    # define __init__ to please mypy683    def __init__(  # pylint:disable=super-init-not-called684        self,685        to_datetime_kwargs: Optional[Dict[str, Any]] = None,686    ) -> None:687        object.__setattr__(688            self, "to_datetime_kwargs", to_datetime_kwargs or {}689        )690    def _coerce(691        self, data_container: PandasObject, pandas_dtype: Any692    ) -> PandasObject:693        to_datetime_fn = self._get_to_datetime_fn(data_container)694        def _to_datetime(col: PandasObject) -> PandasObject:695            col = to_datetime_fn(col, **self.to_datetime_kwargs)696            return col.astype(pandas_dtype).dt.date697        if isinstance(data_container, pd.DataFrame):698            # pd.to_datetime transforms a df input into a series.699            # We actually want to coerce every columns.700            return data_container.transform(_to_datetime)701        return _to_datetime(data_container)702    def coerce(self, data_container: PandasObject) -> PandasObject:703        return self._coerce(data_container, pandas_dtype="datetime64[ns]")704    def coerce_value(self, value: Any) -> Any:705        coerced = self._get_to_datetime_fn(value)(706            value, **self.to_datetime_kwargs707        )708        return coerced.date() if coerced is not None else pd.NaT709    def check(  # type: ignore710        self,711        pandera_dtype: DataType,712        data_container: Optional[pd.Series] = None,713    ) -> Union[bool, Iterable[bool]]:714        if not DataType.check(self, pandera_dtype, data_container):715            if data_container is None:716                return False717            else:718                return np.full_like(data_container, False)719        if data_container is None:...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.