Function make_meta

dask/dataframe/dispatch.py:95–135

This method creates meta-data based on the type of ``x``, and ``parent_meta`` if supplied. Parameters ---------- x : Object of any type. Object to construct meta-data from. index : Index, optional Any index to use in the metadata. This is a pass-through

(x, index=None, parent_meta=None)

def make_meta(x, index=None, parent_meta=None):
    """
    This method creates meta-data based on the type of ``x``,
    and ``parent_meta`` if supplied.

    Parameters
    ----------
    x : Object of any type.
        Object to construct meta-data from.
    index :  Index, optional
        Any index to use in the metadata. This is a pass-through
        parameter to dispatches registered.
    parent_meta : Object, default None
        If ``x`` is of arbitrary types and thus Dask cannot determine
        which back-end to be used to generate the meta-data for this
        object type, in which case ``parent_meta`` will be used to
        determine which back-end to select and dispatch to. To
        utilize this parameter ``make_meta_obj`` has be dispatched.
        If ``parent_meta`` is ``None``, a pandas DataFrame is used for
        ``parent_meta`` that chooses pandas as the backend.

    Returns
    -------
    A valid meta-data
    """

    if not isinstance(x, (pd.Series, pd.DataFrame, pd.Index)) and hasattr(x, "_meta"):
        if is_dask_collection(x):
            return x._meta

    try:
        return make_meta_dispatch(x, index=index)
    except TypeError:
        if parent_meta is not None:
            func = make_meta_obj.dispatch(type(parent_meta))
            return func(x, index=index)
        else:
            # Default to using the pandas backend
            # if ``parent_meta`` is not specified
            func = make_meta_obj.dispatch(pd.DataFrame)
            return func(x, index=index)


def union_categoricals(to_union, sort_categories=False, ignore_order=False):