def __init__()

in src/python/turicreate/data_structures/sarray.py [0:0]

• 80 lines of code
• 39 McCabe index (conditional complexity)

    def __init__(self, data=[], dtype=None, ignore_cast_failure=False, _proxy=None):
        """
        __init__(data=list(), dtype=None, ignore_cast_failure=False)

        Construct a new SArray. The source of data includes: list,
        range, generators, map, filter, numpy.ndarray, pandas.Series, and urls.
        """

        if dtype is not None and type(dtype) != type:
            raise TypeError("dtype must be a type, e.g. use int rather than 'int'")

        if _proxy:
            self.__proxy__ = _proxy
        elif isinstance(data, SArray):
            if dtype is None:
                self.__proxy__ = data.__proxy__
            else:
                self.__proxy__ = data.astype(dtype).__proxy__
        else:
            self.__proxy__ = UnitySArrayProxy()

            ## data transfromation from generator and other iterable to list
            if self._is_iterable_required_to_listify(data):
                data = list(data)

            # we need to perform type inference
            if dtype is None:
                if HAS_PANDAS and isinstance(data, pandas.Series):
                    # if it is a pandas series get the dtype of the series
                    dtype = pytype_from_dtype(data.dtype)
                    if dtype == object:
                        # we need to get a bit more fine grained than that
                        dtype = infer_type_of_sequence(data.values)

                elif HAS_NUMPY and isinstance(data, numpy.ndarray):
                    # first try the fast inproc method
                    try:
                        from .. import numpy_loader

                        if numpy_loader.numpy_activation_successful():
                            from ..numpy import _fast_numpy_to_sarray

                            ret = _fast_numpy_to_sarray(data)
                            # conversion is good!
                            # swap the proxy.
                            self.__proxy__, ret.__proxy__ = (
                                ret.__proxy__,
                                self.__proxy__,
                            )
                            return
                        else:
                            dtype = infer_type_of_sequence(data)
                    except:
                        pass

                    # if it is a numpy array, get the dtype of the array
                    dtype = pytype_from_dtype(data.dtype)
                    if dtype == object:
                        # we need to get a bit more fine grained than that
                        dtype = infer_type_of_sequence(data)
                    if len(data.shape) == 2:
                        # we need to make it an array or a list
                        if dtype == float or dtype == int:
                            dtype = array.array
                        else:
                            dtype = list
                    elif len(data.shape) > 2:
                        raise TypeError(
                            "Cannot convert Numpy arrays of greater than 2 dimensions"
                        )

                elif isinstance(data, str) or (
                    sys.version_info.major < 3 and isinstance(data, unicode)
                ):
                    # if it is a file, we default to string
                    dtype = str
                elif isinstance(data, array.array):
                    dtype = pytype_from_array_typecode(data.typecode)
                elif isinstance(data, collections.Sequence):
                    # Covers any ordered python container and arrays.
                    # Convert it to a list first.
                    dtype = infer_type_of_sequence(data)
                else:
                    dtype = None

            if HAS_PANDAS and isinstance(data, pandas.Series):
                with cython_context():
                    self.__proxy__.load_from_iterable(
                        data.values, dtype, ignore_cast_failure
                    )
            elif isinstance(data, str) or (
                sys.version_info.major <= 2 and isinstance(data, unicode)
            ):
                internal_url = _make_internal_url(data)
                with cython_context():
                    self.__proxy__.load_autodetect(internal_url, dtype)
            elif (
                (HAS_NUMPY and isinstance(data, numpy.ndarray))
                or isinstance(data, array.array)
                or isinstance(data, collections.Sequence)
            ):

                with cython_context():
                    self.__proxy__.load_from_iterable(data, dtype, ignore_cast_failure)
            else:
                raise TypeError(
                    "Unexpected data source. "
                    "Possible data source types are: list, "
                    "numpy.ndarray, pandas.Series, and string(url)"
                )