Python numpy.core方法代碼示例

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import core [as 別名]
def block(arrays):
    # We need to override block since the numpy implementation can take two
    # different paths, one for concatenation, one for creating a large empty
    # result array in which parts are set.  Each assumes array input and
    # cannot be used directly.  Since it would be very costly to inspect all
    # arrays and then turn them back into a nested list, we just copy here the
    # second implementation, np.core.shape_base._block_slicing, since it is
    # shortest and easiest.
    (arrays, list_ndim, result_ndim,
     final_size) = np.core.shape_base._block_setup(arrays)
    shape, slices, arrays = np.core.shape_base._block_info_recursion(
        arrays, list_ndim, result_ndim)
    # Here, one line of difference!
    arrays, unit = _quantities2arrays(*arrays)
    # Back to _block_slicing
    dtype = np.result_type(*[arr.dtype for arr in arrays])
    F_order = all(arr.flags['F_CONTIGUOUS'] for arr in arrays)
    C_order = all(arr.flags['C_CONTIGUOUS'] for arr in arrays)
    order = 'F' if F_order and not C_order else 'C'
    result = np.empty(shape=shape, dtype=dtype, order=order)
    for the_slice, arr in zip(slices, arrays):
        result[(Ellipsis,) + the_slice] = arr
    return result, unit, None 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import core [as 別名]
def get_include():
    """
    Return the directory that contains the NumPy \\*.h header files.

    Extension modules that need to compile against NumPy should use this
    function to locate the appropriate include directory.

    Notes
    -----
    When using ``distutils``, for example in ``setup.py``.
    ::

        import numpy as np
        ...
        Extension('extension_name', ...
                include_dirs=[np.get_include()])
        ...

    """
    import numpy
    if numpy.show_config is None:
        # running from numpy source directory
        d = os.path.join(os.path.dirname(numpy.__file__), 'core', 'include')
    else:
        # using installed numpy core headers
        import numpy.core as core
        d = os.path.join(os.path.dirname(core.__file__), 'include')
    return d 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import core [as 別名]
def array_repr(arr, *args, **kwargs):
    # TODO: The addition of "unit='...'" doesn't worry about line
    # length.  Could copy & adapt _array_repr_implementation from
    # numpy.core.arrayprint.py
    cls_name = arr.__class__.__name__
    fake_name = '_' * len(cls_name)
    fake_cls = type(fake_name, (np.ndarray,), {})
    no_unit = np.array_repr(arr.view(fake_cls),
                            *args, **kwargs).replace(fake_name, cls_name)
    unit_part = f"unit='{arr.unit}'"
    pre, dtype, post = no_unit.rpartition('dtype')
    if dtype:
        return f"{pre}{unit_part}, {dtype}{post}", None, None
    else:
        return f"{no_unit[:-1]}, {unit_part})", None, None 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import core [as 別名]
def array_str(arr, *args, **kwargs):
    # TODO: The addition of the unit doesn't worry about line length.
    # Could copy & adapt _array_repr_implementation from
    # numpy.core.arrayprint.py
    no_unit = np.array_str(arr.value, *args, **kwargs)
    return no_unit + arr._unitstr, None, None 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import core [as 別名]
def array2string(a, *args, **kwargs):
    # array2string breaks on quantities as it tries to turn individual
    # items into float, which works only for dimensionless.  Since the
    # defaults would not keep any unit anyway, this is rather pointless -
    # we're better off just passing on the array view.  However, one can
    # also work around this by passing on a formatter (as is done in Angle).
    # So, we do nothing if the formatter argument is present and has the
    # relevant formatter for our dtype.
    formatter = args[6] if len(args) >= 7 else kwargs.get('formatter', None)

    if formatter is None:
        a = a.value
    else:
        # See whether it covers our dtype.
        from numpy.core.arrayprint import _get_format_function

        with np.printoptions(formatter=formatter) as options:
            try:
                ff = _get_format_function(a.value, **options)
            except Exception:
                # Shouldn't happen, but possibly we're just not being smart
                # enough, so let's pass things on as is.
                pass
            else:
                # If the selected format function is that of numpy, we know
                # things will fail
                if 'numpy' in ff.__module__:
                    a = a.value

    return (a,) + args, kwargs, None, None