Python numpy.dtype() Examples

def asnumpy(self):
        """Returns a ``numpy.ndarray`` object with value copied from this array.

        Examples
        --------
        >>> x = mx.nd.ones((2,3))
        >>> y = x.asnumpy()
        >>> type(y)
        <type 'numpy.ndarray'>
        >>> y
        array([[ 1.,  1.,  1.],
               [ 1.,  1.,  1.]], dtype=float32)
        >>> z = mx.nd.ones((2,3), dtype='int32')
        >>> z.asnumpy()
        array([[1, 1, 1],
               [1, 1, 1]], dtype=int32)
        """
        data = np.empty(self.shape, dtype=self.dtype)
        check_call(_LIB.MXNDArraySyncCopyToCPU(
            self.handle,
            data.ctypes.data_as(ctypes.c_void_p),
            ctypes.c_size_t(data.size)))
        return data 

def serialize_ndarray_b64(o):
    """
    Serializes a :obj:`numpy.ndarray` in a format where the datatype and shape are
    human-readable, but the array data itself is binary64 encoded.

    Args:
        o (:obj:`numpy.ndarray`): :obj:`ndarray` to be serialized.

    Returns:
        A dictionary that can be passed to :obj:`json.dumps`.
    """
    if o.flags['C_CONTIGUOUS']:
        o_data = o.data
    else:
        o_data = np.ascontiguousarray(o).data
    data_b64 = base64.b64encode(o_data)
    return dict(
        _type='np.ndarray',
        data=data_b64.decode('utf-8'),
        dtype=o.dtype,
        shape=o.shape) 

def asscipy(self):
        """Returns a ``scipy.sparse.csr.csr_matrix`` object with value copied from this array

        Examples
        --------
        >>> x = mx.nd.sparse.zeros('csr', (2,3))
        >>> y = x.asscipy()
        >>> type(y)
        <type 'scipy.sparse.csr.csr_matrix'>
        >>> y
        <2x3 sparse matrix of type '<type 'numpy.float32'>'
        with 0 stored elements in Compressed Sparse Row format>
        """
        data = self.data.asnumpy()
        indices = self.indices.asnumpy()
        indptr = self.indptr.asnumpy()
        if not spsp:
            raise ImportError("scipy is not available. \
                               Please check if the scipy python bindings are installed.")
        return spsp.csr_matrix((data, indices, indptr), shape=self.shape, dtype=self.dtype)

# pylint: disable=abstract-method 

def __init__(self, model, dtypestr='float32'):
        """
        :param model: An instance of the cleverhans.model.Model class.
        :param back: The backend to use. Inherited from AttackBase class.
        :param dtypestr: datatype of the input data samples and crafted
                        adversarial attacks.
        """
        # Validate the input arguments.
        if dtypestr != 'float32' and dtypestr != 'float64':
            raise ValueError("Unexpected input for argument dtypestr.")
        import tensorflow as tf
        tfe = tf.contrib.eager
        self.tf_dtype = tf.as_dtype(dtypestr)
        self.np_dtype = np.dtype(dtypestr)

        if not isinstance(model, Model):
            raise ValueError("The model argument should be an instance of"
                             " the cleverhans.model.Model class.")
        # Prepare attributes
        self.model = model
        self.dtypestr = dtypestr 

def _new_alloc_handle(shape, ctx, delay_alloc, dtype=mx_real_t):
    """Return a new handle with specified shape and context.

    Empty handle is only used to hold results.

    Returns
    -------
    handle
        A new empty `NDArray` handle.
    """
    hdl = NDArrayHandle()
    check_call(_LIB.MXNDArrayCreateEx(
        c_array_buf(mx_uint, native_array('I', shape)),
        mx_uint(len(shape)),
        ctypes.c_int(ctx.device_typeid),
        ctypes.c_int(ctx.device_id),
        ctypes.c_int(int(delay_alloc)),
        ctypes.c_int(int(_DTYPE_NP_TO_MX[np.dtype(dtype).type])),
        ctypes.byref(hdl)))
    return hdl 

def _prepare_value_nd(self, value, vshape):
        """Given value and vshape, create an `NDArray` from value with the same
        context and dtype as the current one and broadcast it to vshape."""
        if isinstance(value, numeric_types):
            value_nd = full(shape=vshape, val=value, ctx=self.context, dtype=self.dtype)
        elif isinstance(value, NDArray):
            value_nd = value.as_in_context(self.context)
            if value_nd.dtype != self.dtype:
                value_nd = value_nd.astype(self.dtype)
        else:
            try:
                value_nd = array(value, ctx=self.context, dtype=self.dtype)
            except:
                raise TypeError('NDArray does not support assignment with non-array-like'
                                ' object %s of type %s' % (str(value), str(type(value))))
        if value_nd.shape != vshape:
            value_nd = value_nd.broadcast_to(vshape)
        return value_nd 

def astype(self, dtype, copy=True):
        """Returns a copy of the array after casting to a specified type.
        Parameters
        ----------
        dtype : numpy.dtype or str
            The type of the returned array.
        copy : bool
            Default `True`. By default, astype always returns a newly
            allocated ndarray on the same context. If this is set to
            `False`, and the dtype requested is the same as the ndarray's
            dtype, the ndarray is returned instead of a copy.
        Examples
        --------
        >>> x = mx.nd.sparse.zeros('row_sparse', (2,3), dtype='float32')
        >>> y = x.astype('int32')
        >>> y.dtype
        <type 'numpy.int32'>
        """
        if not copy and np.dtype(dtype) == self.dtype:
            return self

        res = zeros(shape=self.shape, ctx=self.context,
                    dtype=dtype, stype=self.stype)
        self.copyto(res)
        return res 

def dtype(self):
        """Data-type of the array's elements.

        Returns
        -------
        numpy.dtype
            This NDArray's data type.

        Examples
        --------
        >>> x = mx.nd.zeros((2,3))
        >>> x.dtype
        <type 'numpy.float32'>
        >>> y = mx.nd.zeros((2,3), dtype='int32')
        >>> y.dtype
        <type 'numpy.int32'>
        """
        mx_dtype = ctypes.c_int()
        check_call(_LIB.MXNDArrayGetDType(
            self.handle, ctypes.byref(mx_dtype)))
        return _DTYPE_MX_TO_NP[mx_dtype.value] 

def deserialize_ndarray(d):
    """
    Deserializes a JSONified :obj:`numpy.ndarray`. Can handle arrays serialized
    using any of the methods in this module: :obj:`"npy"`, :obj:`"b64"`,
    :obj:`"readable"`.

    Args:
        d (`dict`): A dictionary representation of an :obj:`ndarray` object.

    Returns:
        An :obj:`ndarray` object.
    """
    if 'data' in d:
        x = np.fromstring(
            base64.b64decode(d['data']),
            dtype=d['dtype'])
        x.shape = d['shape']
        return x
    elif 'value' in d:
        return np.array(d['value'], dtype=d['dtype'])
    elif 'npy' in d:
        return deserialize_ndarray_npy(d)
    else:
        raise ValueError('Malformed np.ndarray encoding.') 

def T(self):
        """Returns a copy of the array with axes transposed.

        Equivalent to ``mx.nd.transpose(self)`` except that
        self is returned if ``self.ndim < 2``.

        Unlike ``numpy.ndarray.T``, this function returns a copy
        rather than a view of the array unless ``self.ndim < 2``.

        Examples
        --------
        >>> x = mx.nd.arange(0,6).reshape((2,3))
        >>> x.asnumpy()
        array([[ 0.,  1.,  2.],
               [ 3.,  4.,  5.]], dtype=float32)
        >>> x.T.asnumpy()
        array([[ 0.,  3.],
               [ 1.,  4.],
               [ 2.,  5.]], dtype=float32)

        """
        if len(self.shape) < 2:
            return self
        return op.transpose(self)
    # pylint: enable= invalid-name, undefined-variable 

def deserialize_dtype(d):
    """
    Deserializes a JSONified :obj:`numpy.dtype`.

    Args:
        d (:obj:`dict`): A dictionary representation of a :obj:`dtype` object.

    Returns:
        A :obj:`dtype` object.
    """
    if isinstance(d['descr'], six.string_types):
        return np.dtype(d['descr'])
    descr = []
    for col in d['descr']:
        col_descr = []
        for c in col:
            if isinstance(c, six.string_types):
                col_descr.append(str(c))
            elif type(c) is list:
                col_descr.append(tuple(c))
            else:
                col_descr.append(c)
        descr.append(tuple(col_descr))
    return np.dtype(descr) 

def serialize_dtype(o):
    """
    Serializes a :obj:`numpy.dtype`.

    Args:
        o (:obj:`numpy.dtype`): :obj:`dtype` to be serialized.

    Returns:
        A dictionary that can be passed to :obj:`json.dumps`.
    """
    if len(o) == 0:
        return dict(
            _type='np.dtype',
            descr=str(o))
    return dict(
        _type='np.dtype',
        descr=o.descr)
    # res = []
    # for k in range(len(o)):
    #     res.append((o.names[k], str(o[k])))
    # return dict(
    #     _type='np.dtype',
    #     desc=res) 

def __init__(self, resolution=1024, num_channels=3, dtype='uint8', dynamic_range=[0,255], label_size=0, label_dtype='float32'):
        self.resolution         = resolution
        self.resolution_log2    = int(np.log2(resolution))
        self.shape              = [num_channels, resolution, resolution]
        self.dtype              = dtype
        self.dynamic_range      = dynamic_range
        self.label_size         = label_size
        self.label_dtype        = label_dtype
        self._tf_minibatch_var  = None
        self._tf_lod_var        = None
        self._tf_minibatch_np   = None
        self._tf_labels_np      = None

        assert self.resolution == 2 ** self.resolution_log2
        with tf.name_scope('Dataset'):
            self._tf_minibatch_var = tf.Variable(np.int32(0), name='minibatch_var')
            self._tf_lod_var = tf.Variable(np.int32(0), name='lod_var') 

def asscalar(self):
        """Returns a scalar whose value is copied from this array.

        This function is equivalent to ``self.asnumpy()[0]``. This NDArray must have shape (1,).

        Examples
        --------
        >>> x = mx.nd.ones((1,), dtype='int32')
        >>> x.asscalar()
        1
        >>> type(x.asscalar())
        <type 'numpy.int32'>
        """
        if self.shape != (1,):
            raise ValueError("The current array is not a scalar")
        return self.asnumpy()[0] 

def copy(self):
        """Makes a copy of this ``NDArray``, keeping the same context.

        Returns
        -------
        NDArray, CSRNDArray or RowSparseNDArray
            The copied array

        Examples
        --------
        >>> x = mx.nd.ones((2,3))
        >>> y = x.copy()
        >>> y.asnumpy()
        array([[ 1.,  1.,  1.],
               [ 1.,  1.,  1.]], dtype=float32)
        """
        return self.copyto(self.context) 

def __init__(self, server):
        super().__init__()
        self.setDaemon(True)
        self._stop = threading.Event()
        self._reading = thread_lock()
        self.dt = np.dtype(np.uint16)
        self.dt = self.dt.newbyteorder('<')
        self._data = [np.zeros(WS_FRAME_SIZE, self.dt),
                      np.zeros(WS_FRAME_SIZE, self.dt)]
        self._buf = False
        self.ws = create_connection("ws://{}/".format(server)) 

def image_reslice(image, spec, method=None, fill=0, dtype=None, weights=None, image_type=None):
    '''
    image_reslice(image, spec) yields a duplicate of the given image resliced to have the voxels
      indicated by the given image spec. Note that spec may be an image itself.

    Optional arguments that can be passed to image_interpolate() (asside from affine) are allowed
    here and are passed through.
    '''
    if image_type is None and is_image(image): image_type = to_image_type(image)
    spec = to_image_spec(spec)
    image = to_image(image)
    # we make a big mesh and interpolate at these points...
    imsh = spec['image_shape']
    (args, kw) = ([np.arange(n) for n in imsh[:3]], {'indexing': 'ij'})
    ijk = np.asarray([u.flatten() for u in np.meshgrid(*args, **kw)])
    ijk = np.dot(spec['affine'], np.vstack([ijk, np.ones([1,ijk.shape[1]])]))[:3]
    # interpolate here...
    u = image_interpolate(image, ijk, method=method, fill=fill, dtype=dtype, weights=weights)
    return to_image((np.reshape(u, imsh), spec), image_type=image_type) 

def finto(x, ii, n, null=0):
    '''
    finto(x,ii,n) yields a vector u of length n such that u[ii] = x.

    Notes:
      * The ii index may be a tuple (as can be passed to numpy arrays' getitem method) in order to
        specify that the specific elements of a multidimensional output be set. In this case, the
        argument n should be a tuple of sizes (a single integer is taken to be a square/cube/etc).
      * x may be a sparse-array, but in it will be reified by this function.

    The following optional arguments are allowed:
      * null (defaut: 0) specifies the value that should appear in the elements of u that are not
        set.
    '''
    x  = x.toarray() if sps.issparse(x) else np.asarray(x)
    shx = x.shape
    if isinstance(ii, tuple):
        if not pimms.is_vector(n): n = tuple([n for _ in ii])
        if len(n) != len(ii): raise ValueError('%d-dim index but %d-dim output' % (len(ii),len(n)))
        sh = n + shx[1:]
    elif pimms.is_int(ii): sh = (n,) + shx
    else:                  sh = (n,) + shx[1:]
    u = np.zeros(sh, dtype=x.dtype) if null == 0 else np.full(sh, null, dtype=x.dtype)
    u[ii] = x
    return u

# Potential Functions ############################################################################## 

def _at(self, idx):
        """Returns a view of the array sliced at `idx` in the first dim.
        This is called through ``x[idx]``.

        Parameters
        ----------
        idx : int
            index for slicing the `NDArray` in the first dim.

        Returns
        -------
        NDArray
            `NDArray` sharing the memory with the current one sliced at `idx` in the first dim.

        Examples
        --------
        >>> a = mx.nd.array([[1,2], [3, 4]])
        >>> a[1].asnumpy()
        array([ 3.,  4.], dtype=float32)
        >>> b = mx.nd.array([1, 2, 3, 4])
        >>> b[0].asnumpy()
        array([ 1.], dtype=float32)
        """
        handle = NDArrayHandle()
        if idx < 0:
            length = self.shape[0]
            idx += length
            if idx < 0:
                raise IndexError('index %d is out of bounds for axis 0 with size %d'
                                 % (idx-length, length))
        check_call(_LIB.MXNDArrayAt(
            self.handle, mx_uint(idx), ctypes.byref(handle)))
        return NDArray(handle=handle, writable=self.writable) 

def ones(shape, ctx=None, dtype=None, **kwargs):
    """Returns a new array filled with all ones, with the given shape and type.

    Parameters
    ----------
    shape : int or tuple of int or list of int
        The shape of the empty array.
    ctx : Context, optional
        An optional device context.
        Defaults to the current default context (``mxnet.context.current_context()``).
    dtype : str or numpy.dtype, optional
        An optional value type (default is `float32`).
    out : NDArray, optional
        The output NDArray (default is `None`).

    Returns
    -------
    NDArray
        A new array of the specified shape filled with all ones.

    Examples
    --------
    >>> mx.nd.ones(1).asnumpy()
    array([ 1.], dtype=float32)
    >>> mx.nd.ones((1,2), mx.gpu(0))
    <NDArray 1x2 @gpu(0)>
    >>> mx.nd.ones((1,2), dtype='float16').asnumpy()
    array([[ 1.,  1.]], dtype=float16)
    """
    # pylint: disable= unused-argument
    if ctx is None:
        ctx = current_context()
    dtype = mx_real_t if dtype is None else dtype
    # pylint: disable= no-member, protected-access
    return _internal._ones(shape=shape, ctx=ctx, dtype=dtype, **kwargs)
    # pylint: enable= no-member, protected-access 

def _prepare_src_array(source_array, dtype):
    """Prepare `source_array` so that it can be used to construct NDArray.
    `source_array` is converted to a `np.ndarray` if it's neither an `NDArray` \
    nor an `np.ndarray`.
    """
    if not isinstance(source_array, NDArray) and not isinstance(source_array, np.ndarray):
        try:
            source_array = np.array(source_array, dtype=dtype)
        except:
            raise TypeError('values must be array like object')
    return source_array 

def broadcast_like(self, other):
        """Broadcasts the input array to the shape of other.

        Broadcasting is only allowed on axes with size 1. The new shape cannot change
        the number of dimensions.
        For example, you could broadcast from shape (2, 1) to (2, 3), but not from
        shape (2, 3) to (2, 3, 3).

        Parameters
        ----------
        other : NDArray
            Array with shape of the desired array.

        Returns
        -------
        NDArray
            A NDArray with the desired shape that is not sharing data with this
            array, even if the new shape is the same as ``self.shape``.

        Examples
        --------
        >>> x = mx.nd.arange(0,3).reshape((1,3,1))
        >>> x.asnumpy()
        array([[[ 0.],
                [ 1.],
                [ 2.]]], dtype=float32)
        >>> y = x.broadcast_like(mx.nd.ones((2,3,3)))
        >>> y.asnumpy()
        array([[[ 0.,  0.,  0.],
                [ 1.,  1.,  1.],
                [ 2.,  2.,  2.]],
        <BLANKLINE>
               [[ 0.,  0.,  0.],
                [ 1.,  1.,  1.],
                [ 2.,  2.,  2.]]], dtype=float32)
        """
        return self.broadcast_to(other.shape) 

def _slice(self, start, stop):
        """Returns a sliced NDArray that shares memory with the current one.
        This is called through ``x[start:stop]``.

        Parameters
        ----------
        start : int
            Starting inclusive index of slice in the first dim.
        stop : int
            Finishing exclusive index of slice in the first dim.

        Returns
        -------
            `NDArray` sharing the memory with the current one sliced from
            start to stop in the first dim.

        Examples:
        >>> a = mx.nd.array([[1,2], [3, 4], [5, 6], [7, 8]])
        >>> a[1:2].asnumpy()
        array([[ 3.,  4.]], dtype=float32)
        >>> a[1:1].asnumpy()
        array([], shape=(0, 2), dtype=float32)
        """
        handle = NDArrayHandle()
        start, stop, _ = _get_index_range(start, stop, self.shape[0])

        check_call(_LIB.MXNDArraySlice(
            self.handle, mx_uint(start), mx_uint(stop), ctypes.byref(handle)))
        return NDArray(handle=handle, writable=self.writable) 

def __getitem__(self, key):
        """x.__getitem__(i) <=> x[i]

        Returns a sliced view of this array.

        Parameters
        ----------
        key : slice
            Indexing key.

        Examples
        --------
        >>> x = mx.nd.sparse.zeros('row_sparse', (2, 3))
        >>> x[:].asnumpy()
        array([[ 0.,  0.,  0.],
               [ 0.,  0.,  0.]], dtype=float32)
        """
        if isinstance(key, int):
            raise Exception("__getitem__ with int key is not implemented for RowSparseNDArray yet")
        if isinstance(key, py_slice):
            if key.step is not None or key.start is not None or key.stop is not None:
                raise Exception('RowSparseNDArray only supports [:] for __getitem__')
            else:
                return self
        if isinstance(key, tuple):
            raise ValueError('Multi-dimension indexing is not supported')
        raise ValueError('Undefined behaviour for {}'.format(key)) 

def copyto(self, other):
        """Copies the value of this array to another array.

        Parameters
        ----------
        other : NDArray or CSRNDArray or RowSparseNDArray or Context
            The destination array or context.

        Returns
        -------
        NDArray or CSRNDArray or RowSparseNDArray
            The copied array.
        """
        # pylint: disable= no-member, protected-access
        if isinstance(other, NDArray):
            if other.handle is self.handle:
                warnings.warn('You are attempting to copy an array to itself', RuntimeWarning)
                return False
            return _internal._copyto(self, out=other)
        elif isinstance(other, Context):
            hret = _ndarray_cls(_new_alloc_handle(self.stype, self.shape, other,
                                                  True, self.dtype, self._aux_types))
            return _internal._copyto(self, out=hret)
        else:
            raise TypeError('copyto does not support type ' + str(type(other)))
        # pylint: enable= no-member, protected-access 

def _aux_type(self, i):
        """Data-type of the array's ith aux data.

        Returns
        -------
        numpy.dtype
            This BaseSparseNDArray's aux data type.
        """
        aux_type = ctypes.c_int()
        check_call(_LIB.MXNDArrayGetAuxType(self.handle, i, ctypes.byref(aux_type)))
        return _DTYPE_MX_TO_NP[aux_type.value] 

def _new_alloc_handle(stype, shape, ctx, delay_alloc, dtype, aux_types, aux_shapes=None):
    """Return a new handle with specified storage type, shape, dtype and context.

    Empty handle is only used to hold results

    Returns
    -------
    handle
        A new empty ndarray handle
    """
    hdl = NDArrayHandle()
    for aux_t in aux_types:
        if np.dtype(aux_t) != np.dtype("int64"):
            raise NotImplementedError("only int64 is supported for aux types")
    aux_type_ids = [int(_DTYPE_NP_TO_MX[np.dtype(aux_t).type]) for aux_t in aux_types]
    aux_shapes = [(0,) for aux_t in aux_types] if aux_shapes is None else aux_shapes
    aux_shape_lens = [len(aux_shape) for aux_shape in aux_shapes]
    aux_shapes = py_sum(aux_shapes, ())
    num_aux = mx_uint(len(aux_types))
    check_call(_LIB.MXNDArrayCreateSparseEx(
        ctypes.c_int(int(_STORAGE_TYPE_STR_TO_ID[stype])),
        c_array_buf(mx_uint, native_array('I', shape)),
        mx_uint(len(shape)),
        ctypes.c_int(ctx.device_typeid),
        ctypes.c_int(ctx.device_id),
        ctypes.c_int(int(delay_alloc)),
        ctypes.c_int(int(_DTYPE_NP_TO_MX[np.dtype(dtype).type])),
        num_aux,
        c_array_buf(ctypes.c_int, native_array('i', aux_type_ids)),
        c_array_buf(mx_uint, native_array('I', aux_shape_lens)),
        c_array_buf(mx_uint, native_array('I', aux_shapes)),
        ctypes.byref(hdl)))
    return hdl 

def _sync_copyfrom(self, source_array):
        """Performs a synchronized copy from the `source_array` to the current array.
        This is called through ``x[:] = source_array``, where the `source_array`
        is a `numpy.ndarray` or array-like object.
        This function blocks until all the pending read/write operations with respect
        to the current `NDArray` are finished and carry out the copy operation to the
        current NDArray.

        Parameters
        ----------
        source_array : array_like
            The data source we would like to copy from.

        Example
        -------
        >>> a = mx.nd.array([1, 2])
        >>> a.asnumpy()
        array([ 1.,  2.], dtype=float32)
        >>> a[:] = np.array([3, 4])
        >> a.asnumpy()
        array([ 3.,  4.], dtype=float32)
        """
        if not isinstance(source_array, np.ndarray):
            try:
                source_array = np.array(source_array, dtype=self.dtype)
            except:
                raise TypeError('array must consist of array-like data,' +
                                'type %s is not supported' % str(type(array)))
        source_array = np.asarray(source_array, dtype=self.dtype, order='C')
        if source_array.shape != self.shape:
            raise ValueError('Shape inconsistent: expected %s vs got %s'%(
                str(source_array.shape), str(self.shape)))
        check_call(_LIB.MXNDArraySyncCopyFromCPU(
            self.handle,
            source_array.ctypes.data_as(ctypes.c_void_p),
            ctypes.c_size_t(source_array.size))) 

def image_spec_to_image(imspec, image_type=None, fill=0):
    '''
    image_spec_to_image(imspec) yields an empty with the given image-spec.
    image_spec_to_image(imspec, image_type) creates an image with the given type
    image_spec_to_image(imspec, image_type, fill) fills the resulting image with the given
      fill-value.
    '''
    imsh = imspec_lookup(imspec, 'image_shape')
    if fill == 0: imarr = np.zeros(imsh, dtype=imspec_lookup(imspec, 'voxel_type'))
    else:         imarr = np.full(imsh, fill, dtype=imspec_lookup(imspec, 'voxel_type'))
    # okay, we have the image array...
    image_type = to_image_type('nifti1' if image_type is None else image_type)
    return image_type.create(imarr, meta_data=imspec) 

def parse_type(self, hdat, dataobj=None):
        dtype = super(MGHImageType, self).parse_type(hdat, dataobj=dataobj)
        if   np.issubdtype(dtype, np.floating): dtype = np.float32
        elif np.issubdtype(dtype, np.int8):     dtype = np.int8
        elif np.issubdtype(dtype, np.int16):    dtype = np.int16
        elif np.issubdtype(dtype, np.integer):  dtype = np.int32
        else: raise ValueError('Could not deduce appropriate MGH type for dtype %s' % dtype)
        return dtype