# tag: numpy
# cannot be named "numpy" in order to not clash with the numpy module!

cimport numpy as np
cimport cython

import re
import sys


def little_endian():
    cdef int endian_detector = 1
    return (<char*>&endian_detector)[0] != 0

__test__ = {}

def testcase(f):
    __test__[f.__name__] = f.__doc__
    return f

def testcase_have_buffer_interface(f):
    major, minor, *rest = np.__version__.split('.')
    if (int(major), int(minor)) >= (1, 5) and sys.version_info[:2] >= (2, 6):
        __test__[f.__name__] = f.__doc__
    return f

if little_endian():
    my_endian = '<'
    other_endian = '>'
else:
    my_endian = '>'
    other_endian = '<'

try:
    import numpy as np
    __doc__ = u"""

    >>> assert_dtype_sizes()

    >>> basic()
    [[0 1 2 3 4]
     [5 6 7 8 9]]
    2 0 9 5

    >>> three_dim()  # doctest: +NORMALIZE_WHITESPACE
    [[[0.   1.   2.   3.]
      [4.   5.   6.   7.]]
    <BLANKLINE>
     [[8.   9.  10.  11.]
      [12.  13.  14.  15.]]
    <BLANKLINE>
     [[16.  17.  18.  19.]
      [20.  21.  22.  23.]]]
    6.0 0.0 13.0 8.0

    >>> obj_array()
    [a 1 {}]
    a 1 {}

    Test various forms of slicing, picking etc.
    >>> a = np.arange(10, dtype='l').reshape(2, 5)
    >>> print_long_2d(a)
    0 1 2 3 4
    5 6 7 8 9
    >>> print_long_2d(a[::-1, ::-1])
    9 8 7 6 5
    4 3 2 1 0
    >>> print_long_2d(a[1:2, 1:3])
    6 7
    >>> print_long_2d(a[::2, ::2])
    0 2 4
    >>> print_long_2d(a[::4, :])
    0 1 2 3 4
    >>> print_long_2d(a[:, 1:5:2])
    1 3
    6 8
    >>> print_long_2d(a[:, 5:1:-2])
    4 2
    9 7
    >>> print_long_2d(a[:, [3, 1]])
    3 1
    8 6
    >>> print_long_2d(a.T)
    0 5
    1 6
    2 7
    3 8
    4 9

    Write to slices
    >>> b = a.copy()
    >>> put_range_long_1d(b[:, 3])
    >>> print (b)
    [[0 1 2 0 4]
     [5 6 7 1 9]]
    >>> put_range_long_1d(b[::-1, 3])
    >>> print (b)
    [[0 1 2 1 4]
     [5 6 7 0 9]]
    >>> a = np.zeros(9, dtype='l')
    >>> put_range_long_1d(a[1::3])
    >>> print (a)
    [0 0 0 0 1 0 0 2 0]

    Write to picked subarrays. This should NOT change the original
    array as picking creates a new mutable copy.
    >>> a = np.zeros(10, dtype='l').reshape(2, 5)
    >>> put_range_long_1d(a[[0, 0, 1, 1, 0], [0, 1, 2, 4, 3]])
    >>> print (a)
    [[0 0 0 0 0]
     [0 0 0 0 0]]

    Test contiguous access modes:
    >>> c_arr = np.array(np.arange(12, dtype='i').reshape(3,4), order='C')
    >>> f_arr = np.array(np.arange(12, dtype='i').reshape(3,4), order='F')
    >>> test_c_contig(c_arr)
    0 1 2 3
    4 5 6 7
    8 9 10 11
    >>> test_f_contig(f_arr)
    0 1 2 3
    4 5 6 7
    8 9 10 11
    >>> test_c_contig(f_arr) #doctest: +ELLIPSIS
    Traceback (most recent call last):
       ...
    ValueError: ndarray is not C...contiguous
    >>> test_f_contig(c_arr) #doctest: +ELLIPSIS
    Traceback (most recent call last):
       ...
    ValueError: ndarray is not Fortran contiguous
    >>> test_c_contig(c_arr[::2,::2]) #doctest: +ELLIPSIS
    Traceback (most recent call last):
       ...
    ValueError: ndarray is not C...contiguous

    >>> test_dtype('b', inc1_byte)
    >>> test_dtype('B', inc1_ubyte)
    >>> test_dtype('h', inc1_short)
    >>> test_dtype('H', inc1_ushort)
    >>> test_dtype('i', inc1_int)
    >>> test_dtype('I', inc1_uint)
    >>> test_dtype('l', inc1_long)
    >>> test_dtype('L', inc1_ulong)

    >>> test_dtype('f', inc1_float)
    >>> test_dtype('d', inc1_double)
    >>> test_dtype('g', inc1_longdouble)
    >>> test_dtype('O', inc1_object)
    >>> test_dtype('F', inc1_cfloat) # numpy format codes differ from buffer ones here
    >>> test_dtype('D', inc1_cdouble)
    >>> test_dtype('G', inc1_clongdouble)
    >>> test_dtype('F', inc1_cfloat_struct)
    >>> test_dtype('D', inc1_cdouble_struct)
    >>> test_dtype('G', inc1_clongdouble_struct)

    >>> test_dtype(np.int, inc1_int_t)
    >>> test_dtype(np.longlong, inc1_longlong_t)
    >>> test_dtype(np.float, inc1_float_t)
    >>> test_dtype(np.double, inc1_double_t)
    >>> test_dtype(np.intp, inc1_intp_t)
    >>> test_dtype(np.uintp, inc1_uintp_t)

    >>> test_dtype(np.longdouble, inc1_longdouble_t)

    >>> test_dtype(np.int32, inc1_int32_t)
    >>> test_dtype(np.float64, inc1_float64_t)

    Endian tests:
    >>> test_dtype('%si' % my_endian, inc1_int)
    >>> test_dtype('%si' % other_endian, inc1_int)  #doctest: +ELLIPSIS
    Traceback (most recent call last):
       ...
    ValueError: ...



    >>> test_recordarray()

    >>> print(test_nested_dtypes(np.zeros((3,), dtype=np.dtype([\
            ('a', np.dtype('i,i')),\
            ('b', np.dtype('i,i'))\
        ]))))                              # doctest: +NORMALIZE_WHITESPACE
    array([((0, 0), (0, 0)), ((1, 2), (1, 4)), ((1, 2), (1, 4))], 
          dtype=[('a', [('f0', '!i4'), ('f1', '!i4')]), ('b', [('f0', '!i4'), ('f1', '!i4')])])

    >>> print(test_nested_dtypes(np.zeros((3,), dtype=np.dtype([\
            ('a', np.dtype('i,f')),\
            ('b', np.dtype('i,i'))\
        ]))))
    Traceback (most recent call last):
        ...
    ValueError: Buffer dtype mismatch, expected 'int' but got 'float' in 'DoubleInt.y'

    >>> print(test_packed_align(np.zeros((1,), dtype=np.dtype('b,i', align=False))))
    [(22, 23)]


    The output changed in Python 3:
    >> print(test_unpacked_align(np.zeros((1,), dtype=np.dtype('b,i', align=True))))
    array([(22, 23)],
          dtype=[('f0', '|i1'), ('', '|V3'), ('f1', '!i4')])

    ->

    array([(22, 23)],
          dtype={'names':['f0','f1'], 'formats':['i1','!i4'], 'offsets':[0,4], 'itemsize':8, 'aligned':True})


    >>> print(test_unpacked_align(np.zeros((1,), dtype=np.dtype('b,i', align=True))))
    [(22, 23)]

    >>> print(test_packed_align(np.zeros((1,), dtype=np.dtype('b,i', align=True)))) #doctest: +ELLIPSIS
    Traceback (most recent call last):
        ...
    ValueError: ...

    >>> print(test_unpacked_align(np.zeros((1,), dtype=np.dtype('b,i', align=False)))) #doctest: +ELLIPSIS
    Traceback (most recent call last):
        ...
    ValueError: ...


    >>> test_good_cast()
    True
    >>> test_bad_cast()
    Traceback (most recent call last):
        ...
    ValueError: Item size of buffer (1 byte) does not match size of 'int' (4 bytes)

    >>> test_complextypes()
    1,1
    1,1
    8,16

    >>> test_point_record()         # doctest: +NORMALIZE_WHITESPACE
    array([(0., 0.), (1., -1.), (2., -2.)], 
          dtype=[('x', '!f8'), ('y', '!f8')])

"""

    if np.__version__ >= '1.6' and False:
        __doc__ += u"""
        Tests are DISABLED as the buffer format parser does not align members
        of aligned structs in padded structs in relation to the possibly
        unaligned initial offset.

        The following expose bugs in Numpy (versions prior to 2011-04-02):
        >>> print(test_partially_packed_align(np.zeros((1,), dtype=np.dtype([('a', 'b'), ('b', 'i'), ('sub', np.dtype('b,i')), ('c', 'i')], align=True))))
        array([(22, 23, (24, 25), 26)],
              dtype=[('a', '|i1'), ('', '|V3'), ('b', '!i4'), ('sub', [('f0', '|i1'), ('f1', '!i4')]), ('', '|V3'), ('c', '!i4')])

        >>> print(test_partially_packed_align_2(np.zeros((1,), dtype=np.dtype([('a', 'b'), ('b', 'i'), ('c', 'b'), ('sub', np.dtype('b,i', align=True))]))))
        array([(22, 23, 24, (27, 28))],
              dtype=[('a', '|i1'), ('b', '!i4'), ('c', '|i1'), ('sub', [('f0', '|i1'), ('', '|V3'), ('f1', '!i4')])])

        >>> print(test_partially_packed_align(np.zeros((1,), dtype=np.dtype([('a', 'b'), ('b', 'i'), ('sub', np.dtype('b,i')), ('c', 'i')], align=False)))) #doctest: +ELLIPSIS
        Traceback (most recent call last):
            ...
        ValueError: ...

        >>> print(test_partially_packed_align_2(np.zeros((1,), dtype=np.dtype([('a', 'b'), ('b', 'i'), ('c', 'b'), ('sub', np.dtype('b,i', align=False))])))) #doctest: +ELLIPSIS
        Traceback (most recent call last):
            ...
        ValueError: ...
        """

except:
    __doc__ = u""

__test__[__name__] = __doc__

def assert_dtype_sizes():
    assert sizeof(np.int8_t) == 1
    assert sizeof(np.int16_t) == 2
    assert sizeof(np.int32_t) == 4
    assert sizeof(np.int64_t) == 8
    assert sizeof(np.uint8_t) == 1
    assert sizeof(np.uint16_t) == 2
    assert sizeof(np.uint32_t) == 4
    assert sizeof(np.uint64_t) == 8
    assert sizeof(np.float32_t) == 4
    assert sizeof(np.float64_t) == 8
    assert sizeof(np.complex64_t) == 8
    assert sizeof(np.complex128_t) == 16

def ndarray_str(arr):
    u"""
    Work around display differences in NumPy 1.14.
    """
    return re.sub(ur'\[ +', '[', unicode(arr))

def basic():
    cdef object[int, ndim=2] buf = np.arange(10, dtype='i').reshape((2, 5))
    print buf
    print buf[0, 2], buf[0, 0], buf[1, 4], buf[1, 0]

def three_dim():
    cdef object[double, ndim=3] buf = np.arange(24, dtype='d').reshape((3,2,4))
    print ndarray_str(buf)
    print buf[0, 1, 2], buf[0, 0, 0], buf[1, 1, 1], buf[1, 0, 0]

def obj_array():
    cdef object[object, ndim=1] buf = np.array(["a", 1, {}])
    print str(buf).replace('"', '').replace("'", '')
    print buf[0], buf[1], buf[2]


def print_long_2d(np.ndarray[long, ndim=2] arr):
    cdef int i, j
    for i in range(arr.shape[0]):
        print u" ".join([unicode(arr[i, j]) for j in range(arr.shape[1])])

def put_range_long_1d(np.ndarray[long] arr):
    u"""Writes 0,1,2,... to array and returns array"""
    cdef int value = 0, i
    for i in range(arr.shape[0]):
        arr[i] = value
        value += 1

def test_c_contig(np.ndarray[int, ndim=2, mode='c'] arr):
    cdef int i, j
    for i in range(arr.shape[0]):
        print u" ".join([unicode(arr[i, j]) for j in range(arr.shape[1])])

def test_f_contig(np.ndarray[int, ndim=2, mode='fortran'] arr):
    cdef int i, j
    for i in range(arr.shape[0]):
        print u" ".join([unicode(arr[i, j]) for j in range(arr.shape[1])])

# Exhaustive dtype tests -- increments element [1] by 1 (or 1+1j) for all dtypes
def inc1_byte(np.ndarray[char] arr):                    arr[1] += 1
def inc1_ubyte(np.ndarray[unsigned char] arr):          arr[1] += 1
def inc1_short(np.ndarray[short] arr):                  arr[1] += 1
def inc1_ushort(np.ndarray[unsigned short] arr):        arr[1] += 1
def inc1_int(np.ndarray[int] arr):                      arr[1] += 1
def inc1_uint(np.ndarray[unsigned int] arr):            arr[1] += 1
def inc1_long(np.ndarray[long] arr):                    arr[1] += 1
def inc1_ulong(np.ndarray[unsigned long] arr):          arr[1] += 1
def inc1_longlong(np.ndarray[long long] arr):           arr[1] += 1
def inc1_ulonglong(np.ndarray[unsigned long long] arr): arr[1] += 1

def inc1_float(np.ndarray[float] arr):                  arr[1] += 1
def inc1_double(np.ndarray[double] arr):                arr[1] += 1
def inc1_longdouble(np.ndarray[long double] arr):       arr[1] += 1

def inc1_cfloat(np.ndarray[float complex] arr):            arr[1] = arr[1] + 1 + 1j
def inc1_cdouble(np.ndarray[double complex] arr):          arr[1] = (arr[1] + 1) + 1j
def inc1_clongdouble(np.ndarray[long double complex] arr): arr[1] = arr[1] + (1 + 1j)

def inc1_cfloat_struct(np.ndarray[np.cfloat_t] arr):
    arr[1].real += 1
    arr[1].imag += 1

def inc1_cdouble_struct(np.ndarray[np.cdouble_t] arr):
    arr[1].real += 1
    arr[1].imag += 1

def inc1_clongdouble_struct(np.ndarray[np.clongdouble_t] arr):
    cdef long double x
    x = arr[1].real + 1
    arr[1].real = x
    arr[1].imag = arr[1].imag + 1

def inc1_object(np.ndarray[object] arr):
    o = arr[1]
    o += 1
    arr[1] = o # unfortunately, += segfaults for objects


def inc1_int_t(np.ndarray[np.int_t] arr):               arr[1] += 1
def inc1_long_t(np.ndarray[np.long_t] arr):             arr[1] += 1
def inc1_longlong_t(np.ndarray[np.longlong_t] arr):     arr[1] += 1
def inc1_float_t(np.ndarray[np.float_t] arr):           arr[1] += 1
def inc1_double_t(np.ndarray[np.double_t] arr):         arr[1] += 1
def inc1_longdouble_t(np.ndarray[np.longdouble_t] arr): arr[1] += 1
def inc1_intp_t(np.ndarray[np.intp_t] arr):             arr[1] += 1
def inc1_uintp_t(np.ndarray[np.uintp_t] arr):           arr[1] += 1

# The tests below only work on platforms that has the given types
def inc1_int32_t(np.ndarray[np.int32_t] arr):           arr[1] += 1
def inc1_float64_t(np.ndarray[np.float64_t] arr):       arr[1] += 1


def test_dtype(dtype, inc1):
    if dtype in ("g", np.longdouble,
                 "G", np.clongdouble):
        if sizeof(double) == sizeof(long double): # MSVC
            return
    if dtype in ('F', 'D', 'G'):
        a = np.array([0, 10+10j], dtype=dtype)
        inc1(a)
        if a[1] != (11 + 11j): print u"failed!", a[1]
    else:
        a = np.array([0, 10], dtype=dtype)
        inc1(a)
        if a[1] != 11: print u"failed!"

cdef struct DoubleInt:
    int x, y

def test_recordarray():
    cdef object[DoubleInt] arr
    arr = np.array([(5,5), (4, 6)], dtype=np.dtype('i,i'))
    cdef DoubleInt rec
    rec = arr[0]
    if rec.x != 5: print u"failed"
    if rec.y != 5: print u"failed"
    rec.y += 5
    arr[1] = rec
    arr[0].x -= 2
    arr[0].y += 3
    if arr[0].x != 3: print u"failed"
    if arr[0].y != 8: print u"failed"
    if arr[1].x != 5: print u"failed"
    if arr[1].y != 10: print u"failed"

cdef struct NestedStruct:
    DoubleInt a
    DoubleInt b

cdef struct BadDoubleInt:
    float x
    int y

cdef struct BadNestedStruct:
    DoubleInt a
    BadDoubleInt b

def test_nested_dtypes(obj):
    cdef object[NestedStruct] arr = obj
    arr[1].a.x = 1
    arr[1].a.y = 2
    arr[1].b.x = arr[0].a.y + 1
    arr[1].b.y = 4
    arr[2] = arr[1]
    return repr(arr).replace('<', '!').replace('>', '!')

def test_bad_nested_dtypes():
    cdef object[BadNestedStruct] arr

def test_good_cast():
    # Check that a signed int can round-trip through casted unsigned int access
    cdef np.ndarray[unsigned int, cast=True] arr = np.array([-100], dtype='i')
    cdef unsigned int data = arr[0]
    return -100 == <int>data

def test_bad_cast():
    # This should raise an exception
    cdef np.ndarray[int, cast=True] arr = np.array([1], dtype='b')

cdef packed struct PackedStruct:
    char a
    int b

cdef struct UnpackedStruct:
    char a
    int b

cdef struct PartiallyPackedStruct:
    char a
    int b
    PackedStruct sub
    int c

cdef packed struct PartiallyPackedStruct2:
    char a
    int b
    char c
    UnpackedStruct sub

def test_packed_align(np.ndarray[PackedStruct] arr):
    arr[0].a = 22
    arr[0].b = 23
    return list(arr)

def test_unpacked_align(np.ndarray[UnpackedStruct] arr):
    arr[0].a = 22
    arr[0].b = 23
    # return repr(arr).replace('<', '!').replace('>', '!')
    return list(arr)

def test_partially_packed_align(np.ndarray[PartiallyPackedStruct] arr):
    arr[0].a = 22
    arr[0].b = 23
    arr[0].sub.a = 24
    arr[0].sub.b = 25
    arr[0].c = 26
    return repr(arr).replace('<', '!').replace('>', '!')

def test_partially_packed_align_2(np.ndarray[PartiallyPackedStruct2] arr):
    arr[0].a = 22
    arr[0].b = 23
    arr[0].c = 24
    arr[0].sub.a = 27
    arr[0].sub.b = 28
    return repr(arr).replace('<', '!').replace('>', '!')

def test_complextypes():
    cdef np.complex64_t x64 = 1, y64 = 1j
    cdef np.complex128_t x128 = 1, y128 = 1j
    x64 = x64 + y64
    print "%.0f,%.0f" % (x64.real, x64.imag)
    x128 = x128 + y128
    print "%.0f,%.0f" % (x128.real, x128.imag)
    print "%d,%d" % (sizeof(x64), sizeof(x128))


cdef struct Point:
    np.float64_t x, y

def test_point_record():
    cdef np.ndarray[Point] test
    Point_dtype = np.dtype([('x', np.float64), ('y', np.float64)])
    test = np.zeros(3, Point_dtype)
    cdef int i
    for i in range(3):
        test[i].x = i
        test[i].y = -i
    print re.sub(
        r'\.0+\b', '.', repr(test).replace('<', '!').replace('>', '!')
                                  .replace('( ', '(').replace(',  ', ', '))

# Test fused np.ndarray dtypes and runtime dispatch
@testcase
def test_fused_ndarray_floating_dtype(np.ndarray[cython.floating, ndim=1] a):
    """
    >>> import cython
    >>> sorted(test_fused_ndarray_floating_dtype.__signatures__)
    ['double', 'float']


    >>> test_fused_ndarray_floating_dtype[cython.double](np.arange(10, dtype=np.float64))
    ndarray[double,ndim=1] ndarray[double,ndim=1] 5.0 6.0
    >>> test_fused_ndarray_floating_dtype(np.arange(10, dtype=np.float64))
    ndarray[double,ndim=1] ndarray[double,ndim=1] 5.0 6.0

    >>> test_fused_ndarray_floating_dtype[cython.float](np.arange(10, dtype=np.float32))
    ndarray[float,ndim=1] ndarray[float,ndim=1] 5.0 6.0
    >>> test_fused_ndarray_floating_dtype(np.arange(10, dtype=np.float32))
    ndarray[float,ndim=1] ndarray[float,ndim=1] 5.0 6.0
    """
    cdef np.ndarray[cython.floating, ndim=1] b = a
    print cython.typeof(a), cython.typeof(b), a[5], b[6]


double_array = np.linspace(0, 1, 100)
int32_array = np.arange(100, dtype=np.int32)

cdef fused fused_external:
    np.int32_t
    np.int64_t
    np.float32_t
    np.float64_t

@testcase
def test_fused_external(np.ndarray[fused_external, ndim=1] a):
    """
    >>> import cython
    >>> sorted(test_fused_external.__signatures__)
    ['float32_t', 'float64_t', 'int32_t', 'int64_t']

    >>> test_fused_external["float64_t"](double_array)
    float64

    >>> test_fused_external["int32_t"](int32_array)
    int32

    >>> test_fused_external(np.arange(100, dtype=np.int64))
    int64
    """
    print a.dtype

cdef fused fused_buffers:
    np.ndarray[np.int32_t, ndim=1]
    np.int64_t[::1]

@testcase
def test_fused_buffers(fused_buffers arg):
    """
    >>> sorted(test_fused_buffers.__signatures__)
    ['int64_t[::1]', 'ndarray[int32_t,ndim=1]']
    """

cpdef _fused_cpdef_buffers(np.ndarray[fused_external] a):
    print a.dtype

@testcase
def test_fused_cpdef_buffers():
    """
    >>> test_fused_cpdef_buffers()
    int32
    int32
    """
    _fused_cpdef_buffers[np.int32_t](int32_array)

    cdef np.ndarray[np.int32_t] typed_array = int32_array
    _fused_cpdef_buffers(typed_array)

@testcase
def test_fused_ndarray_integral_dtype(np.ndarray[cython.integral, ndim=1] a):
    """
    >>> import cython
    >>> sorted(test_fused_ndarray_integral_dtype.__signatures__)
    ['int', 'long', 'short']

    >>> test_fused_ndarray_integral_dtype[cython.int](np.arange(10, dtype=np.dtype('i')))
    5 6
    >>> test_fused_ndarray_integral_dtype(np.arange(10, dtype=np.dtype('i')))
    5 6

    >>> test_fused_ndarray_integral_dtype[cython.long](np.arange(10, dtype='l'))
    5 6
    >>> test_fused_ndarray_integral_dtype(np.arange(10, dtype='l'))
    5 6
    """
    cdef np.ndarray[cython.integral, ndim=1] b = a
    # Don't print the types, the platform specific sizes can make the dispatcher
    # select different integer types with equal sizeof()
    print a[5], b[6]

cdef fused fused_dtype:
    float complex
    double complex
    object

@testcase
def test_fused_ndarray_other_dtypes(np.ndarray[fused_dtype, ndim=1] a):
    """
    >>> import cython
    >>> sorted(test_fused_ndarray_other_dtypes.__signatures__)
    ['double complex', 'float complex', 'object']
    >>> test_fused_ndarray_other_dtypes(np.arange(10, dtype=np.complex64))
    ndarray[float complex,ndim=1] ndarray[float complex,ndim=1] (5+0j) (6+0j)
    >>> test_fused_ndarray_other_dtypes(np.arange(10, dtype=np.complex128))
    ndarray[double complex,ndim=1] ndarray[double complex,ndim=1] (5+0j) (6+0j)
    >>> test_fused_ndarray_other_dtypes(np.arange(10, dtype=np.object))
    ndarray[Python object,ndim=1] ndarray[Python object,ndim=1] 5 6
    """
    cdef np.ndarray[fused_dtype, ndim=1] b = a
    print cython.typeof(a), cython.typeof(b), a[5], b[6]


# Test fusing the array types together and runtime dispatch
cdef struct Foo:
    int a
    float b

cdef fused fused_FooArray:
    np.ndarray[Foo, ndim=1]

cdef fused fused_ndarray:
    np.ndarray[float, ndim=1]
    np.ndarray[double, ndim=1]
    np.ndarray[Foo, ndim=1]

def get_Foo_array():
    cdef Foo data[10]
    for i in range(10):
        data[i] = [0, 0]
    data[5].b = 9.0
    return np.asarray(<Foo[:]>data).copy()

@testcase_have_buffer_interface
def test_fused_ndarray(fused_ndarray a):
    """
    >>> import cython
    >>> sorted(test_fused_ndarray.__signatures__)
    ['ndarray[Foo,ndim=1]', 'ndarray[double,ndim=1]', 'ndarray[float,ndim=1]']

    >>> test_fused_ndarray(get_Foo_array())
    ndarray[Foo,ndim=1] ndarray[Foo,ndim=1]
    9.0
    >>> test_fused_ndarray(np.arange(10, dtype=np.float64))
    ndarray[double,ndim=1] ndarray[double,ndim=1]
    5.0
    >>> test_fused_ndarray(np.arange(10, dtype=np.float32))
    ndarray[float,ndim=1] ndarray[float,ndim=1]
    5.0
    """
    cdef fused_ndarray b = a
    print cython.typeof(a), cython.typeof(b)

    if fused_ndarray in fused_FooArray:
        print b[5].b
    else:
        print b[5]

cpdef test_fused_cpdef_ndarray(fused_ndarray a):
    """
    >>> import cython
    >>> sorted(test_fused_cpdef_ndarray.__signatures__)
    ['ndarray[Foo,ndim=1]', 'ndarray[double,ndim=1]', 'ndarray[float,ndim=1]']

    >>> test_fused_cpdef_ndarray(get_Foo_array())
    ndarray[Foo,ndim=1] ndarray[Foo,ndim=1]
    9.0
    >>> test_fused_cpdef_ndarray(np.arange(10, dtype=np.float64))
    ndarray[double,ndim=1] ndarray[double,ndim=1]
    5.0
    >>> test_fused_cpdef_ndarray(np.arange(10, dtype=np.float32))
    ndarray[float,ndim=1] ndarray[float,ndim=1]
    5.0
    """
    cdef fused_ndarray b = a
    print cython.typeof(a), cython.typeof(b)

    if fused_ndarray in fused_FooArray:
        print b[5].b
    else:
        print b[5]

testcase_have_buffer_interface(test_fused_cpdef_ndarray)

@testcase_have_buffer_interface
def test_fused_cpdef_ndarray_cdef_call():
    """
    >>> test_fused_cpdef_ndarray_cdef_call()
    ndarray[Foo,ndim=1] ndarray[Foo,ndim=1]
    9.0
    """
    cdef np.ndarray[Foo, ndim=1] foo_array = get_Foo_array()
    test_fused_cpdef_ndarray(foo_array)

cdef fused int_type:
    np.int32_t
    np.int64_t

float64_array = np.arange(10, dtype=np.float64)
float32_array = np.arange(10, dtype=np.float32)
int32_array = np.arange(10, dtype=np.int32)
int64_array = np.arange(10, dtype=np.int64)

@testcase
def test_dispatch_non_clashing_declarations_repeating_types(np.ndarray[cython.floating] a1,
                                                            np.ndarray[int_type] a2,
                                                            np.ndarray[cython.floating] a3,
                                                            np.ndarray[int_type] a4):
    """
    >>> test_dispatch_non_clashing_declarations_repeating_types(float64_array, int32_array, float64_array, int32_array)
    1.0 2 3.0 4
    >>> test_dispatch_non_clashing_declarations_repeating_types(float64_array, int64_array, float64_array, int64_array)
    1.0 2 3.0 4
    >>> test_dispatch_non_clashing_declarations_repeating_types(float64_array, int32_array, float64_array, int64_array)  # doctest: +ELLIPSIS
    Traceback (most recent call last):
    ValueError: Buffer dtype mismatch, expected 'int32_t'...
    >>> test_dispatch_non_clashing_declarations_repeating_types(float64_array, int64_array, float64_array, int32_array)  # doctest: +ELLIPSIS
    Traceback (most recent call last):
    ValueError: Buffer dtype mismatch, expected 'int64_t'...
    """
    print a1[1], a2[2], a3[3], a4[4]

ctypedef np.int32_t typedeffed_type

cdef fused typedeffed_fused_type:
    typedeffed_type
    int
    long

@testcase
def test_dispatch_typedef(np.ndarray[typedeffed_fused_type] a):
    """
    >>> test_dispatch_typedef(int32_array)
    5
    """
    print a[5]


cdef extern from "types.h":
    ctypedef char actually_long_t

cdef fused confusing_fused_typedef:
    actually_long_t
    int
    unsigned long
    double complex
    unsigned char
    signed char

def test_dispatch_external_typedef(np.ndarray[confusing_fused_typedef] a):
    """
    >>> test_dispatch_external_typedef(np.arange(-5, 5, dtype=np.long))
    -2
    """
    print a[3]

# test fused memoryview slices
cdef fused memslice_fused_dtype:
    float
    double
    int
    long
    float complex
    double complex
    object

@testcase
def test_fused_memslice_other_dtypes(memslice_fused_dtype[:] a):
    """
    >>> import cython
    >>> sorted(test_fused_memslice_other_dtypes.__signatures__)
    ['double', 'double complex', 'float', 'float complex', 'int', 'long', 'object']
    >>> test_fused_memslice_other_dtypes(np.arange(10, dtype=np.complex64))
    float complex[:] float complex[:] (5+0j) (6+0j)
    >>> test_fused_memslice_other_dtypes(np.arange(10, dtype=np.complex128))
    double complex[:] double complex[:] (5+0j) (6+0j)
    >>> test_fused_memslice_other_dtypes(np.arange(10, dtype=np.float32))
    float[:] float[:] 5.0 6.0
    >>> test_fused_memslice_other_dtypes(np.arange(10, dtype=np.dtype('i')))
    int[:] int[:] 5 6
    >>> test_fused_memslice_other_dtypes(np.arange(10, dtype=np.object))
    object[:] object[:] 5 6
    """
    cdef memslice_fused_dtype[:] b = a
    print cython.typeof(a), cython.typeof(b), a[5], b[6]

cdef fused memslice_fused:
    float[:]
    double[:]
    int[:]
    long[:]
    float complex[:]
    double complex[:]
    object[:]

@testcase
def test_fused_memslice(memslice_fused a):
    """
    >>> import cython
    >>> sorted(test_fused_memslice.__signatures__)
    ['double complex[:]', 'double[:]', 'float complex[:]', 'float[:]', 'int[:]', 'long[:]', 'object[:]']
    >>> test_fused_memslice(np.arange(10, dtype=np.complex64))
    float complex[:] float complex[:] (5+0j) (6+0j)
    >>> test_fused_memslice(np.arange(10, dtype=np.complex128))
    double complex[:] double complex[:] (5+0j) (6+0j)
    >>> test_fused_memslice(np.arange(10, dtype=np.float32))
    float[:] float[:] 5.0 6.0
    >>> test_fused_memslice(np.arange(10, dtype=np.dtype('i')))
    int[:] int[:] 5 6
    >>> test_fused_memslice(np.arange(10, dtype=np.object))
    object[:] object[:] 5 6
    """
    cdef memslice_fused b = a
    print cython.typeof(a), cython.typeof(b), a[5], b[6]

@testcase
def test_dispatch_memoryview_object():
    """
    >>> test_dispatch_memoryview_object()
    int[:] int[:] 5 6
    """
    cdef int[:] m = np.arange(10, dtype=np.dtype('i'))
    cdef int[:] m2 = m
    cdef int[:] m3 = <object> m
    test_fused_memslice(m3)

cdef fused ndim_t:
    double[:]
    double[:, :]
    double[:, :, :]

@testcase
def test_dispatch_ndim(ndim_t array):
    """
    >>> test_dispatch_ndim(np.empty(5, dtype=np.double))
    double[:] 1
    >>> test_dispatch_ndim(np.empty((5, 5), dtype=np.double))
    double[:, :] 2
    >>> test_dispatch_ndim(np.empty((5, 5, 5), dtype=np.double))
    double[:, :, :] 3

    Test indexing using Cython.Shadow
    >>> import cython
    >>> test_dispatch_ndim[cython.double[:]](np.empty(5, dtype=np.double))
    double[:] 1
    >>> test_dispatch_ndim[cython.double[:, :]](np.empty((5, 5), dtype=np.double))
    double[:, :] 2
    """
    print cython.typeof(array), np.asarray(array).ndim


@testcase
def test_copy_buffer(np.ndarray[double, ndim=1] a):
    """
    >>> a = test_copy_buffer(np.ones(10, dtype=np.double))
    >>> len(a)
    10
    >>> print(a.dtype)
    float64
    >>> a[0]
    1.0
    """
    a = a.copy()
    a = a.copy()
    a = a.copy()
    a = a.copy()
    a = a.copy()
    return a


include "numpy_common.pxi"