def dynd_to_np_mask(x):
if is_option(x.dtype):
arr = nd.as_numpy(x.view_scalars(x.dtype.value_type))
missing = is_missing(arr)
else:
arr = nd.as_numpy(x)
missing = np.full_like(arr, False, dtype='bool')
return arr, missing
def test_scalar_agg_bool(op):
np_c = nd.as_numpy(c)
np_d = nd.as_numpy(d)
assert_dynd_eq(op(s_c, s_d)._data, op(np_c, np_d), False)
assert_dynd_eq(op(s_c, True)._data, op(np_c, True), False)
assert_dynd_eq(op(s_d, True)._data, op(np_d, True), False)
assert_dynd_eq(op(s_c, True)._data, op(np_c, True), False)
assert_dynd_eq(op(s_d, True)._data, op(np_d, True), False)
assert_dynd_eq(op(True, s_c)._data, op(True, np_c), False)
assert_dynd_eq(op(True, s_d)._data, op(True, np_d), False)
assert_dynd_eq(op(True, s_c)._data, op(True, np_c), False)
assert_dynd_eq(op(True, s_d)._data, op(True, np_d), False)
def test_string_as_numpy(self):
a = nd.array(["this", "is", "a", "test of varlen strings"])
b = nd.as_numpy(a, allow_copy=True)
self.assertEqual(b.dtype, np.dtype('O'))
assert_equal(b, np.array(["this", "is", "a", "test of varlen strings"],
dtype='O'))
# Also in a struct
a = nd.array([(1, "testing", 1.5), (10, "abc", 2)],
type="strided * {x: int, y: string, z: real}")
b = nd.as_numpy(a, allow_copy=True)
self.assertEqual(b.dtype, np.dtype([('x', 'int32'),
('y', 'O'),
('z', 'float64')], align=True))
self.assertEqual(b.tolist(), [(1, "testing", 1.5), (10, "abc", 2)])
def test_string_as_numpy(self):
a = nd.array(["this", "is", "a", "test of varlen strings"])
b = nd.as_numpy(a, allow_copy=True)
self.assertEqual(b.dtype, np.dtype('O'))
assert_equal(b, np.array(["this", "is", "a", "test of varlen strings"],
dtype='O'))
# Also in a struct
a = nd.array([(1, "testing", 1.5), (10, "abc", 2)],
type="Fixed * {x: int, y: string, z: real}")
b = nd.as_numpy(a, allow_copy=True)
self.assertEqual(b.dtype, np.dtype([('x', 'int32'),
('y', 'O'),
('z', 'float64')], align=True))
self.assertEqual(b.tolist(), [(1, "testing", 1.5), (10, "abc", 2)])
def create_hdf5(self, name):
import tables as tb
a1 = nd.array([[1, 2, 3], [4, 5, 6]], dtype="int32")
a2 = nd.array([[1, 2, 3], [3, 2, 1]], dtype="int32")
fname = os.path.join(self.arrdir, '%s.h5' % name)
with tb.open_file(fname, 'w') as f:
f.create_array(f.root, "a1", nd.as_numpy(a1))
mg = f.create_group(f.root, "mygroup")
f.create_array(mg, "a2", nd.as_numpy(a2))
with open(os.path.join(self.arrdir, '%s.array' % name), 'w') as f:
f.write('type: hdf5\n')
f.write('import: {\n')
f.write(' datapath: /mygroup/a2\n')
f.write(' }\n')
def create_hdf5(self, name):
import tables as tb
a1 = nd.array([[1, 2, 3], [4, 5, 6]], dtype="int32")
a2 = nd.array([[1, 2, 3], [3, 2, 1]], dtype="int32")
fname = os.path.join(self.arrdir, '%s.h5' % name)
with tb.open_file(fname, 'w') as f:
f.create_array(f.root, "a1", nd.as_numpy(a1))
mg = f.create_group(f.root, "mygroup")
f.create_array(mg, "a2", nd.as_numpy(a2))
with open(os.path.join(self.arrdir, '%s.array' % name), 'w') as f:
f.write('type: hdf5\n')
f.write('import: {\n')
f.write(' datapath: /mygroup/a2\n')
f.write(' }\n')
def test_cstruct_as_numpy(self):
# Aligned cstruct
a = nd.array([[1, 2], [3, 4]], dtype='c{x : int32, y: int64}')
b = nd.as_numpy(a)
self.assertEqual(b.dtype,
np.dtype([('x', np.int32), ('y', np.int64)], align=True))
self.assertEqual(nd.as_py(a.x), b['x'].tolist())
self.assertEqual(nd.as_py(a.y), b['y'].tolist())
# Unaligned cstruct
a = nd.array([[1, 2], [3, 4]],
dtype='c{x : unaligned[int32], y: unaligned[int64]}')
b = nd.as_numpy(a)
self.assertEqual(b.dtype, np.dtype([('x', np.int32), ('y', np.int64)]))
self.assertEqual(nd.as_py(a.x), b['x'].tolist())
self.assertEqual(nd.as_py(a.y), b['y'].tolist())
def test_struct_as_numpy(self):
# Aligned struct
a = nd.array([[1, 2], [3, 4]], type='2 * {x : int32, y: int64}')
b = nd.as_numpy(a)
self.assertEqual(b.dtype,
np.dtype([('x', np.int32), ('y', np.int64)], align=True))
self.assertEqual(nd.as_py(a.x), b['x'].tolist())
self.assertEqual(nd.as_py(a.y), b['y'].tolist())
# Unaligned struct
a = nd.array([[1, 2], [3, 4]],
type='2 * {x : unaligned[int32], y: unaligned[int64]}')
b = nd.as_numpy(a)
self.assertEqual(b.dtype, np.dtype([('x', np.int32), ('y', np.int64)]))
self.assertEqual(nd.as_py(a.x), b['x'].tolist())
self.assertEqual(nd.as_py(a.y), b['y'].tolist())
def test_datetime_as_numpy(self):
a = nd.array(['2000-12-13T12:30',
'1995-05-02T2:15:33'],
dtype='datetime[tz="UTC"]')
b = nd.as_numpy(a, allow_copy=True)
assert_equal(b, np.array(['2000-12-13T12:30Z', '1995-05-02T02:15:33Z'],
dtype='M8[us]'))
def test__type_from_h5py_special(self):
# h5py 2.3 style "special dtype"
dt = np.dtype(object, metadata={'vlen' : str})
self.assertEqual(ndt.type(dt), ndt.string)
if sys.version_info < (3, 0):
dt = np.dtype(object, metadata={'vlen' : unicode})
self.assertEqual(ndt.type(dt), ndt.string)
# h5py 2.2 style "special dtype"
dt = np.dtype(('O', [( ({'type': str},'vlen'), 'O' )] ))
self.assertEqual(ndt.type(dt), ndt.string)
if sys.version_info < (3, 0):
dt = np.dtype(('O', [( ({'type': unicode},'vlen'), 'O' )] ))
self.assertEqual(ndt.type(dt), ndt.string)
# Should be able to roundtrip dynd -> numpy -> dynd
x = nd.array(['testing', 'one', 'two'])
self.assertEqual(nd.type_of(x), ndt.type('3 * string'))
y = nd.as_numpy(x, allow_copy=True)
self.assertEqual(y.shape, (3,))
self.assertEqual(y[0], 'testing')
self.assertEqual(y[1], 'one')
self.assertEqual(y[2], 'two')
self.assertEqual(y.dtype.kind, 'O')
if sys.version_info < (3, 0):
self.assertEqual(y.dtype.metadata, {'vlen' : unicode})
else:
self.assertEqual(y.dtype.metadata, {'vlen' : str})
z = nd.array(y)
self.assertEqual(nd.type_of(z), nd.type_of(x))
self.assertEqual(nd.as_py(z), nd.as_py(x))
def test_ndt_type_from_h5py_special(self):
# h5py 2.3 style "special dtype"
dt = np.dtype(object, metadata={'vlen' : str})
self.assertEqual(ndt.type(dt), ndt.string)
if sys.version_info < (3, 0):
dt = np.dtype(object, metadata={'vlen' : unicode})
self.assertEqual(ndt.type(dt), ndt.string)
# h5py 2.2 style "special dtype"
dt = np.dtype(('O', [( ({'type': str},'vlen'), 'O' )] ))
self.assertEqual(ndt.type(dt), ndt.string)
if sys.version_info < (3, 0):
dt = np.dtype(('O', [( ({'type': unicode},'vlen'), 'O' )] ))
self.assertEqual(ndt.type(dt), ndt.string)
# Should be able to roundtrip dynd -> numpy -> dynd
x = nd.array(['testing', 'one', 'two'])
self.assertEqual(nd.type_of(x), ndt.type('3 * string'))
y = nd.as_numpy(x, allow_copy=True)
self.assertEqual(y.dtype.kind, 'O')
if sys.version_info < (3, 0):
self.assertEqual(y.dtype.metadata, {'vlen' : unicode})
else:
self.assertEqual(y.dtype.metadata, {'vlen' : str})
z = nd.array(y)
self.assertEqual(nd.type_of(z), nd.type_of(x))
self.assertEqual(nd.as_py(z), nd.as_py(x))
def sort(self, json_cmd):
import numpy as np
print('sorting')
cmd = json.loads(json_cmd)
array_url = cmd.get('input', self.base_url + self.array_name)
if not array_url.startswith(self.base_url):
raise RuntimeError('Input array must start with the base url')
array_name = array_url[len(self.base_url):]
field = cmd['field']
arr = self.get_session_array(array_name)
nparr = as_numpy(arr)
idxs = np.argsort(nparr[field])
res = nd.ndobject(nparr[idxs])
defarr = self.array_provider.create_deferred_array_filename(
self.session_name, 'sort_', res)
dshape = nd.dshape_of(res)
defarr[0].write(
json.dumps({
'dshape': dshape,
'command': 'sort',
'params': {
'field': field,
}
}))
defarr[0].close()
content_type = 'application/json; charset=utf-8'
body = json.dumps({
'session': self.base_url + self.session_name,
'output': self.base_url + defarr[1],
'dshape': dshape
})
return (content_type, body)
def ones(dshape, ddesc=None):
"""Create an array and fill it with ones.
Parameters
----------
dshape : datashape
The datashape for the resulting array.
ddesc : data descriptor instance
This comes with the necessary info for storing the data. If
None, a DyND_DDesc will be used.
Returns
-------
out: a concrete blaze array.
"""
dshape = _normalize_dshape(dshape)
if ddesc is None:
ddesc = DyND_DDesc(nd.ones(str(dshape), access='rw'))
return Array(ddesc)
if isinstance(ddesc, BLZ_DDesc):
shape, dt = to_numpy(dshape)
ddesc.blzarr = blz.ones(
shape, dt, rootdir=ddesc.path, mode=ddesc.mode, **ddesc.kwargs)
elif isinstance(ddesc, HDF5_DDesc):
obj = nd.as_numpy(nd.empty(str(dshape)))
with tb.open_file(ddesc.path, mode=ddesc.mode) as f:
where, name = split_path(ddesc.datapath)
f.create_earray(where, name, filters=ddesc.filters, obj=obj)
ddesc.mode = 'a' # change into 'a'ppend mode for further operations
return Array(ddesc)
def sort(self, json_cmd):
import numpy as np
print ('sorting')
cmd = json.loads(json_cmd)
array_url = cmd.get('input', self.base_url + self.array_name)
if not array_url.startswith(self.base_url):
raise RuntimeError('Input array must start with the base url')
array_name = array_url[len(self.base_url):]
field = cmd['field']
arr = self.get_session_array(array_name)
nparr = as_numpy(arr)
idxs = np.argsort(nparr[field])
res = nd.ndobject(nparr[idxs])
defarr = self.array_provider.create_deferred_array_filename(
self.session_name, 'sort_', res)
dshape = nd.dshape_of(res)
defarr[0].write(json.dumps({
'dshape': dshape,
'command': 'sort',
'params': {
'field': field,
}
}))
defarr[0].close()
content_type = 'application/json; charset=utf-8'
body = json.dumps({
'session': self.base_url + self.session_name,
'output': self.base_url + defarr[1],
'dshape': dshape
})
return (content_type, body)
def finalize(bases):
shape = bases[0].shape[:2]
out = nd.empty(shape, dshape)
for path, finalizer, inds in zip(paths, finalizers, indices):
arr = reduce(getattr, path, out)
np_arr = nd.as_numpy(arr.view_scalars(arr.dtype.value_type))
np_arr[:] = finalizer(*get(inds, bases))
return out
def finalize(bases):
shape = bases[0].shape[:2]
out = nd.empty(shape, dshape)
for path, finalizer, inds in zip(paths, finalizers, indices):
arr = reduce(getattr, path, out)
np_arr = nd.as_numpy(arr.view_scalars(arr.dtype.value_type))
np_arr[:] = finalizer(*get(inds, bases))
return out
def test_datetime_as_numpy(self):
a = nd.array(['2000-12-13T12:30', '1995-05-02T2:15:33'],
dtype='datetime[tz="UTC"]')
b = nd.as_numpy(a, allow_copy=True)
assert_equal(
b,
np.array(['2000-12-13T12:30Z', '1995-05-02T02:15:33Z'],
dtype='M8[us]'))
def test_expr_struct_conversion(self):
a = nd.array([date(2000, 12, 13), date(1995, 5, 2)]).to_struct
b = nd.as_numpy(a, allow_copy=True)
self.assertTrue(isinstance(b, np.ndarray))
# Use the NumPy assertions which support arrays
assert_equal(b['year'], [2000, 1995])
assert_equal(b['month'], [12, 5])
assert_equal(b['day'], [13, 2])
def test_expr_struct_conversion(self):
a = nd.array([date(2000, 12, 13), date(1995, 5, 2)]).to_struct()
b = nd.as_numpy(a, allow_copy=True)
self.assertTrue(isinstance(b, np.ndarray))
# Use the NumPy assertions which support arrays
assert_equal(b['year'], [2000, 1995])
assert_equal(b['month'], [12, 5])
assert_equal(b['day'], [13, 2])
def test_count_cat():
agg = c.points(df, 'x', 'y', ds.count_cat('cat'))
sol = np.array([[[5, 0, 0, 0],
[0, 0, 5, 0]],
[[0, 5, 0, 0],
[0, 0, 0, 5]]])
assert (nd.as_numpy(agg._data) == sol).all()
assert agg._cats == ('a', 'b', 'c', 'd')
def test_var_dim_conversion(self):
# A simple instantiated var_dim array should be
# viewable with numpy without changes
a = nd.array([1, 2, 3, 4, 5], type='var * int32')
b = nd.as_numpy(a)
self.assertTrue(isinstance(b, np.ndarray))
self.assertEqual(b.dtype, np.dtype('int32'))
# Use the NumPy assertions which support arrays
assert_equal(b, [1, 2, 3, 4, 5])
def test_var_dim_conversion(self):
# A simple instantiated var_dim array should be
# viewable with numpy without changes
a = nd.array([1, 2, 3, 4, 5], type='var * int32')
b = nd.as_numpy(a)
self.assertTrue(isinstance(b, np.ndarray))
self.assertEqual(b.dtype, np.dtype('int32'))
# Use the NumPy assertions which support arrays
assert_equal(b, [1, 2, 3, 4, 5])
def _extend_chunks(self, chunks):
with h5py.File(self.path, mode='a') as f:
dset = f[self.datapath]
dtype = dset.dtype
shape = dset.shape
for chunk in chunks:
arr = nd.as_numpy(chunk, allow_copy=True)
shape = list(dset.shape)
shape[0] += len(arr)
dset.resize(shape)
dset[-len(arr):] = arr
def __getitem__(self, key):
try:
if isinstance(key, list):
# List of categories
inds = [self._cats.index(k) for k in key]
dtype = self._data.dtype
if is_option(dtype):
out = nd.as_numpy(self._data.view_scalars(
dtype.value_type))
else:
out = nd.as_numpy(self._data)
out = nd.asarray(out[:, :, inds]).view_scalars(dtype)
return CategoricalAggregate(out, key, self.x_axis, self.y_axis)
else:
# Single category
i = self._cats.index(key)
return ScalarAggregate(self._data[:, :, i],
self.x_axis, self.y_axis)
except ValueError:
raise KeyError("'{0}'".format(key))
def _extend_chunks(self, chunks):
with h5py.File(self.path, mode='a') as f:
dset = f[self.datapath]
dtype = dset.dtype
shape = dset.shape
for chunk in chunks:
arr = nd.as_numpy(chunk, allow_copy=True)
shape = list(dset.shape)
shape[0] += len(arr)
dset.resize(shape)
dset[-len(arr):] = arr
def _extend_chunks(self, chunks):
if 'w' not in self.mode and 'a' not in self.mode:
raise ValueError('Read only')
with h5py.File(self.path, mode=self.mode) as f:
dset = f[self.datapath]
dtype = dset.dtype
shape = dset.shape
for chunk in chunks:
arr = nd.as_numpy(chunk, allow_copy=True)
shape = list(dset.shape)
shape[0] += len(arr)
dset.resize(shape)
dset[-len(arr):] = arr
def create_hdf5(self, name):
import tables as tb
a1 = nd.array([[1, 2, 3], [4, 5, 6]], dtype="int32")
a2 = nd.array([[1, 2, 3], [3, 2, 1]], dtype="int32")
a3 = nd.array([[1, 3, 2], [2, 1, 3]], dtype="int32")
fname = os.path.join(self.arrdir, '%s_arr.h5' % name)
with tb.open_file(fname, 'w') as f:
f.create_array(f.root, "a1", nd.as_numpy(a1))
mg = f.create_group(f.root, "mygroup")
f.create_array(mg, "a2", nd.as_numpy(a2))
f.create_array(mg, "a3", nd.as_numpy(a3))
mg2 = f.create_group(mg, "mygroup2")
# Create a .array file for locating the dataset inside the file
with open(os.path.join(self.arrdir, '%s_arr.array' % name), 'w') as f:
f.write('type: hdf5\n')
f.write('import: {\n')
f.write(' datapath: /mygroup/a2\n')
f.write(' }\n')
# Create a .dir file for listing datasets inside the file
with open(os.path.join(self.arrdir, '%s_dir.dir' % name), 'w') as f:
f.write('type: hdf5\n')
f.write('import: {\n')
f.write(' filename: "%s"\n' % fname.replace('\\', '\\\\'))
f.write(' }\n')
def colorize(agg, color_key, how="log", min_alpha=20):
"""Color a CategoricalAggregate by field.
Parameters
----------
agg : CategoricalAggregate
color_key : dict or iterable
A mapping of fields to colors. Can be either a ``dict`` mapping from
field name to colors, or an iterable of colors in the same order as the
record fields.
how : string or callable
The interpolation method to use. Valid strings are 'log' [default],
'cbrt', and 'linear'. Callables take a 2-dimensional array of
magnitudes at each pixel, and should return a numeric array of the same
shape.
min_alpha : float, optional
The minimum alpha value to use for non-empty pixels, in [0, 255].
"""
if not isinstance(agg, CategoricalAggregate):
raise TypeError("agg must be instance of CategoricalAggregate")
if not isinstance(color_key, dict):
color_key = dict(zip(agg.cats, color_key))
if len(color_key) != len(agg.cats):
raise ValueError("Number of colors doesn't match number of fields")
if not (0 <= min_alpha <= 255):
raise ValueError("min_alpha must be between 0 and 255")
colors = [rgb(color_key[c]) for c in agg.cats]
rs, gs, bs = map(np.array, zip(*colors))
data = nd.as_numpy(agg._data).astype("f8")
total = data.sum(axis=2)
r = (data.dot(rs) / total).astype(np.uint8)
g = (data.dot(gs) / total).astype(np.uint8)
b = (data.dot(bs) / total).astype(np.uint8)
a = _normalize_interpolate_how(how)(total)
a = ((255 - min_alpha) * a / a.max() + min_alpha).astype(np.uint8)
white = total == 0
r[white] = g[white] = b[white] = 255
a[white] = 0
return Image(np.dstack([r, g, b, a]).view(np.uint32).reshape(a.shape), agg.x_axis, agg.y_axis)
def ones(dshape, ddesc=None):
"""Create an array and fill it with ones.
Parameters
----------
dshape : datashape
The datashape for the resulting array.
ddesc : data descriptor instance
This comes with the necessary info for storing the data. If
None, a DyND_DDesc will be used.
Returns
-------
out: a concrete blaze array.
"""
dshape = _normalize_dshape(dshape)
if ddesc is None:
ddesc = DyND_DDesc(nd.ones(str(dshape), access='rw'))
return Array(ddesc)
if isinstance(ddesc, BLZ_DDesc):
shape, dt = to_numpy(dshape)
ddesc.blzarr = blz.ones(shape,
dt,
rootdir=ddesc.path,
mode=ddesc.mode,
**ddesc.kwargs)
elif isinstance(ddesc, HDF5_DDesc):
obj = nd.as_numpy(nd.empty(str(dshape)))
with tb.open_file(ddesc.path, mode=ddesc.mode) as f:
where, name = split_path(ddesc.datapath)
f.create_earray(where, name, filters=ddesc.filters, obj=obj)
ddesc.mode = 'a' # change into 'a'ppend mode for further operations
return Array(ddesc)
def into(a, b):
return nd.as_numpy(b, allow_copy=True)
def test_date_as_numpy(self):
a = nd.array([date(2000, 12, 13), date(1995, 5, 2)])
b = nd.as_numpy(a, allow_copy=True)
assert_equal(b, np.array(['2000-12-13', '1995-05-02'], dtype='M8[D]'))
def into(a, b, **kwargs):
return nd.as_numpy(b, allow_copy=True)
def eq(agg, b):
a = nd.as_numpy(agg.view_scalars(agg.dtype.value_type))
assert np.allclose(a, b)
assert a.dtype == b.dtype
def eq(agg, b):
agg = agg._data
a = nd.as_numpy(agg.view_scalars(getattr(agg.dtype, 'value_type', agg.dtype)))
assert np.allclose(a, b)
assert a.dtype == b.dtype
def dynd_to_numpy(x, **kwargs):
return nd.as_numpy(x, allow_copy=True)
def test_fixed_dim(self):
a = nd.array([1, 3, 5], type='3 * int32')
b = nd.as_numpy(a)
self.assertEqual(b.dtype, np.dtype('int32'))
self.assertEqual(b.tolist(), [1, 3, 5])
def test_date_as_numpy(self):
a = nd.array([date(2000, 12, 13), date(1995, 5, 2)])
b = nd.as_numpy(a, allow_copy=True)
assert_equal(b, np.array(['2000-12-13', '1995-05-02'], dtype='M8[D]'))
def dynd_to_numpy(x, **kwargs):
return nd.as_numpy(x, allow_copy=True)
def __array__(self):
return nd.as_numpy(self.dynd_arr())
def __array__(self):
return nd.as_numpy(self.as_dynd())
def __array__(self):
return nd.as_numpy(self.dynd_arr())
def array(obj, dshape=None, ddesc=None):
"""Create a Blaze array.
Parameters
----------
obj : array_like
Initial contents for the array.
dshape : datashape
The datashape for the resulting array. By default the
datashape will be inferred from data. If an explicit dshape is
provided, the input data will be coerced into the provided
dshape.
ddesc : data descriptor instance
This comes with the necessary info for storing the data. If
None, a DyND_DDesc will be used.
Returns
-------
out : a concrete blaze array.
"""
dshape = _normalize_dshape(dshape)
if ((obj is not None) and (not inspect.isgenerator(obj))
and (dshape is not None)):
dt = ndt.type(str(dshape))
if dt.ndim > 0:
obj = nd.array(obj, type=dt, access='rw')
else:
obj = nd.array(obj, dtype=dt, access='rw')
if obj is None and ddesc is None:
raise ValueError('you need to specify at least `obj` or `ddesc`')
if isinstance(obj, Array):
return obj
elif isinstance(obj, DDesc):
if ddesc is None:
ddesc = obj
return Array(ddesc)
else:
raise ValueError(('you cannot specify `ddesc` when `obj` '
'is already a DDesc instance'))
if ddesc is None:
# Use a dynd ddesc by default
try:
array = nd.asarray(obj, access='rw')
except:
raise ValueError(('failed to construct a dynd array from '
'object %r') % obj)
ddesc = DyND_DDesc(array)
return Array(ddesc)
# The DDesc has been specified
if isinstance(ddesc, DyND_DDesc):
if obj is not None:
raise ValueError(('you cannot specify simultaneously '
'`obj` and a DyND `ddesc`'))
return Array(ddesc)
elif isinstance(ddesc, BLZ_DDesc):
if inspect.isgenerator(obj):
dt = None if dshape is None else to_numpy_dtype(dshape)
# TODO: Generator logic could go inside barray
ddesc.blzarr = blz.fromiter(obj,
dtype=dt,
count=-1,
rootdir=ddesc.path,
mode=ddesc.mode,
**ddesc.kwargs)
else:
if isinstance(obj, nd.array):
obj = nd.as_numpy(obj)
if dshape and isinstance(dshape.measure, datashape.Record):
ddesc.blzarr = blz.btable(obj,
rootdir=ddesc.path,
mode=ddesc.mode,
**ddesc.kwargs)
else:
ddesc.blzarr = blz.barray(obj,
rootdir=ddesc.path,
mode=ddesc.mode,
**ddesc.kwargs)
elif isinstance(ddesc, HDF5_DDesc):
if isinstance(obj, nd.array):
obj = nd.as_numpy(obj)
with tb.open_file(ddesc.path, mode=ddesc.mode) as f:
where, name = split_path(ddesc.datapath)
if dshape and isinstance(dshape.measure, datashape.Record):
# Convert the structured array to unaligned dtype
# We need that because PyTables only accepts unaligned types,
# which are the default in NumPy
obj = np.array(obj, datashape.to_numpy_dtype(dshape.measure))
f.create_table(where, name, filters=ddesc.filters, obj=obj)
else:
f.create_earray(where, name, filters=ddesc.filters, obj=obj)
ddesc.mode = 'a' # change into 'a'ppend mode for further operations
return Array(ddesc)
def eq(agg, b):
a = nd.as_numpy(agg.view_scalars(agg.dtype.value_type))
assert np.allclose(a, b)
assert a.dtype == b.dtype
def into(a, b, **kwargs):
return nd.as_numpy(b, allow_copy=True)
def array(obj, dshape=None, ddesc=None):
"""Create a Blaze array.
Parameters
----------
obj : array_like
Initial contents for the array.
dshape : datashape
The datashape for the resulting array. By default the
datashape will be inferred from data. If an explicit dshape is
provided, the input data will be coerced into the provided
dshape.
ddesc : data descriptor instance
This comes with the necessary info for storing the data. If
None, a DyND_DDesc will be used.
Returns
-------
out : a concrete blaze array.
"""
dshape = _normalize_dshape(dshape)
if ((obj is not None) and
(not inspect.isgenerator(obj)) and
(dshape is not None)):
dt = ndt.type(str(dshape))
if dt.ndim > 0:
obj = nd.array(obj, type=dt, access='rw')
else:
obj = nd.array(obj, dtype=dt, access='rw')
if obj is None and ddesc is None:
raise ValueError('you need to specify at least `obj` or `ddesc`')
if isinstance(obj, Array):
return obj
elif isinstance(obj, DDesc):
if ddesc is None:
ddesc = obj
return Array(ddesc)
else:
raise ValueError(('you cannot specify `ddesc` when `obj` '
'is already a DDesc instance'))
if ddesc is None:
# Use a dynd ddesc by default
try:
array = nd.asarray(obj, access='rw')
except:
raise ValueError(('failed to construct a dynd array from '
'object %r') % obj)
ddesc = DyND_DDesc(array)
return Array(ddesc)
# The DDesc has been specified
if isinstance(ddesc, DyND_DDesc):
if obj is not None:
raise ValueError(('you cannot specify simultaneously '
'`obj` and a DyND `ddesc`'))
return Array(ddesc)
elif isinstance(ddesc, BLZ_DDesc):
if inspect.isgenerator(obj):
dt = None if dshape is None else to_numpy_dtype(dshape)
# TODO: Generator logic could go inside barray
ddesc.blzarr = blz.fromiter(obj, dtype=dt, count=-1,
rootdir=ddesc.path, mode=ddesc.mode,
**ddesc.kwargs)
else:
if isinstance(obj, nd.array):
obj = nd.as_numpy(obj)
if dshape and isinstance(dshape.measure, datashape.Record):
ddesc.blzarr = blz.btable(
obj, rootdir=ddesc.path, mode=ddesc.mode, **ddesc.kwargs)
else:
ddesc.blzarr = blz.barray(
obj, rootdir=ddesc.path, mode=ddesc.mode, **ddesc.kwargs)
elif isinstance(ddesc, HDF5_DDesc):
if isinstance(obj, nd.array):
obj = nd.as_numpy(obj)
with tb.open_file(ddesc.path, mode=ddesc.mode) as f:
where, name = split_path(ddesc.datapath)
if dshape and isinstance(dshape.measure, datashape.Record):
# Convert the structured array to unaligned dtype
# We need that because PyTables only accepts unaligned types,
# which are the default in NumPy
obj = np.array(obj, datashape.to_numpy_dtype(dshape.measure))
f.create_table(where, name, filters=ddesc.filters, obj=obj)
else:
f.create_earray(where, name, filters=ddesc.filters, obj=obj)
ddesc.mode = 'a' # change into 'a'ppend mode for further operations
return Array(ddesc)
def __array__(self):
return nd.as_numpy(self.as_dynd())
def test_fixed_dim(self):
a = nd.array([1, 3, 5], type='3 * int32')
b = nd.as_numpy(a)
self.assertEqual(b.dtype, np.dtype('int32'))
self.assertEqual(b.tolist(), [1, 3, 5])
