def test_compress_level_error(wrong_compress):
# Verify that passing an invalid compress argument raises an error.
exception_msg = ('Non valid compress level given: '
'"{0}"'.format(wrong_compress))
with raises(ValueError) as excinfo:
numpy_pickle.dump('dummy', 'foo', compress=wrong_compress)
excinfo.match(exception_msg)
def cache_value(value, filename, decimal=7):
"""Helper function for checking that a value hasn't changed between
two invocations.
First call: write value is a file
Second call: check that what was written is identical to the value
provided in the second call.
TODO: only numpy arrays are compared, other values still have to
be compared.
Parameters
==========
value: arbitrary Python value
this could include numpy objects. Uses persistence from joblib to
achieve high efficiency.
"""
from joblib.numpy_pickle import dump, load
base_dir = os.path.split(filename)[0]
if not os.path.isdir(base_dir):
os.makedirs(base_dir)
if os.path.isfile(filename):
cached = load(filename)
np.testing.assert_almost_equal(cached, value, decimal=decimal)
else:
dump(value, filename)
def test_numpy_persistence_bufferred_array_compression():
big_array = np.ones((_IO_BUFFER_SIZE + 100), dtype=np.uint8)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(big_array, filename, compress=True)
arr_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(big_array, arr_reloaded)
def test_file_handle_persistence_compressed_mmap():
obj = np.random.random((10, 10))
filename = env['filename'] + str(random.randint(0, 1000))
with open(filename, 'wb') as f:
numpy_pickle.dump(obj, f, compress=('gzip', 3))
with closing(gzip.GzipFile(filename, 'rb')) as f:
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.load(f, mmap_mode='r+')
nose.tools.assert_equal(len(caught_warnings), 1)
for warn in caught_warnings:
nose.tools.assert_equal(warn.category, UserWarning)
nose.tools.assert_equal(
warn.message.args[0], 'File "%(filename)s" is compressed '
'using "%(compressor)s" which is not '
'compatible with mmap_mode '
'"%(mmap_mode)s" flag '
'passed. mmap_mode option will be '
'ignored.' % {
'filename': "",
'mmap_mode': 'r+',
'compressor': 'GzipFile'
})
def test_joblib_compression_formats():
compresslevels = (1, 3, 6)
filename = env['filename'] + str(random.randint(0, 1000))
objects = (np.ones(shape=(100, 100), dtype='f8'),
range(10),
{'a': 1, 2: 'b'}, [], (), {}, 0, 1.0)
for compress in compresslevels:
for cmethod in _COMPRESSORS:
dump_filename = filename + "." + cmethod
for obj in objects:
if not PY3_OR_LATER and cmethod in ('xz', 'lzma'):
# Lzma module only available for python >= 3.3
msg = "{} compression is only available".format(cmethod)
assert_raises_regex(NotImplementedError, msg,
numpy_pickle.dump, obj, dump_filename,
compress=(cmethod, compress))
else:
numpy_pickle.dump(obj, dump_filename,
compress=(cmethod, compress))
# Verify the file contains the right magic number
with open(dump_filename, 'rb') as f:
assert _detect_compressor(f) == cmethod
# Verify the reloaded object is correct
obj_reloaded = numpy_pickle.load(dump_filename)
assert isinstance(obj_reloaded, type(obj))
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
else:
assert obj_reloaded == obj
os.remove(dump_filename)
def test_numpy_subclass(tmpdir):
filename = tmpdir.join('test.pkl').strpath
a = SubArray((10, ))
numpy_pickle.dump(a, filename)
c = numpy_pickle.load(filename)
assert isinstance(c, SubArray)
np.testing.assert_array_equal(c, a)
def test_numpy_persistence_bufferred_array_compression():
big_array = np.ones((_IO_BUFFER_SIZE + 100), dtype=np.uint8)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(big_array, filename, compress=True)
arr_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(big_array, arr_reloaded)
def test_compress_tuple_argument():
compress_tuples = (('zlib', 3),
('gzip', 3))
# Verify the tuple is correctly taken into account.
filename = env['filename'] + str(random.randint(0, 1000))
for compress in compress_tuples:
numpy_pickle.dump("dummy", filename,
compress=compress)
# Verify the file contains the right magic number
with open(filename, 'rb') as f:
assert _detect_compressor(f) == compress[0]
# Verify setting a wrong compress tuple raises a ValueError.
assert_raises_regex(ValueError,
'Compress argument tuple should contain exactly '
'2 elements',
numpy_pickle.dump, "dummy", filename,
compress=('zlib', 3, 'extra'))
# Verify a tuple with a wrong compress method raises a ValueError.
msg = 'Non valid compression method given: "{}"'.format('wrong')
assert_raises_regex(ValueError, msg,
numpy_pickle.dump, "dummy", filename,
compress=('wrong', 3))
# Verify a tuple with a wrong compress level raises a ValueError.
msg = 'Non valid compress level given: "{}"'.format('wrong')
assert_raises_regex(ValueError, msg,
numpy_pickle.dump, "dummy", filename,
compress=('zlib', 'wrong'))
def test_file_handle_persistence():
objs = [np.random.random((10, 10)),
"some data",
np.matrix([0, 1, 2])]
fobjs = [bz2.BZ2File, gzip.GzipFile]
if PY3_OR_LATER:
import lzma
fobjs += [lzma.LZMAFile]
filename = env['filename'] + str(random.randint(0, 1000))
for obj in objs:
for fobj in fobjs:
with fobj(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
# using the same decompressor prevents from internally
# decompress again.
with fobj(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f)
# when needed, the correct decompressor should be used when
# passing a raw file handle.
with open(filename, 'rb') as f:
obj_reloaded_2 = numpy_pickle.load(f)
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
np.testing.assert_array_equal(obj_reloaded_2, obj)
else:
assert obj_reloaded == obj
assert obj_reloaded_2 == obj
os.remove(filename)
def test_numpy_persistence_bufferred_array_compression(tmpdir):
big_array = np.ones((_IO_BUFFER_SIZE + 100), dtype=np.uint8)
filename = tmpdir.join('test.pkl').strpath
numpy_pickle.dump(big_array, filename, compress=True)
arr_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(big_array, arr_reloaded)
def test_compression_using_file_extension():
# test that compression method corresponds to the given filename extension.
extensions_dict = {
# valid compressor extentions
'.z': 'zlib',
'.gz': 'gzip',
'.bz2': 'bz2',
'.lzma': 'lzma',
'.xz': 'xz',
# invalid compressor extensions
'.pkl': 'not-compressed',
'': 'not-compressed'
}
filename = env['filename'] + str(random.randint(0, 1000))
obj = "object to dump"
for ext, cmethod in extensions_dict.items():
dump_fname = filename + ext
if not PY3_OR_LATER and cmethod in ('xz', 'lzma'):
# Lzma module only available for python >= 3.3
msg = "{0} compression is only available".format(cmethod)
assert_raises_regex(NotImplementedError, msg, numpy_pickle.dump,
obj, dump_fname)
else:
numpy_pickle.dump(obj, dump_fname)
# Verify the file contains the right magic number
with open(dump_fname, 'rb') as f:
nose.tools.assert_equal(_detect_compressor(f), cmethod)
# Verify the reloaded object is correct
obj_reloaded = numpy_pickle.load(dump_fname)
nose.tools.assert_true(isinstance(obj_reloaded, type(obj)))
nose.tools.assert_equal(obj_reloaded, obj)
os.remove(dump_fname)
def test_numpy_subclass():
filename = env['filename']
a = SubArray((10,))
numpy_pickle.dump(a, filename)
c = numpy_pickle.load(filename)
assert isinstance(c, SubArray)
np.testing.assert_array_equal(c, a)
def test_numpy_subclass():
filename = env['filename']
a = SubArray((10, ))
numpy_pickle.dump(a, filename)
c = numpy_pickle.load(filename)
nose.tools.assert_true(isinstance(c, SubArray))
np.testing.assert_array_equal(c, a)
def test_joblib_compression_formats(tmpdir, compress, cmethod):
filename = tmpdir.join('test.pkl').strpath
objects = (np.ones(shape=(100, 100), dtype='f8'), range(10), {
'a': 1,
2: 'b'
}, [], (), {}, 0, 1.0)
if cmethod in ("lzma", "xz") and lzma is None:
pytest.skip("lzma is support not available")
elif cmethod == 'lz4' and with_lz4.args[0]:
# Skip the test if lz4 is not installed. We here use the with_lz4
# skipif fixture whose argument is True when lz4 is not installed
pytest.skip("lz4 is not installed.")
dump_filename = filename + "." + cmethod
for obj in objects:
numpy_pickle.dump(obj, dump_filename, compress=(cmethod, compress))
# Verify the file contains the right magic number
with open(dump_filename, 'rb') as f:
assert _detect_compressor(f) == cmethod
# Verify the reloaded object is correct
obj_reloaded = numpy_pickle.load(dump_filename)
assert isinstance(obj_reloaded, type(obj))
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
else:
assert obj_reloaded == obj
def test_compress_string_argument(tmpdir, compress_string):
# Verify the string is correctly taken into account.
filename = tmpdir.join('test.pkl').strpath
numpy_pickle.dump("dummy", filename, compress=compress_string)
# Verify the file contains the right magic number
with open(filename, 'rb') as f:
assert _detect_compressor(f) == compress_string
def test_file_handle_persistence(tmpdir):
objs = [np.random.random((10, 10)), "some data", np.matrix([0, 1, 2])]
fobjs = [bz2.BZ2File, gzip.GzipFile]
if lzma is not None:
fobjs += [lzma.LZMAFile]
filename = tmpdir.join('test.pkl').strpath
for obj in objs:
for fobj in fobjs:
with fobj(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
# using the same decompressor prevents from internally
# decompress again.
with fobj(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f)
# when needed, the correct decompressor should be used when
# passing a raw file handle.
with open(filename, 'rb') as f:
obj_reloaded_2 = numpy_pickle.load(f)
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
np.testing.assert_array_equal(obj_reloaded_2, obj)
else:
assert obj_reloaded == obj
assert obj_reloaded_2 == obj
def test_compressed_pickle_dump_and_load():
expected_list = [
np.arange(5, dtype=np.int64),
np.arange(5, dtype=np.float64),
# .tostring actually returns bytes and is a
# compatibility alias for .tobytes which was
# added in 1.9.0
np.arange(256, dtype=np.uint8).tostring(),
u"C'est l'\xe9t\xe9 !"
]
with tempfile.NamedTemporaryFile(suffix='.gz', dir=env['dir']) as f:
fname = f.name
try:
numpy_pickle.dump(expected_list, fname, compress=1)
result_list = numpy_pickle.load(fname)
for result, expected in zip(result_list, expected_list):
if isinstance(expected, np.ndarray):
nose.tools.assert_equal(result.dtype, expected.dtype)
np.testing.assert_equal(result, expected)
else:
nose.tools.assert_equal(result, expected)
finally:
os.remove(fname)
def test_compression_using_file_extension():
# test that compression method corresponds to the given filename extension.
extensions_dict = {
# valid compressor extentions
'.z': 'zlib',
'.gz': 'gzip',
'.bz2': 'bz2',
'.lzma': 'lzma',
'.xz': 'xz',
# invalid compressor extensions
'.pkl': 'not-compressed',
'': 'not-compressed'
}
filename = env['filename'] + str(random.randint(0, 1000))
obj = "object to dump"
for ext, cmethod in extensions_dict.items():
dump_fname = filename + ext
if not PY3_OR_LATER and cmethod in ('xz', 'lzma'):
# Lzma module only available for python >= 3.3
msg = "{} compression is only available".format(cmethod)
assert_raises_regex(NotImplementedError, msg,
numpy_pickle.dump, obj, dump_fname)
else:
numpy_pickle.dump(obj, dump_fname)
# Verify the file contains the right magic number
with open(dump_fname, 'rb') as f:
assert _detect_compressor(f) == cmethod
# Verify the reloaded object is correct
obj_reloaded = numpy_pickle.load(dump_fname)
assert isinstance(obj_reloaded, type(obj))
assert obj_reloaded == obj
os.remove(dump_fname)
def test_file_handle_persistence():
objs = [np.random.random((10, 10)), "some data", np.matrix([0, 1, 2])]
fobjs = [open]
if not PY26:
fobjs += [bz2.BZ2File, gzip.GzipFile]
if PY3_OR_LATER:
import lzma
fobjs += [lzma.LZMAFile]
filename = env['filename'] + str(random.randint(0, 1000))
for obj in objs:
for fobj in fobjs:
with fobj(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
# using the same decompressor prevents from internally
# decompress again.
with fobj(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f)
# when needed, the correct decompressor should be used when
# passing a raw file handle.
with open(filename, 'rb') as f:
obj_reloaded_2 = numpy_pickle.load(f)
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
np.testing.assert_array_equal(obj_reloaded_2, obj)
else:
nose.tools.assert_equal(obj_reloaded, obj)
nose.tools.assert_equal(obj_reloaded_2, obj)
os.remove(filename)
def test_joblib_compression_formats(tmpdir, compress, cmethod):
filename = tmpdir.join('test.pkl').strpath
objects = (np.ones(shape=(100, 100), dtype='f8'), range(10), {
'a': 1,
2: 'b'
}, [], (), {}, 0, 1.0)
dump_filename = filename + "." + cmethod
for obj in objects:
if not PY3_OR_LATER and cmethod in ('lzma', 'xz', 'lz4'):
# Lzma module only available for python >= 3.3
msg = "{} compression is only available".format(cmethod)
error = NotImplementedError
if cmethod == 'lz4':
error = ValueError
with raises(error) as excinfo:
numpy_pickle.dump(obj,
dump_filename,
compress=(cmethod, compress))
excinfo.match(msg)
else:
numpy_pickle.dump(obj, dump_filename, compress=(cmethod, compress))
# Verify the file contains the right magic number
with open(dump_filename, 'rb') as f:
assert _detect_compressor(f) == cmethod
# Verify the reloaded object is correct
obj_reloaded = numpy_pickle.load(dump_filename)
assert isinstance(obj_reloaded, type(obj))
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
else:
assert obj_reloaded == obj
def test_numpy_persistence():
filename = env['filename']
rnd = np.random.RandomState(0)
a = rnd.random_sample((10, 2))
for compress in (False, True, 0, 3):
# We use 'a.T' to have a non C-contiguous array.
for index, obj in enumerate(((a,), (a.T,), (a, a), [a, a, a])):
# Change the file name to avoid side effects between tests
this_filename = filename + str(random.randint(0, 1000))
filenames = numpy_pickle.dump(obj, this_filename,
compress=compress)
# All is cached in one file
nose.tools.assert_equal(len(filenames), 1)
# Check that only one file was created
nose.tools.assert_equal(filenames[0], this_filename)
# Check that this file does exist
nose.tools.assert_true(
os.path.exists(os.path.join(env['dir'], filenames[0])))
# Unpickle the object
obj_ = numpy_pickle.load(this_filename)
# Check that the items are indeed arrays
for item in obj_:
nose.tools.assert_true(isinstance(item, np.ndarray))
# And finally, check that all the values are equal.
np.testing.assert_array_equal(np.array(obj), np.array(obj_))
# Now test with array subclasses
for obj in (np.matrix(np.zeros(10)),
np.memmap(filename + str(random.randint(0, 1000)) + 'mmap',
mode='w+', shape=4, dtype=np.float)):
this_filename = filename + str(random.randint(0, 1000))
filenames = numpy_pickle.dump(obj, this_filename,
compress=compress)
# All is cached in one file
nose.tools.assert_equal(len(filenames), 1)
obj_ = numpy_pickle.load(this_filename)
if (type(obj) is not np.memmap and
hasattr(obj, '__array_prepare__')):
# We don't reconstruct memmaps
nose.tools.assert_true(isinstance(obj_, type(obj)))
np.testing.assert_array_equal(obj_, obj)
# Test with an object containing multiple numpy arrays
obj = ComplexTestObject()
filenames = numpy_pickle.dump(obj, this_filename,
compress=compress)
# All is cached in one file
nose.tools.assert_equal(len(filenames), 1)
obj_loaded = numpy_pickle.load(this_filename)
nose.tools.assert_true(isinstance(obj_loaded, type(obj)))
np.testing.assert_array_equal(obj_loaded.array_float, obj.array_float)
np.testing.assert_array_equal(obj_loaded.array_int, obj.array_int)
np.testing.assert_array_equal(obj_loaded.array_obj, obj.array_obj)
def test_memmap_persistence():
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
nose.tools.assert_true(isinstance(b, np.memmap))
def test_memmap_persistence():
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
if [int(x) for x in np.__version__.split('.', 2)[:2]] >= [1, 3]:
nose.tools.assert_true(isinstance(b, np.memmap))
def persist_input(self, args_tuple, kwargs_dict, filtered_args_dict):
DirectoryJob.persist_input(self, args_tuple, kwargs_dict,
filtered_args_dict)
call_info = dict(func=self.func,
version_info=self.func.version_info,
args=args_tuple,
kwargs=kwargs_dict)
numpy_pickle.dump(call_info, pjoin(self._work_path, 'input.pkl'))
def test_memmap_persistence():
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
nose.tools.assert_true(isinstance(b, np.memmap))
def test_standard_types(tmpdir, compress, member):
# Test pickling and saving with standard types.
filename = tmpdir.join('test.pkl').strpath
numpy_pickle.dump(member, filename, compress=compress)
_member = numpy_pickle.load(filename)
# We compare the pickled instance to the reloaded one only if it
# can be compared to a copied one
if member == copy.deepcopy(member):
assert member == _member
def test_masked_array_persistence():
# The special-case picker fails, because saving masked_array
# not implemented, but it just delegates to the standard pickler.
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
a = np.ma.masked_greater(a, 0.5)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
nose.tools.assert_true(isinstance(b, np.ma.masked_array))
def atomic_pickle(data, path, filename):
fd, workfile = tempfile.mkstemp(prefix=filename + '-', dir=path)
try:
os.close(fd)
numpy_pickle.dump(data, workfile)
os.rename(workfile, pjoin(path, filename))
except:
if os.path.exists(workfile):
os.unlink(workfile)
raise
def atomic_pickle(data, path, filename):
fd, workfile = tempfile.mkstemp(prefix=filename + '-', dir=path)
try:
os.close(fd)
numpy_pickle.dump(data, workfile)
os.rename(workfile, pjoin(path, filename))
except:
if os.path.exists(workfile):
os.unlink(workfile)
raise
def test_in_memory_persistence():
objs = [np.random.random((10, 10)), "some data", np.matrix([0, 1, 2])]
for obj in objs:
f = io.BytesIO()
numpy_pickle.dump(obj, f)
obj_reloaded = numpy_pickle.load(f)
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
else:
assert obj_reloaded == obj
def test_masked_array_persistence():
# The special-case picker fails, because saving masked_array
# not implemented, but it just delegates to the standard pickler.
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
a = np.ma.masked_greater(a, 0.5)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
assert isinstance(b, np.ma.masked_array)
def test_masked_array_persistence(tmpdir):
# The special-case picker fails, because saving masked_array
# not implemented, but it just delegates to the standard pickler.
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
a = np.ma.masked_greater(a, 0.5)
filename = tmpdir.join('test.pkl').strpath
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
assert isinstance(b, np.ma.masked_array)
def test_file_handle_persistence_mmap(tmpdir):
obj = np.random.random((10, 10))
filename = tmpdir.join('test.pkl').strpath
with open(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
with open(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f, mmap_mode='r+')
np.testing.assert_array_equal(obj_reloaded, obj)
def test_lz4_compression_without_lz4(tmpdir):
# Check that lz4 cannot be used when dependency is not available.
fname = tmpdir.join('test.nolz4').strpath
data = 'test data'
with raises(ValueError) as excinfo:
numpy_pickle.dump(data, fname, compress='lz4')
excinfo.match(LZ4_NOT_INSTALLED_ERROR)
with raises(ValueError) as excinfo:
numpy_pickle.dump(data, fname + '.lz4')
excinfo.match(LZ4_NOT_INSTALLED_ERROR)
def test_file_handle_persistence_mmap():
obj = np.random.random((10, 10))
filename = env['filename'] + str(random.randint(0, 1000))
with open(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
with open(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f, mmap_mode='r+')
np.testing.assert_array_equal(obj_reloaded, obj)
def test_file_handle_persistence_mmap():
obj = np.random.random((10, 10))
filename = env['filename'] + str(random.randint(0, 1000))
with open(filename, 'wb') as f:
numpy_pickle.dump(obj, f)
with open(filename, 'rb') as f:
obj_reloaded = numpy_pickle.load(f, mmap_mode='r+')
np.testing.assert_array_equal(obj_reloaded, obj)
def test_pickle_highest_protocol():
# ensure persistence of a numpy array is valid even when using
# the pickle HIGHEST_PROTOCOL.
# see https://github.com/joblib/joblib/issues/362
filename = env['filename'] + str(random.randint(0, 1000))
test_array = np.zeros(10)
numpy_pickle.dump(test_array, filename, protocol=pickle.HIGHEST_PROTOCOL)
array_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(array_reloaded, test_array)
def test_pathlib():
try:
from pathlib import Path
except ImportError:
pass
else:
filename = env['filename']
value = 123
numpy_pickle.dump(value, Path(filename))
nose.tools.assert_equal(numpy_pickle.load(filename), value)
numpy_pickle.dump(value, filename)
nose.tools.assert_equal(numpy_pickle.load(Path(filename)), value)
def test_in_memory_persistence():
objs = [np.random.random((10, 10)),
"some data",
np.matrix([0, 1, 2])]
for obj in objs:
f = io.BytesIO()
numpy_pickle.dump(obj, f)
obj_reloaded = numpy_pickle.load(f)
if isinstance(obj, np.ndarray):
np.testing.assert_array_equal(obj_reloaded, obj)
else:
assert obj_reloaded == obj
def test_memmap_with_padding(tmpdir):
# Test that memmaped arrays return by numpy.load are correctly aligned
fname = tmpdir.join('test.mmap').strpath
arr = np.random.randn(10)
numpy_pickle.dump(arr, fname)
memmap = numpy_pickle.load(fname, mmap_mode='r')
assert isinstance(memmap, np.memmap)
np.testing.assert_array_equal(arr, memmap)
assert isinstance(memmap.ctypes.data / 8, int)
def test_pathlib():
try:
from pathlib import Path
except ImportError:
pass
else:
filename = env['filename']
value = 123
numpy_pickle.dump(value, Path(filename))
assert numpy_pickle.load(filename) == value
numpy_pickle.dump(value, filename)
assert numpy_pickle.load(Path(filename)) == value
def test_pickle_highest_protocol(tmpdir):
# ensure persistence of a numpy array is valid even when using
# the pickle HIGHEST_PROTOCOL.
# see https://github.com/joblib/joblib/issues/362
filename = tmpdir.join('test.pkl').strpath
test_array = np.zeros(10)
numpy_pickle.dump(test_array, filename, protocol=pickle.HIGHEST_PROTOCOL)
array_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(array_reloaded, test_array)
def test_pathlib(tmpdir):
try:
from pathlib import Path
except ImportError:
pass
else:
filename = tmpdir.join('test.pkl').strpath
value = 123
numpy_pickle.dump(value, Path(filename))
assert numpy_pickle.load(filename) == value
numpy_pickle.dump(value, filename)
assert numpy_pickle.load(Path(filename)) == value
def test_pickle_highest_protocol():
# ensure persistence of a numpy array is valid even when using
# the pickle HIGHEST_PROTOCOL.
# see https://github.com/joblib/joblib/issues/362
filename = env['filename'] + str(random.randint(0, 1000))
test_array = np.zeros(10)
numpy_pickle.dump(test_array, filename, protocol=pickle.HIGHEST_PROTOCOL)
array_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(array_reloaded, test_array)
def test_pathlib():
try:
from pathlib import Path
except ImportError:
pass
else:
filename = env['filename']
value = 123
numpy_pickle.dump(value, Path(filename))
nose.tools.assert_equal(numpy_pickle.load(filename), value)
numpy_pickle.dump(value, filename)
nose.tools.assert_equal(numpy_pickle.load(Path(filename)), value)
def test_pathlib():
try:
from pathlib import Path
except ImportError:
pass
else:
filename = env['filename']
value = 123
numpy_pickle.dump(value, Path(filename))
assert numpy_pickle.load(filename) == value
numpy_pickle.dump(value, filename)
assert numpy_pickle.load(Path(filename)) == value
def test_load_memmap_with_big_offset(tmpdir):
# Test that numpy memmap offset is set correctly if greater than
# mmap.ALLOCATIONGRANULARITY, see
# https://github.com/joblib/joblib/issues/451 and
# https://github.com/numpy/numpy/pull/8443 for more details.
fname = tmpdir.join('test.mmap').strpath
size = mmap.ALLOCATIONGRANULARITY
obj = [np.zeros(size, dtype='uint8'), np.ones(size, dtype='uint8')]
numpy_pickle.dump(obj, fname)
memmaps = numpy_pickle.load(fname, mmap_mode='r')
assert isinstance(memmaps[1], np.memmap)
assert memmaps[1].offset > size
np.testing.assert_array_equal(obj, memmaps)
def test_standard_types():
# Test pickling and saving with standard types.
filename = env['filename']
for compress in [0, 1]:
for member in typelist:
# Change the file name to avoid side effects between tests
this_filename = filename + str(random.randint(0, 1000))
numpy_pickle.dump(member, this_filename, compress=compress)
_member = numpy_pickle.load(this_filename)
# We compare the pickled instance to the reloaded one only if it
# can be compared to a copied one
if member == copy.deepcopy(member):
yield assert_equal, member, _member
def test_non_contiguous_array_pickling():
filename = env['filename'] + str(random.randint(0, 1000))
for array in [ # Array that triggers a contiguousness issue with nditer,
# see https://github.com/joblib/joblib/pull/352 and see
# https://github.com/joblib/joblib/pull/353
np.asfortranarray([[1, 2], [3, 4]])[1:],
# Non contiguous array with works fine with nditer
np.ones((10, 50, 20), order='F')[:, :1, :]]:
assert not array.flags.c_contiguous
assert not array.flags.f_contiguous
numpy_pickle.dump(array, filename)
array_reloaded = numpy_pickle.load(filename)
np.testing.assert_array_equal(array_reloaded, array)
os.remove(filename)
def test_numpy_persistence():
filename = env['filename']
rnd = np.random.RandomState(0)
a = rnd.random_sample((10, 2))
for compress, cache_size in ((0, 0), (1, 0), (1, 10)):
# We use 'a.T' to have a non C-contiguous array.
for index, obj in enumerate(((a,), (a.T,), (a, a), [a, a, a])):
# Change the file name to avoid side effects between tests
this_filename = filename + str(random.randint(0, 1000))
filenames = numpy_pickle.dump(obj, this_filename,
compress=compress,
cache_size=cache_size)
# Check that one file was created per array
if not compress:
nose.tools.assert_equal(len(filenames), len(obj) + 1)
# Check that these files do exist
for file in filenames:
nose.tools.assert_true(
os.path.exists(os.path.join(env['dir'], file)))
# Unpickle the object
obj_ = numpy_pickle.load(this_filename)
# Check that the items are indeed arrays
for item in obj_:
nose.tools.assert_true(isinstance(item, np.ndarray))
# And finally, check that all the values are equal.
nose.tools.assert_true(np.all(np.array(obj) ==
np.array(obj_)))
# Now test with array subclasses
for obj in (
np.matrix(np.zeros(10)),
np.core.multiarray._reconstruct(np.memmap, (), np.float)
):
this_filename = filename + str(random.randint(0, 1000))
filenames = numpy_pickle.dump(obj, this_filename,
compress=compress,
cache_size=cache_size)
obj_ = numpy_pickle.load(this_filename)
if (type(obj) is not np.memmap
and hasattr(obj, '__array_prepare__')):
# We don't reconstruct memmaps
nose.tools.assert_true(isinstance(obj_, type(obj)))
# Finally smoke test the warning in case of compress + mmap_mode
this_filename = filename + str(random.randint(0, 1000))
numpy_pickle.dump(a, this_filename, compress=1)
numpy_pickle.load(this_filename, mmap_mode='r')
def test_compressed_pickle_dump_and_load():
# XXX: temporarily disable this test on non little-endian machines
if sys.byteorder != 'little':
raise nose.SkipTest('Skipping this test on non little-endian machines')
expected_list = [np.arange(5, dtype=np.dtype('<i8')),
np.arange(5, dtype=np.dtype('<f8')),
np.array([1, 'abc', {'a': 1, 'b': 2}], dtype='O'),
# .tostring actually returns bytes and is a
# compatibility alias for .tobytes which was
# added in 1.9.0
np.arange(256, dtype=np.uint8).tostring(),
u"C'est l'\xe9t\xe9 !"]
with tempfile.NamedTemporaryFile(suffix='.gz', dir=env['dir']) as f:
fname = f.name
# Need to test both code branches (whether array size is greater
# or smaller than cache_size)
for cache_size in [0, 1e9]:
try:
dumped_filenames = numpy_pickle.dump(
expected_list, fname, compress=1,
cache_size=cache_size)
result_list = numpy_pickle.load(fname)
for result, expected in zip(result_list, expected_list):
if isinstance(expected, np.ndarray):
nose.tools.assert_equal(result.dtype, expected.dtype)
np.testing.assert_equal(result, expected)
else:
nose.tools.assert_equal(result, expected)
finally:
for fn in dumped_filenames:
os.remove(fn)
def test_compressed_pickle_dump_and_load():
expected_list = [np.arange(5, dtype=np.dtype('<i8')),
np.arange(5, dtype=np.dtype('>i8')),
np.arange(5, dtype=np.dtype('<f8')),
np.arange(5, dtype=np.dtype('>f8')),
np.array([1, 'abc', {'a': 1, 'b': 2}], dtype='O'),
# .tostring actually returns bytes and is a
# compatibility alias for .tobytes which was
# added in 1.9.0
np.arange(256, dtype=np.uint8).tostring(),
# np.matrix is a subclass of np.ndarray, here we want
# to verify this type of object is correctly unpickled
# among versions.
np.matrix([0, 1, 2], dtype=np.dtype('<i8')),
np.matrix([0, 1, 2], dtype=np.dtype('>i8')),
u"C'est l'\xe9t\xe9 !"]
with tempfile.NamedTemporaryFile(suffix='.gz', dir=env['dir']) as f:
fname = f.name
try:
dumped_filenames = numpy_pickle.dump(expected_list, fname, compress=1)
assert len(dumped_filenames) == 1
result_list = numpy_pickle.load(fname)
for result, expected in zip(result_list, expected_list):
if isinstance(expected, np.ndarray):
assert result.dtype == expected.dtype
np.testing.assert_equal(result, expected)
else:
assert result == expected
finally:
os.remove(fname)
def test_file_handle_persistence_in_memory_mmap():
obj = np.random.random((10, 10))
buf = io.BytesIO()
numpy_pickle.dump(obj, buf)
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.load(buf, mmap_mode='r+')
assert len(caught_warnings) == 1
for warn in caught_warnings:
assert warn.category == UserWarning
assert (warn.message.args[0] ==
'In memory persistence is not compatible with '
'mmap_mode "%(mmap_mode)s" flag passed. mmap_mode '
'option will be ignored.' % {'mmap_mode': 'r+'})
def test_memmap_persistence_mixed_dtypes():
# loading datastructures that have sub-arrays with dtype=object
# should not prevent memmaping on fixed size dtype sub-arrays.
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
b = np.array([1, 'b'], dtype=object)
construct = (a, b)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(construct, filename)
a_clone, b_clone = numpy_pickle.load(filename, mmap_mode='r')
# the floating point array has been memory mapped
assert isinstance(a_clone, np.memmap)
# the object-dtype array has been loaded in memory
assert not isinstance(b_clone, np.memmap)
def test_memmap_persistence():
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, filename)
b = numpy_pickle.load(filename, mmap_mode='r')
assert isinstance(b, np.memmap)
# Test with an object containing multiple numpy arrays
filename = env['filename'] + str(random.randint(0, 1000))
obj = ComplexTestObject()
numpy_pickle.dump(obj, filename)
obj_loaded = numpy_pickle.load(filename, mmap_mode='r')
assert isinstance(obj_loaded, type(obj))
assert isinstance(obj_loaded.array_float, np.memmap)
assert not obj_loaded.array_float.flags.writeable
assert isinstance(obj_loaded.array_int, np.memmap)
assert not obj_loaded.array_int.flags.writeable
# Memory map not allowed for numpy object arrays
assert not isinstance(obj_loaded.array_obj, np.memmap)
np.testing.assert_array_equal(obj_loaded.array_float,
obj.array_float)
np.testing.assert_array_equal(obj_loaded.array_int,
obj.array_int)
np.testing.assert_array_equal(obj_loaded.array_obj,
obj.array_obj)
# Test we can write in memmaped arrays
obj_loaded = numpy_pickle.load(filename, mmap_mode='r+')
assert obj_loaded.array_float.flags.writeable
obj_loaded.array_float[0:10] = 10.0
assert obj_loaded.array_int.flags.writeable
obj_loaded.array_int[0:10] = 10
obj_reloaded = numpy_pickle.load(filename, mmap_mode='r')
np.testing.assert_array_equal(obj_reloaded.array_float,
obj_loaded.array_float)
np.testing.assert_array_equal(obj_reloaded.array_int,
obj_loaded.array_int)
# Test w+ mode is caught and the mode has switched to r+
numpy_pickle.load(filename, mmap_mode='w+')
assert obj_loaded.array_int.flags.writeable
assert obj_loaded.array_int.mode == 'r+'
assert obj_loaded.array_float.flags.writeable
assert obj_loaded.array_float.mode == 'r+'
def test_compress_mmap_mode_warning():
# Test the warning in case of compress + mmap_mode
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
this_filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, this_filename, compress=1)
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.load(this_filename, mmap_mode='r+')
assert len(caught_warnings) == 1
for warn in caught_warnings:
assert warn.category == UserWarning
assert (warn.message.args[0] ==
'mmap_mode "%(mmap_mode)s" is not compatible with '
'compressed file %(filename)s. "%(mmap_mode)s" flag will '
'be ignored.' % {'filename': this_filename,
'mmap_mode': 'r+'})
def test_file_handle_persistence_compressed_mmap():
obj = np.random.random((10, 10))
filename = env['filename'] + str(random.randint(0, 1000))
with open(filename, 'wb') as f:
numpy_pickle.dump(obj, f, compress=('gzip', 3))
with closing(gzip.GzipFile(filename, 'rb')) as f:
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.load(f, mmap_mode='r+')
assert len(caught_warnings) == 1
for warn in caught_warnings:
assert warn.category == UserWarning
assert (warn.message.args[0] ==
'"%(fileobj)r" is not a raw file, mmap_mode '
'"%(mmap_mode)s" flag will be ignored.'
% {'fileobj': f, 'mmap_mode': 'r+'})
def test_cache_size_warning():
# Check deprecation warning raised when cache size is not None
filename = env['filename'] + str(random.randint(0, 1000))
rnd = np.random.RandomState(0)
a = rnd.random_sample((10, 2))
for cache_size in (None, 0, 10):
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.dump(a, filename, cache_size=cache_size)
expected_nb_warnings = 1 if cache_size is not None else 0
assert len(caught_warnings) == expected_nb_warnings
for warn in caught_warnings:
assert warn.category == DeprecationWarning
assert (warn.message.args[0] ==
"Please do not set 'cache_size' in joblib.dump, this "
"parameter has no effect and will be removed. You "
"used 'cache_size={0}'".format(cache_size))
def test_compressed_pickle_dump_and_load():
expected_list = [np.arange(5, dtype=np.int64),
np.arange(5, dtype=np.float64),
# .tostring actually returns bytes and is a
# compatibility alias for .tobytes which was
# added in 1.9.0
np.arange(256, dtype=np.uint8).tostring(),
u"C'est l'\xe9t\xe9 !"]
with tempfile.NamedTemporaryFile(suffix='.gz', dir=env['dir']) as f:
numpy_pickle.dump(expected_list, f.name, compress=1)
result_list = numpy_pickle.load(f.name)
for result, expected in zip(result_list, expected_list):
if isinstance(expected, np.ndarray):
nose.tools.assert_equal(result.dtype, expected.dtype)
np.testing.assert_equal(result, expected)
else:
nose.tools.assert_equal(result, expected)
def test_compress_mmap_mode_warning():
# Test the warning in case of compress + mmap_mode
rnd = np.random.RandomState(0)
a = rnd.random_sample(10)
this_filename = env['filename'] + str(random.randint(0, 1000))
numpy_pickle.dump(a, this_filename, compress=1)
with warnings.catch_warnings(record=True) as caught_warnings:
warnings.simplefilter("always")
numpy_pickle.load(this_filename, mmap_mode='r+')
nose.tools.assert_equal(len(caught_warnings), 1)
for warn in caught_warnings:
nose.tools.assert_equal(warn.category, DeprecationWarning)
nose.tools.assert_equal(warn.message.args[0],
'File "%(filename)s" is compressed using '
'"%(compressor)s" which is not compatible '
'with mmap_mode "%(mmap_mode)s" flag '
'passed.' % {'filename': this_filename,
'mmap_mode': 'r+',
'compressor': 'zlib'})
