def test_serialize_to_components_invalid_cases():
buf = pa.py_buffer(b'hello')
components = {
'num_tensors': 0,
'num_sparse_tensors': {
'coo': 0, 'csr': 0, 'csc': 0, 'csf': 0, 'ndim_csf': 0
},
'num_ndarrays': 0,
'num_buffers': 1,
'data': [buf]
}
with pytest.raises(pa.ArrowInvalid):
pa.deserialize_components(components)
components = {
'num_tensors': 0,
'num_sparse_tensors': {
'coo': 0, 'csr': 0, 'csc': 0, 'csf': 0, 'ndim_csf': 0
},
'num_ndarrays': 1,
'num_buffers': 0,
'data': [buf, buf]
}
with pytest.raises(pa.ArrowInvalid):
pa.deserialize_components(components)
def test_serialize_to_components_invalid_cases():
buf = pa.frombuffer(b'hello')
components = {'num_tensors': 0, 'num_buffers': 1, 'data': [buf]}
with pytest.raises(pa.ArrowException):
pa.deserialize_components(components)
components = {'num_tensors': 1, 'num_buffers': 0, 'data': [buf, buf]}
with pytest.raises(pa.ArrowException):
pa.deserialize_components(components)
def loads(buf):
mv = memoryview(buf)
header = read_file_header(mv)
compress = header.compress
if compress == CompressType.NONE:
data = buf[HEADER_LENGTH:]
else:
data = decompressors[compress](mv[HEADER_LENGTH:])
if header.type == SerialType.ARROW:
try:
return pyarrow.deserialize(memoryview(data),
mars_serialize_context())
except pyarrow.lib.ArrowInvalid: # pragma: no cover
# reconstruct value from buffers of arrow components
data_view = memoryview(data)
meta_block_size = np.frombuffer(data_view[0:4],
dtype='int32').item()
meta = pickle.loads(data_view[4:4 + meta_block_size]) # nosec
buffer_sizes = meta.pop('buffer_sizes')
bounds = np.cumsum([4 + meta_block_size] + buffer_sizes)
meta['data'] = [
pyarrow.py_buffer(data_view[bounds[idx]:bounds[idx + 1]])
for idx in range(len(buffer_sizes))
]
return pyarrow.deserialize_components(meta,
mars_serialize_context())
else:
return pickle.loads(data)
def test_serialize_to_components_invalid_cases():
buf = pa.py_buffer(b'hello')
components = {
'num_tensors': 0,
'num_buffers': 1,
'data': [buf]
}
with pytest.raises(pa.ArrowInvalid):
pa.deserialize_components(components)
components = {
'num_tensors': 1,
'num_buffers': 0,
'data': [buf, buf]
}
with pytest.raises(pa.ArrowInvalid):
pa.deserialize_components(components)
def get(self, key, fill_cache_func):
value = self._cache.get(key, default=None)
if value is None:
value = fill_cache_func()
table_pandas = value.to_pandas()
serialized_df = pa.serialize(table_pandas)
components = serialized_df.to_components()
self._cache.set(key, components)
else:
original_df = pa.deserialize_components(value)
value = pa.Table.from_pandas(original_df, preserve_index=False)
return value
def time_deserialize_from_components(self):
pa.deserialize_components(self.as_components)
def __call__(self, data):
components = pickle.loads(data[0])
data = list(map(pa.py_buffer, data[1:]))
components['data'] = data
return pa.deserialize_components(components, context=self.context)
print("Time for serializing array in-memory (pickle4, copy): %.3fs" %
(t1 - t0))
t0 = time()
carr2 = cat.from_sframe(sframe_nocopy, copy=False)
t1 = time()
print("Time for de-serializing array in-memory (caterva, no-copy): %.3fs" %
(t1 - t0))
if check_roundtrip:
print("The roundtrip is... ", end="", flush=True)
np.testing.assert_allclose(carr2, arr)
print("ok!")
t0 = time()
arr2 = pa.deserialize_components(pyarrow_nocopy)
t1 = time()
print("Time for de-serializing array in-memory (arrow, no-copy): %.3fs" %
(t1 - t0))
if check_roundtrip:
print("The roundtrip is... ", end="", flush=True)
np.testing.assert_allclose(arr2, arr)
print("ok!")
t0 = time()
arr2 = pa.deserialize(pyarrow_copy)
t1 = time()
print("Time for de-serializing array in-memory (arrow, copy): %.3fs" %
(t1 - t0))
def deserialize(data):
symbol_infos = pa.deserialize_components(data)
return Hishty(start=min(symbol_infos), end=max(symbol_infos), symbol_infos=symbol_infos)
