def test_serialize_pandas_no_preserve_index():
df = pd.DataFrame({'a': [1, 2, 3]}, index=[1, 2, 3])
expected = pd.DataFrame({'a': [1, 2, 3]})
buf = pa.serialize_pandas(df, preserve_index=False)
result = pa.deserialize_pandas(buf)
assert_frame_equal(result, expected)
buf = pa.serialize_pandas(df, preserve_index=True)
result = pa.deserialize_pandas(buf)
assert_frame_equal(result, df)
async def post_handler(request):
# print(request.body)
# bb = io.BytesIO(request.body)
bb = pa.py_buffer(request.body)
# br = pa.BufferReader(request.body)
df = pa.deserialize_pandas(bb)
print(df.shape)
return raw(b'HI')
def test_pandas_serialize_round_trip_nthreads():
index = pd.Index([1, 2, 3], name='my_index')
columns = ['foo', 'bar']
df = pd.DataFrame(
{'foo': [1.5, 1.6, 1.7], 'bar': list('abc')},
index=index, columns=columns
)
buf = pa.serialize_pandas(df)
result = pa.deserialize_pandas(buf, nthreads=2)
assert_frame_equal(result, df)
def test_pandas_serialize_round_trip_multi_index():
index1 = pd.Index([1, 2, 3], name='level_1')
index2 = pd.Index(list('def'), name=None)
index = pd.MultiIndex.from_arrays([index1, index2])
columns = ['foo', 'bar']
df = pd.DataFrame(
{'foo': [1.5, 1.6, 1.7], 'bar': list('abc')},
index=index,
columns=columns,
)
buf = pa.serialize_pandas(df)
result = pa.deserialize_pandas(buf)
assert_frame_equal(result, df)
def get(cls, list_quandle_series, start_date, end_date):
r = requests.get(
'http://localhost:{}/v1.0/data/multiple'.format(_port),
params={
'list_series': list_quandle_series,
'start_date': start_date,
'end_date': end_date
})
if r.status_code == 200:
return pyarrow.deserialize_pandas(r.content)
else:
raise Exception("unable to acquire data: HTTP code: ".format(
r.status_code))
def test_pandas_serialize_round_trip_not_string_columns():
df = pd.DataFrame(list(zip([1.5, 1.6, 1.7], 'abc')))
buf = pa.serialize_pandas(df)
result = pa.deserialize_pandas(buf)
assert_frame_equal(result, df)
def time_deserialize_pandas(self):
pa.deserialize_pandas(self.serialized)
def test_pandas_serialize_round_trip_not_string_columns():
df = pd.DataFrame(list(zip([1.5, 1.6, 1.7], 'abc')))
buf = pa.serialize_pandas(df)
result = pa.deserialize_pandas(buf)
assert_frame_equal(result, df)
def _check_serialize_pandas_round_trip(df, nthreads=1):
buf = pa.serialize_pandas(df, nthreads=nthreads)
result = pa.deserialize_pandas(buf, nthreads=nthreads)
assert_frame_equal(result, df)
def _check_serialize_pandas_round_trip(df, use_threads=False):
buf = pa.serialize_pandas(df, nthreads=2 if use_threads else 1)
result = pa.deserialize_pandas(buf, use_threads=use_threads)
assert_frame_equal(result, df)
def _deserialize_pandas_series(data):
deserialized = deserialize_pandas(data)
return deserialized[deserialized.columns[0]]
def _deserialize_pandas_dataframe(data):
return deserialize_pandas(data)
def _deserialize_pandas_dataframe(data):
return deserialize_pandas(data)
def _deserialize_pandas_series(data):
deserialized = deserialize_pandas(data)
return deserialized[deserialized.columns[0]]
def _check_serialize_pandas_round_trip(df, use_threads=False):
buf = pa.serialize_pandas(df, nthreads=2 if use_threads else 1)
result = pa.deserialize_pandas(buf, use_threads=use_threads)
assert_frame_equal(result, df)
def deserialize(b):
try:
obj = pickle.loads(b)
except pickle.UnpicklingError:
obj = pa.deserialize_pandas(b)
return obj
