def test_stream_dict_to_ds_created_from_df(self, pd):
data = pd.DataFrame(dict(a=[10], b=[20], c=[30])).set_index('c')
ds = ColumnDataSource(data)
ds._document = "doc"
notify_owners_stuff = {}
def notify_owners_mock(*args, **kw):
notify_owners_stuff['args'] = args
notify_owners_stuff['kw'] = kw
ds.data._notify_owners = notify_owners_mock
stream_stuff = {}
data_stream = ds.data._stream
def stream_wrapper(*args, **kwargs):
stream_stuff['args'] = args
stream_stuff['kwargs'] = kwargs
data_stream(*args, **kwargs)
ds.data._stream = stream_wrapper
ds._stream(
dict(a=[11, 12], b=np.array([21, 22]), c=pd.Series([31, 32])), 7)
assert len(stream_stuff['args']) == 5
expected_stream_args = ("doc", ds,
dict(a=[11, 12],
b=np.array([21, 22]),
c=pd.Series([31, 32])), 7, None)
for i, (arg, ex_arg) in enumerate(
zip(stream_stuff['args'], expected_stream_args)):
if i == 2:
assert arg['a'] == ex_arg['a']
del arg['a'], ex_arg['a']
self._assert_equal_dicts_of_arrays(arg, ex_arg)
else:
assert arg == ex_arg
assert stream_stuff['kwargs'] == {}
assert len(notify_owners_stuff['args']) == 1
self._assert_equal_dicts_of_arrays(
notify_owners_stuff['args'][0],
dict(a=np.array([10]), b=np.array([20]), c=np.array([30])))
self._assert_equal_dicts_of_arrays(
dict(ds.data),
dict(a=np.array([10, 11, 12]),
b=np.array([20, 21, 22]),
c=np.array([30, 31, 32])))
def test_transform_series_force_list_true(pd):
df = pd.Series([1, 3, 5, 6, 8])
out = bus.transform_series(df, force_list=True)
assert isinstance(out, list)
df = pd.Series([1, 3, 5, 6, 8], dtype=np.int32)
out = bus.transform_series(df, force_list=True)
assert isinstance(out, list)
df = pd.Series([1.0, 3, 5, 6, 8])
out = bus.transform_series(df, force_list=True)
assert isinstance(out, list)
df = pd.Series(np.array([np.nan, np.inf, -np.inf, 0]))
out = bus.transform_series(df, force_list=True)
assert isinstance(out, list)
def test_nans_and_infs_pandas(self, pd):
arr = pd.Series(np.array([np.nan, np.inf, -np.inf, 0]))
serialized = self.serialize(arr)
deserialized = self.deserialize(serialized)
assert deserialized[0] == 'NaN'
assert deserialized[1] == 'Infinity'
assert deserialized[2] == '-Infinity'
assert deserialized[3] == 0
def test_transform_series_force_list_default_with_buffers(pd):
# default int seems to be int64, can't be converted to buffer!
df = pd.Series([1, 3, 5, 6, 8])
out = bus.transform_series(df)
assert isinstance(out, list)
assert out == [1, 3, 5, 6, 8]
df = pd.Series([1, 3, 5, 6, 8], dtype=np.int32)
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
df = pd.Series([1.0, 3, 5, 6, 8])
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
df = pd.Series(np.array([np.nan, np.inf, -np.inf, 0]))
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
def test_transform_series_force_list_default(pd):
# default int seems to be int64, can't be encoded!
df = pd.Series([1, 3, 5, 6, 8])
out = bus.transform_series(df)
assert isinstance(out, list)
assert out == [1, 3, 5, 6, 8]
df = pd.Series([1, 3, 5, 6, 8], dtype=np.int32)
out = bus.transform_series(df)
assert isinstance(out, dict)
df = pd.Series([1.0, 3, 5, 6, 8])
out = bus.transform_series(df)
assert isinstance(out, dict)
df = pd.Series(np.array([np.nan, np.inf, -np.inf, 0]))
out = bus.transform_series(df)
assert isinstance(out, dict)
def test_stream_series_to_ds_created_from_df(self, pd):
data = pd.DataFrame(dict(a=[10], b=[20], c=[30]))
ds = ColumnDataSource(data)
ds._document = "doc"
notify_owners_stuff = {}
def notify_owners_mock(*args, **kw):
notify_owners_stuff['args'] = args
notify_owners_stuff['kw'] = kw
ds.data._notify_owners = notify_owners_mock
stream_stuff = {}
data_stream = ds.data._stream
def stream_wrapper(*args, **kwargs):
stream_stuff['args'] = args
stream_stuff['kwargs'] = kwargs
data_stream(*args, **kwargs)
ds.data._stream = stream_wrapper
ds._stream(pd.Series([11, 21, 31], index=list('abc')), 7)
assert len(stream_stuff['args']) == 5
expected_df = pd.DataFrame(
dict(a=np.array([11]), b=np.array([21]), c=np.array([31])))
expected_stream_data = expected_df.to_dict('series')
expected_stream_data['index'] = expected_df.index.values
expected_args = ("doc", ds, expected_stream_data, 7, None)
for i, (arg,
ex_arg) in enumerate(zip(stream_stuff['args'], expected_args)):
if i == 2:
self._assert_equal_dicts_of_arrays(arg, ex_arg)
else:
assert arg == ex_arg
assert stream_stuff['kwargs'] == {}
assert len(notify_owners_stuff['args']) == 1
self._assert_equal_dicts_of_arrays(
notify_owners_stuff['args'][0],
dict(a=np.array([10]),
b=np.array([20]),
c=np.array([30]),
index=np.array([0])))
self._assert_equal_dicts_of_arrays(
dict(ds.data),
dict(a=np.array([10, 11]),
b=np.array([20, 21]),
c=np.array([30, 31]),
index=np.array([0, 0])))
def test_transform_column_source_data_with_buffers(pd, cols, dt1, dt2):
d = dict(a=[1, 2, 3],
b=np.array([4, 5, 6], dtype=dt1),
c=pd.Series([7, 8, 9], dtype=dt2))
bufs = []
out = bus.transform_column_source_data(d, buffers=bufs, cols=cols)
assert set(out) == (set(d) if cols is None else set(cols))
if 'a' in out:
assert out['a'] == [1, 2, 3]
for x in ['b', 'c']:
dt = d[x].dtype
if x in out:
if dt in bus.BINARY_ARRAY_TYPES:
assert isinstance(out[x], dict)
assert 'shape' in out[x]
assert out[x]['shape'] == d[x].shape
assert 'dtype' in out[x]
assert out[x]['dtype'] == d[x].dtype.name
assert '__buffer__' in out[x]
else:
assert isinstance(out[x], list)
assert out[x] == list(d[x])
def test__get_range_with_too_long_series(pd):
with pytest.raises(ValueError):
_get_range(pd.Series([20, 30, 40]))
def test__get_range_with_series(pd):
r = _get_range(pd.Series([20, 30]))
assert isinstance(r, Range1d)
assert r.start == 20
assert r.end == 30
def test_transform_array_with_nans_to_list(pd, values):
s = pd.Series([np.nan, values[0], values[1]])
out = bus.transform_array_to_list(s)
assert isinstance(out, list)
assert out == ['NaN', values[0], values[1]]
def test_transform_series_force_list_default_with_buffers(pd):
# default int seems to be int64, can't be converted to buffer!
df = pd.Series([1, 3, 5, 6, 8])
out = bus.transform_series(df)
assert isinstance(out, list)
assert out == [1, 3, 5, 6, 8]
df = pd.Series([1, 3, 5, 6, 8], dtype=np.int32)
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
df = pd.Series([1.0, 3, 5, 6, 8])
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
df = pd.Series(np.array([np.nan, np.inf, -np.inf, 0]))
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == np.array(df).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == df.dtype.name
assert '__buffer__' in out
# PeriodIndex
df = pd.period_range('1900-01-01', '2000-01-01', freq='A')
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == bus.convert_datetime_array(
df.to_timestamp().values).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == 'float64'
assert '__buffer__' in out
# DatetimeIndex
df = pd.period_range('1900-01-01', '2000-01-01', freq='A').to_timestamp()
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == bus.convert_datetime_array(df.values).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == 'float64'
assert '__buffer__' in out
# TimeDeltaIndex
df = pd.to_timedelta(np.arange(5), unit='s')
bufs = []
out = bus.transform_series(df, buffers=bufs)
assert isinstance(out, dict)
assert len(bufs) == 1
assert len(bufs[0]) == 2
assert isinstance(bufs[0][0], dict)
assert list(bufs[0][0]) == ["id"]
assert bufs[0][1] == bus.convert_datetime_array(df.values).tobytes()
assert 'shape' in out
assert out['shape'] == df.shape
assert 'dtype' in out
assert out['dtype'] == 'float64'
assert '__buffer__' in out
def test_panda_series(self, pd):
s = pd.Series([1, 3, 5, 6, 8])
assert self.encoder.default(s) == [1, 3, 5, 6, 8]
def test_with_pd_series(self, pd):
s = pd.Series([0, 1, 2, 3, 4])
assert self.serialize(s) == '[0,1,2,3,4]'
