def dfs_arrow(tmp_path_factory):
tmpdir = tmp_path_factory.mktemp("vaex")
path = str(tmpdir / 'strings.hdf5')
df = vaex.from_arrays(s=vaex.string_column(string_list),
sr=vaex.string_column(string_list_reverse))
df.export(path) # we write it out so that the memory is read only
return vaex.open(path)
def create_base_ds():
dataset = vaex.dataset.DatasetArrays("dataset")
x = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
y = y = x**2
ints = np.arange(-2, 19, dtype="i8")
ints[0] = 2**62 + 1
ints[1] = -2**62 + 1
ints[2] = -2**62 - 1
ints[0 + 10] = 2**62 + 1
ints[1 + 10] = -2**62 + 1
ints[2 + 10] = -2**62 - 1
dataset.add_column("x", x)
dataset.add_column("y", y)
# m = x.copy()
m = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
ma_value = 77777
m[-1 + 10] = ma_value
m[-1 + 20] = ma_value
m = np.ma.array(m, mask=m == ma_value)
n = x.copy()
n[-2 + 10] = np.nan
n[-2 + 20] = np.nan
nm = x.copy()
nm[-2 + 10] = np.nan
nm[-2 + 20] = np.nan
nm[-1 + 10] = ma_value
nm[-1 + 20] = ma_value
nm = np.ma.array(nm, mask=nm == ma_value)
mi = np.ma.array(m.data.astype(np.int64),
mask=m.data == ma_value,
fill_value=88888)
dataset.add_column("m", m)
dataset.add_column('n', n)
dataset.add_column('nm', nm)
dataset.add_column("mi", mi)
dataset.add_column("ints", ints)
name = np.array(list(map(lambda x: str(x) + "bla" + ('_' * int(x)), x)),
dtype='U') #, dtype=np.string_)
dataset.add_column("name", np.array(name))
dataset.add_column("name_arrow", vaex.string_column(name))
obj_data = np.array([
'train', 'false', True, 1, 30., np.nan, 'something',
'something a bit longer resembling a sentence?!', -10000,
'this should be masked'
],
dtype='object')
obj_mask = np.array([False] * 9 + [True])
obj = nm.copy().astype('object')
obj[2:12] = np.ma.MaskedArray(data=obj_data, mask=obj_mask, dtype='object')
dataset.add_column("obj", obj)
return dataset
# dsf = create_filtered()
def test_arrow_strings():
N = 4
x = ['a', 'bb', 'ccc', 'dddd']
xc = vaex.string_column(x)
df = vaex.from_arrays(x=xc)
assert len(df.columns['x']) == 4
trimmed = df.columns['x'][2:4]
assert trimmed[:].tolist() == x[2:4]
assert trimmed[1:2].tolist() == x[3:4]
assert len(df) == N
assert len(df[1:3]) == 2
assert df[1:3].x.tolist() == x[1:3]
indices = np.array([0, 2, 1, 3])
assert xc.take(indices).tolist() == ['a', 'ccc', 'bb', 'dddd']
indices_masked = np.ma.array(indices, mask=[False, True, False, False])
assert xc.take(indices_masked).tolist() == ['a', None, 'bb', 'dddd']
indices = np.array([0, 2, 1, 3])
assert xc.take(indices).tolist() == ['a', 'ccc', 'bb', 'dddd']
mask = np.array([True, True, False, True])
assert vaex.array_types.filter(xc, mask).tolist() == ['a', 'bb', 'dddd']
mask_masked = np.ma.array(np.array([True, True, False, True]), mask=[False, True, True, False])
assert vaex.array_types.filter(xc, mask_masked).tolist() == ['a', 'dddd']
def test_arrow_strings_null():
N = 4
x = ['a', 'bb', None, 'dddd', None]
xc = vaex.string_column(x)
assert xc.tolist() == x
assert xc[1:].tolist() == x[1:]
assert xc[2:4].tolist() == x[2:4]
def test_unique_arrow(df_factory):
ds = df_factory(x=vaex.string_column(
['a', 'b', 'a', 'a', 'a', 'b', 'b', 'b', 'b', 'a']))
with small_buffer(ds, 2):
assert set(ds.unique(ds.x)) == {'a', 'b'}
values, index = ds.unique(ds.x, return_inverse=True)
assert np.array(values)[index].tolist() == ds.x.tolist()
def test_string_strip_special_case():
strings = [
"Explanation\nWhy the edits made under my username Hardcore Metallica Fan were reverted? "
"They weren't vandalisms, just closure on some GAs after I voted at New York Dolls FAC. "
"And please don't remove the template from the talk page since I'm retired now.89.205.38.27"
]
df = vaex.from_arrays(s=vaex.string_column(strings))
df.s.str.strip(' ').values # .get(0)
def test_arrow_strings():
N = 4
x = ['a', 'bb', 'ccc', 'dddd']
xc = vaex.string_column(x)
df = vaex.from_arrays(x=xc)
assert len(df.columns['x']) == 4
trimmed = df.columns['x'][2:4]
assert trimmed[:].tolist() == x[2:4]
assert len(df) == N
assert len(df[1:3]) == 2
assert df[1:3].x.tolist() == x[1:3]
def test_is_na():
s = vaex.string_column(["aap", None, "noot", "mies"])
o = ["aap", None, False, np.nan]
x = np.arange(4, dtype=np.float64)
x[2] = x[3] = np.nan
m = np.ma.array(x, mask=[0, 1, 0, 1])
df = vaex.from_arrays(x=x, m=m, s=s, o=o)
assert (df.x.isna().tolist() == [False, False, True, True])
assert (df.m.isna().tolist() == [False, True, True, True])
assert (df.s.isna().tolist() == [False, True, False, False])
assert (df.o.isna().tolist() == [False, True, False, True])
def test_arrow_strings():
N = 4
x = ['a', 'bb', 'ccc', 'dddd']
xc = vaex.string_column(x)
df = vaex.from_arrays(x=xc)
assert len(df.columns['x']) == 4
trimmed = df.columns['x'][2:4]
assert trimmed[:].tolist() == x[2:4]
assert len(df) == N
assert len(df[1:3]) == 2
assert df[1:3].x.tolist() == x[1:3]
def test_string_count_stat():
ds = vaex.from_arrays(names=['hello', 'this', 'is', 'long'])
assert ds.count(ds.names) == 4
ds = vaex.from_arrays(names=np.ma.array(['hello', 'this', 'is', 'long'], mask=[0, 0, 1, 0]))
assert ds.count(ds.names) == 3
df = vaex.from_arrays(names=np.array(['hi', 'is', 'l2', np.nan], dtype='O'))
assert df.count(ds.names) == 3
names = vaex.string_column(['hello', 'this', None, 'long'])
x = np.arange(len(names))
df = vaex.from_arrays(names=names, x=x)
assert df.count(df.names, binby='x', limits=[0, 100], shape=1).tolist() == [3]
def test_concat_mixed_types():
x1 = np.zeros(3) + np.nan
x2 = vaex.string_column(['hi', 'there'])
df1 = vaex.from_arrays(x=x1)
df2 = vaex.from_arrays(x=x2)
df = vaex.concat([df1, df2])
assert df2.x.dtype == df.x.dtype, "expect 'upcast' to string"
assert df[:2].x.tolist() == ['nan', 'nan']
assert df[1:4].x.tolist() == ['nan', 'nan', 'hi']
assert df[2:4].x.tolist() == ['nan', 'hi']
assert df[3:4].x.tolist() == ['hi']
assert df[3:5].x.tolist() == ['hi', 'there']
def test_unique_categorical(df_factory, future):
df = df_factory(x=vaex.string_column(['a', 'c', 'b', 'a', 'a']))
df = df.ordinal_encode('x')
df = df._future() if future else df
if future:
assert df.x.dtype == str
assert set(df.x.unique()) == {'a', 'b', 'c'}
assert df.x.nunique() == 3
else:
assert df.x.dtype == int
assert set(df.x.unique()) == {0, 1, 2}
assert df.x.nunique() == 3
def test_null_values():
df = vaex.from_arrays(s=vaex.string_column(['aap', None, 'mies']),
x=[0, 1, 2])
assert df.count() == 3
assert df.count(df.s) == 2
assert df.count(df.s, selection=df.x > 0) == 1
def dfs_arrow():
return vaex.from_arrays(s=vaex.string_column(string_list),
sr=vaex.string_column(string_list_reverse))
def create_base_ds():
x = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
columns = {'x': x}
y = y = x**2
ints = np.arange(-2, 19, dtype="i8")
ints[0] = 2**62 + 1
ints[1] = -2**62 + 1
ints[2] = -2**62 - 1
ints[0 + 10] = 2**62 + 1
ints[1 + 10] = -2**62 + 1
ints[2 + 10] = -2**62 - 1
columns["x"] = x
columns["y"] = y
# m = x.copy()
m = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
ma_value = 77777
m[-1 + 10 + 2] = ma_value
m[-1 + 20] = ma_value
m = np.ma.array(m, mask=m == ma_value)
n = x.copy()
n[-2 + 10] = np.nan
n[-2 + 20] = np.nan
nm = x.copy()
nm[-2 + 10] = np.nan
nm[-2 + 20] = np.nan
nm[-1 + 10] = ma_value
nm[-1 + 20] = ma_value
nm = np.ma.array(nm, mask=nm == ma_value)
mi = np.ma.array(m.data.astype(np.int64),
mask=m.data == ma_value,
fill_value=88888)
columns["m"] = m
columns['n'] = n
columns['nm'] = nm
columns["mi"] = mi
columns["ints"] = ints
name = np.array(list(map(lambda x: str(x) + "bla" + ('_' * int(x)), x)),
dtype='U') #, dtype=np.string_)
columns["name"] = np.array(name)
columns["name_arrow"] = vaex.string_column(name)
obj_data = np.array([
'train', 'false', True, 1, 30., np.nan, 'something',
'something a bit longer resembling a sentence?!', -10000,
'this should be masked'
],
dtype='object')
obj_mask = np.array([False] * 9 + [True])
obj = nm.copy().astype('object')
obj[2:12] = np.ma.MaskedArray(data=obj_data, mask=obj_mask, dtype='object')
columns["obj"] = obj #, dtype=np.dtype('O')
booleans = np.ones(21, dtype=np.bool)
booleans[[4, 6, 8, 14, 16, 19]] = False
columns["bool"] = booleans
datetime = np.array([
'2016-02-29T22:02:02.32',
'2013-01-17T01:02:03.32',
'2017-11-11T08:15:15.00',
'1995-04-01T05:55:55.55',
'2000-01-01T00:00:00.00',
'2019-03-05T09:12:13.51',
'1993-10-15T17:23:47.00',
'2001-09-15T00:00:00.15',
'2019-02-18T13:12:10.09',
'1991-07-12T16:17:33.11',
'2005-05-05T05:05:05.05',
'2011-08-27T03:06:15.00',
'1999-07-09T09:01:33.21',
'2018-04-04T17:30:00.00',
'2012-12-01T21:00:00.01',
'1994-05-02T11:22:33.00',
'2003-07-02T22:33:00.00',
'2014-06-03T06:30:00.00',
'1997-09-04T20:31:00.11',
'2004-02-24T04:00:00.00',
'2000-06-15T12:30:30.00',
],
dtype=np.datetime64)
timedelta = datetime - np.datetime64('1996-05-17T16:45:00.00')
columns["datetime"] = datetime
columns["timedelta"] = timedelta
columns["123456"] = x # a column that will have an alias
df = vaex.from_arrays(**columns)
df.add_virtual_column("z", "x+t*y")
df.set_variable("t", 1.)
return df._readonly()
def create_base_ds():
dataset = vaex.dataset.DatasetArrays("dataset")
x = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
y = y = x**2
ints = np.arange(-2, 19, dtype="i8")
ints[0] = 2**62 + 1
ints[1] = -2**62 + 1
ints[2] = -2**62 - 1
ints[0 + 10] = 2**62 + 1
ints[1 + 10] = -2**62 + 1
ints[2 + 10] = -2**62 - 1
dataset.add_column("x", x)
dataset.add_column("y", y)
# m = x.copy()
m = np.arange(-2, 40, dtype=">f8").reshape((-1, 21)).T.copy()[:, 0]
ma_value = 77777
m[-1 + 10] = ma_value
m[-1 + 20] = ma_value
m = np.ma.array(m, mask=m == ma_value)
n = x.copy()
n[-2 + 10] = np.nan
n[-2 + 20] = np.nan
nm = x.copy()
nm[-2 + 10] = np.nan
nm[-2 + 20] = np.nan
nm[-1 + 10] = ma_value
nm[-1 + 20] = ma_value
nm = np.ma.array(nm, mask=nm == ma_value)
mi = np.ma.array(m.data.astype(np.int64),
mask=m.data == ma_value,
fill_value=88888)
dataset.add_column("m", m)
dataset.add_column('n', n)
dataset.add_column('nm', nm)
dataset.add_column("mi", mi)
dataset.add_column("ints", ints)
name = np.array(list(map(lambda x: str(x) + "bla" + ('_' * int(x)), x)),
dtype='U') #, dtype=np.string_)
dataset.add_column("name", np.array(name))
dataset.add_column("name_arrow", vaex.string_column(name))
obj_data = np.array([
'train', 'false', True, 1, 30., np.nan, 'something',
'something a bit longer resembling a sentence?!', -10000,
'this should be masked'
],
dtype='object')
obj_mask = np.array([False] * 9 + [True])
obj = nm.copy().astype('object')
obj[2:12] = np.ma.MaskedArray(data=obj_data, mask=obj_mask, dtype='object')
dataset.add_column("obj", obj)
booleans = np.ones(21, dtype=np.bool)
booleans[[4, 6, 8, 14, 16, 19]] = False
dataset.add_column("bool", booleans)
datetime = np.array([
'2016-02-29T22:02:02.32',
'2013-01-17T01:02:03.32',
'2017-11-11T08:15:15.00',
'1995-04-01T05:55:55.55',
'2000-01-01T00:00:00.00',
'2019-03-05T09:12:13.51',
'1993-10-15T17:23:47.00',
'2001-09-15T00:00:00.15',
'2019-02-18T13:12:10.09',
'1991-07-12T16:17:33.11',
'2005-05-05T05:05:05.05',
'2011-08-27T03:06:15.00',
'1999-07-09T09:01:33.21',
'2018-04-04T17:30:00.00',
'2012-12-01T21:00:00.01',
'1994-05-02T11:22:33.00',
'2003-07-02T22:33:00.00',
'2014-06-03T06:30:00.00',
'1997-09-04T20:31:00.11',
'2004-02-24T04:00:00.00',
'2000-06-15T12:30:30.00',
],
dtype=np.datetime64)
timedelta = datetime - np.datetime64('1996-05-17T16:45:00.00')
dataset.add_column("datetime", datetime)
dataset.add_column("timedelta", timedelta)
return dataset._readonly()
def create_base_ds():
dataset = vaex.dataset.DatasetArrays("dataset")
x = np.arange(-2, 40, dtype=">f8").reshape((-1,21)).T.copy()[:,0]
y = y = x ** 2
ints = np.arange(-2,19, dtype="i8")
ints[0] = 2**62+1
ints[1] = -2**62+1
ints[2] = -2**62-1
ints[0+10] = 2**62+1
ints[1+10] = -2**62+1
ints[2+10] = -2**62-1
dataset.add_column("x", x)
dataset.add_column("y", y)
# m = x.copy()
m = np.arange(-2, 40, dtype=">f8").reshape((-1,21)).T.copy()[:,0]
ma_value = 77777
m[-1+10] = ma_value
m[-1+20] = ma_value
m = np.ma.array(m, mask=m==ma_value)
n = x.copy()
n[-2+10] = np.nan
n[-2+20] = np.nan
nm = x.copy()
nm[-2+10] = np.nan
nm[-2+20] = np.nan
nm[-1+10] = ma_value
nm[-1+20] = ma_value
nm = np.ma.array(nm, mask=nm==ma_value)
mi = np.ma.array(m.data.astype(np.int64), mask=m.data==ma_value, fill_value=88888)
dataset.add_column("m", m)
dataset.add_column('n', n)
dataset.add_column('nm', nm)
dataset.add_column("mi", mi)
dataset.add_column("ints", ints)
name = np.array(list(map(lambda x: str(x) + "bla" + ('_' * int(x)), x)), dtype='U') #, dtype=np.string_)
dataset.add_column("name", np.array(name))
dataset.add_column("name_arrow", vaex.string_column(name))
obj_data = np.array(['train', 'false' , True, 1, 30., np.nan, 'something', 'something a bit longer resembling a sentence?!', -10000, 'this should be masked'], dtype='object')
obj_mask = np.array([False] * 9 + [True])
obj = nm.copy().astype('object')
obj[2:12] = np.ma.MaskedArray(data=obj_data, mask=obj_mask, dtype='object')
dataset.add_column("obj", obj)
booleans = np.ones(21, dtype=np.bool)
booleans[[4, 6, 8, 14, 16, 19]] = False
dataset.add_column("bool", booleans)
datetime = np.array(['2016-02-29T22:02:02.32', '2013-01-17T01:02:03.32', '2017-11-11T08:15:15.00',
'1995-04-01T05:55:55.55', '2000-01-01T00:00:00.00', '2019-03-05T09:12:13.51',
'1993-10-15T17:23:47.00', '2001-09-15T00:00:00.15', '2019-02-18T13:12:10.09',
'1991-07-12T16:17:33.11', '2005-05-05T05:05:05.05', '2011-08-27T03:06:15.00',
'1999-07-09T09:01:33.21', '2018-04-04T17:30:00.00', '2012-12-01T21:00:00.01',
'1994-05-02T11:22:33.00', '2003-07-02T22:33:00.00', '2014-06-03T06:30:00.00',
'1997-09-04T20:31:00.11', '2004-02-24T04:00:00.00', '2000-06-15T12:30:30.00',
],dtype=np.datetime64)
timedelta = datetime - np.datetime64('1996-05-17T16:45:00.00')
dataset.add_column("datetime", datetime)
dataset.add_column("timedelta", timedelta)
return dataset._readonly()
def test_string_strip_special_case2():
strings = [
'The eunuch in question left me no choice but to reinsert it. Take action as you see fit.·snunɐw·'
]
df = vaex.from_arrays(s=vaex.string_column(strings))
assert df.s.str.upper().tolist() == df.s.str_pandas.upper().tolist()
def test_string_strip_special_case2():
strings = ['ɐa', 'aap']
df = vaex.from_arrays(s=vaex.string_column(strings))
assert df.s.str.capitalize().tolist() == df.s.str_pandas.capitalize(
).tolist()
def test_concat_arrow_strings():
df1 = vaex.from_arrays(x=vaex.string_column(['aap', 'noot', 'mies']))
df2 = vaex.from_arrays(x=vaex.string_column(['a', 'b', 'c']))
df = vaex.concat([df1, df2])
assert df.data_type('x') == df1.data_type('x')
assert df.x.tolist() == ['aap', 'noot', 'mies', 'a', 'b', 'c']
