def test_div_booleans_integers_floats_random(seed):
random.seed(seed)
n = 1000
src1 = [random.randint(-100, 100) for _ in range(n)]
src2 = [random.randint(0, 1) for _ in range(n)]
src3 = [random.random() * 1000 - 500 for _ in range(n)]
DT = dt.Frame(x=src1, y=src2 / dt.bool8, z=src3)
RES = DT[:, [
f.x / f.y, f.y / f.x, f.y / f.z, f.z / f.y, f.z / f.x, f.x / f.z
]]
EXP = dt.Frame([[dt_div(src1[i], src2[i]) for i in range(n)],
[dt_div(src2[i], src1[i]) for i in range(n)],
[dt_div(src2[i], src3[i]) for i in range(n)],
[dt_div(src3[i], src2[i]) for i in range(n)],
[dt_div(src3[i], src1[i]) for i in range(n)],
[dt_div(src1[i], src3[i]) for i in range(n)]])
assert_equals(RES, EXP)
def test_save_to_jay(tempfile_jay):
src = [d(1, 1, 1), d(2001, 12, 13), d(2026, 5, 9), None, d(1956, 11, 11)]
DT = dt.Frame(src)
DT.to_jay(tempfile_jay)
del DT
DT2 = dt.fread(tempfile_jay)
assert DT2.shape == (5, 1)
assert DT2.type == dt.Type.date32
assert DT2.to_list()[0] == src
def test_corr_random(numpy, seed):
numpy.random.seed(seed)
arr1 = numpy.random.rand(100)
arr2 = numpy.random.rand(100)
np_corr = numpy.corrcoef(arr1, arr2)[0, 1]
DT = dt.Frame([arr1, arr2])
dt_corr = DT[:, corr(f[0], f[1])][0, 0]
assert numpy.isclose(np_corr, dt_corr, atol=0, rtol=1e-12)
def test_select_dates():
DT = dt.Frame(A=[12], B=[d(2000, 12, 20)], C=[True])
assert DT.types == [dt.Type.int32, dt.Type.date32, dt.Type.bool8]
RES1 = DT[:, f[d]]
RES2 = DT[:, d]
RES3 = DT[:, dt.ltype.time]
assert_equals(RES1, DT['B'])
assert_equals(RES2, DT['B'])
assert_equals(RES3, DT['B'])
def test_assign_nonexisting_column():
# See #1983: if column `B` is created at a wrong moment in the evaluation
# sequence, this may seg.fault
DT = dt.Frame(A=range(5))
with pytest.raises(KeyError,
match="Column B does not exist in the "
"Frame"):
DT[:, "B"] = f.B + 1
frame_integrity_check(DT)
def test_slice_wrong_types2(ttype):
DT = dt.Frame(A=[None], B=["Hello"], types={"A": ttype})
msg = 'Non-integer expressions cannot be used inside a slice'
with pytest.raises(TypeError, match=msg):
DT[:, f.B[f.A:]]
with pytest.raises(TypeError, match=msg):
DT[:, f.B[:f.A]]
with pytest.raises(TypeError, match=msg):
DT[:, f.B[::f.A]]
def test_ymdt_unnamed():
DT = dt.Frame(Y=[2001, 2003, 2005, 2020, 1960],
M=[1, 5, 4, 11, 8],
D=[12, 18, 30, 1, 14],
h=[7, 14, 22, 23, 12],
m=[15, 30, 0, 59, 0],
s=[12, 23, 0, 59, 27],
ns=[0, 0, 0, 999999000, 123000])
RES = DT[:, ymdt(f.Y, f.M, f.D, f.h, f.m, f.s, f.ns)]
assert_equals(
RES,
dt.Frame([
t(2001, 1, 12, 7, 15, 12),
t(2003, 5, 18, 14, 30, 23),
t(2005, 4, 30, 22, 0, 0),
t(2020, 11, 1, 23, 59, 59, 999999),
t(1960, 8, 14, 12, 0, 27, 123)
]))
def test_bool_create_column_forced():
# When the column is forced into type bool, we effectively apply
# bool(x) to each value in the list,
# with the only exception being `math.nan`, which is converted into None.
DT = dt.Frame([True, False, None, 0, 1, 12, 0.0, -0.0, "", "hi", math.nan],
type=bool)
assert DT.type == dt.Type.bool8
assert DT.to_list() == [[
True, False, None, False, True, True, False, False, False, True, None
]]
def test_sub_booleans_integers_floats_random(seed):
random.seed(seed)
n = 1000
src1 = [random.randint(-100, 100) for _ in range(n)]
src2 = [random.randint(0, 1) for _ in range(n)]
src3 = [random.random() * 1000 - 500 for _ in range(n)]
DT = dt.Frame(x=src1, y=src2/dt.bool8, z=src3)
RES = DT[:, [f.x - f.y, f.y - f.x,
f.y - f.z, f.z - f.y,
f.z - f.x, f.x - f.z]]
EXP = dt.Frame(
C0=[src1[i] - src2[i] for i in range(n)],
C2=[src2[i] - src3[i] for i in range(n)],
C4=[src3[i] - src1[i] for i in range(n)]
)
assert_equals(RES[:, [0, 2, 4]], EXP)
assert_equals(RES[:, [0, 2, 4]],
RES[:, {"C0":-f[1], "C2":-f[3], "C4":-f[5]}])
def test_arr32_arr32_repr():
DT = dt.Frame(V=[[[1, 2, 3], [4, 9]], None, [None], [[-1], [0, 13]]])
assert str(DT) == (" | V \n"
" | arr32(arr32(int32))\n"
"-- + -------------------\n"
" 0 | [[1, 2, 3], [4, 9]]\n"
" 1 | NA \n"
" 2 | [NA] \n"
" 3 | [[-1], [0, 13]] \n"
"[4 rows x 1 column]\n")
def test_median_wrong_stype():
DT = dt.Frame(A=["foo"], B=["moo"], stypes={"A": dt.str32, "B": dt.str64})
with pytest.raises(TypeError) as e:
noop(DT[:, median(f.A)])
assert ("Unable to apply reduce function median() to a column of "
"type str32" in str(e.value))
with pytest.raises(TypeError) as e:
noop(DT[:, median(f.B)])
assert ("Unable to apply reduce function median() to a column of "
"type str64" in str(e.value))
def test_update_misplaced():
DT = dt.Frame(A=range(5))
with pytest.raises(TypeError,
match="Column selector must be an integer "
"or a string"):
DT[update(B=0)]
with pytest.raises(TypeError,
match="Invalid item at position 2 in "
r"DT\[i, j, \.\.\.\] call"):
DT[:, :, update(B=0)]
def test_create_from_python1():
src = [[1, 2, 3], [], [4, 5], [6], None, [7, 8, 10, -1]]
DT = dt.Frame(A=src)
assert DT.shape == (6, 1)
assert DT.type == dt.Type.arr32(dt.Type.int32)
assert DT.names == ("A", )
assert DT.ltypes == (dt.ltype.invalid,
) # These properties are deprecated, also
assert DT.stypes == (dt.stype.arr32, ) # see issue #3142
assert DT.to_list() == [src]
def test_cast_string_to_date32(ttype):
DT = dt.Frame(["2001-02-14",
"2012-11-24",
"noise", "2022-22-22", "2021-02-29", "2003-04-31",
"2020-02-29",
"2003-10-01",
"1969-12-31",
"1970-01-01",
"2000-00-00", "2000-01-1", "2000-01-"],
stype=ttype)
DT[0] = dt.Type.date32
assert_equals(DT, dt.Frame([d(2001, 2, 14),
d(2012, 11, 24),
None, None, None, None,
d(2020, 2, 29),
d(2003, 10, 1),
d(1969, 12, 31),
d(1970, 1, 1),
None, None, None]))
def test_date32_relational():
DT = dt.Frame(A=[d(2000, 1, 1), d(2010, 11, 17), None, d(2020, 3, 30),
None, d(1998, 5, 14), d(999, 9, 9)],
B=[d(2000, 1, 2), d(2010, 11, 17), None, d(2020, 1, 31),
d(1998, 5, 14), None, d(999, 9, 9)])
RES = DT[:, {">": f.A > f.B,
">=": f.A >= f.B,
"==": f.A == f.B,
"!=": f.A != f.B,
"<=": f.A <= f.B,
"<": f.A < f.B}]
assert_equals(RES, dt.Frame({
">": [False, False, False, True, False, False, False],
">=": [False, True, True, True, False, False, True],
"==": [False, True, True, False, False, False, True],
"!=": [True, False, False, True, True, True, False],
"<=": [True, True, True, False, False, False, True],
"<": [True, False, False, False, False, False, False],
}))
def test_mean_empty_frame():
DT = dt.Frame([[]] * 4, names=list("ABCD"),
stypes=(dt.bool8, dt.int32, dt.float32, dt.float64))
assert DT.shape == (0, 4)
RZ = DT[:, mean(f[:])]
frame_integrity_check(RZ)
assert RZ.shape == (1, 4)
assert RZ.names == ("A", "B", "C", "D")
assert RZ.stypes == (dt.float64, dt.float64, dt.float32, dt.float64)
assert RZ.to_list() == [[None]] * 4
def test_time64_create_from_python():
d = datetime.datetime
src = [
d(2000, 10, 18, 3, 30),
d(2010, 11, 13, 15, 11, 59),
d(2020, 2, 29, 20, 20, 20, 20), None
]
DT = dt.Frame(src)
assert DT.types == [dt.Type.time64]
assert DT.to_list() == [src]
def test_count_dt_groupby_string():
df_in = dt.Frame([None, "blue", "green", "indico", None, None, "orange",
"red", "violet", "yellow", "green", None, "blue"])
df_reduce = df_in[:, [count(f.C0), count()], "C0"]
frame_integrity_check(df_reduce)
assert df_reduce.shape == (8, 3)
assert df_reduce.ltypes == (ltype.str, ltype.int, ltype.int,)
assert df_reduce.to_list() == [[None, "blue", "green", "indico", "orange",
"red", "violet", "yellow"],
[0, 2, 2, 1, 1, 1, 1, 1],
[4, 2, 2, 1, 1, 1, 1, 1]]
def test_cast_time64_to_date32():
from datetime import date
DT = dt.Frame([
d(2091, 11, 28, 15, 51, 27, 310000),
d(1970, 1, 3, 21, 5, 2, 475000),
d(1969, 12, 31, 23, 59, 59, 999999),
d(1969, 12, 31, 0, 0, 0),
d(1962, 4, 12, 3, 52, 27, 458000), None
])
assert DT.type == dt.Type.time64
DT[0] = dt.Type.date32
assert_equals(
DT,
dt.Frame([
date(2091, 11, 28),
date(1970, 1, 3),
date(1969, 12, 31),
date(1969, 12, 31),
date(1962, 4, 12), None
]))
def test_time64_minmax():
src = [
None,
d(2000, 10, 18, 3, 30),
d(2010, 11, 13, 15, 11, 59),
d(2020, 2, 29, 20, 20, 20, 20), None
]
DT = dt.Frame(src)
assert DT.min1() == d(2000, 10, 18, 3, 30)
assert DT.max1() == d(2020, 2, 29, 20, 20, 20, 20)
assert DT.countna1() == 2
def test_compare():
DT = dt.Frame(A=[d(2010, 11, 5, i * 17 % 11, 0, 0) for i in range(11)])
DT['B'] = DT.sort(0)
RES = DT[:, {
"EQ": (f.A == f.B),
"NE": (f.A != f.B),
"LT": (f.A < f.B),
"LE": (f.A <= f.B),
"GE": (f.A >= f.B),
"GT": (f.A > f.B)
}]
assert_equals(
RES,
dt.Frame(EQ=[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
NE=[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
LT=[0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1],
LE=[1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1],
GE=[1, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0],
GT=[0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0],
type=dt.bool8))
def test_time64_kurt():
src = [
None,
d(2010, 11, 13, 15, 11, 50),
d(2010, 11, 13, 17, 11, 50), None,
d(2010, 11, 13, 15, 11, 50), None
]
DT_date = dt.Frame(src)
DT_int = DT_date[:, dt.int64(f[0])]
assert DT_date.kurt1() == DT_int.kurt1()
assert_equals(DT_date.kurt(), DT_int.kurt())
def test_count_2d_dt_groupby_integer():
df_in = dt.Frame([[9, 8, 2, 3, None, None, 3, 0, 5, 5, 8, None, 1],
[0, 1, 0, 5, 3, 8, 1, 0, 2, 5, None, 8, 1]])
df_reduce = df_in[:, [count(f.C0), count(f.C1), count()], "C0"]
frame_integrity_check(df_reduce)
assert df_reduce.shape == (8, 4)
assert df_reduce.ltypes == (ltype.int,) * 4
assert df_reduce.to_list() == [[None, 0, 1, 2, 3, 5, 8, 9],
[0, 1, 1, 1, 2, 2, 2, 1],
[3, 1, 1, 1, 2, 2, 1, 1],
[3, 1, 1, 1, 2, 2, 2, 1]]
def test_arr32_repr_in_terminal():
DT = dt.Frame(A=[[1], [2, 3], None, [4, 5, 6], []])
assert str(DT) == (" | A \n"
" | arr32(int32)\n"
"-- + ------------\n"
" 0 | [1] \n"
" 1 | [2, 3] \n"
" 2 | NA \n"
" 3 | [4, 5, 6] \n"
" 4 | [] \n"
"[5 rows x 1 column]\n")
def test_round_int16_positive_ndigits():
DT = dt.Frame(A=[None, 12, 0, 34, -999, 32767, 10001, -32767] / dt.int16)
assert_equals(DT[:, dtround(f.A)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=0)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=1)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=2)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=3)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=5)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=9)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=17)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=314)], DT)
def test_round_bool_positive_ndigits():
DT = dt.Frame(A=[True, False, None])
assert_equals(DT[:, dtround(f.A)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=0)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=1)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=2)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=3)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=5)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=9)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=19)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=999999)], DT)
def test_date32_kurt():
src = [None,
d(2010, 11, 13),
d(2010, 11, 12),
d(2022, 11, 12),
d(2010, 11, 13),
None]
DT_date = dt.Frame(src)
DT_int = DT_date[:, dt.int32(f[0])]
assert DT_date.kurt1() == DT_int.kurt1()
assert_equals(DT_date.kurt(), DT_int.kurt())
def test_round_int8_positive_ndigits():
DT = dt.Frame(A=[None] + list(range(-127, 128)), stype=dt.int8)
assert_equals(DT[:, dtround(f.A)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=0)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=1)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=2)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=3)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=5)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=9)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=17)], DT)
assert_equals(DT[:, dtround(f.A, ndigits=987654321)], DT)
def test_median_grouped():
DT = dt.Frame(A=[0, 0, 0, 0, 1, 1, 1, 1, 1],
B=[2, 6, 1, 0, -3, 4, None, None, -1],
stypes={
"A": dt.int16,
"B": dt.int32
})
RES = DT[:, median(f.B), by(f.A)]
assert RES.shape == (2, 2)
assert RES.stypes == (dt.int16, dt.float64)
assert RES.to_list() == [[0, 1], [1.5, -1.0]]
def test_bool_create_force_from_exceptional():
# Here we check what happens if the objects that we are trying to
# cast into bools throw an exception. An FExpr is just that kind of
# object.
# Two things are expected to happen here: the exception-raising objects
# become `None`s, and exception objects are discarded.
with pytest.raises(TypeError):
assert dt.f.A
DT = dt.Frame([None, dt.f.A, math.nan, dt.f[:], TypeError], type=bool)
assert DT.to_list() == [[None, None, None, None, True]]