def test_dt_invert_invalid(src):
dt0 = dt.Frame(src)
for engine in ["llvm", "eager"]:
if engine == "llvm" and not has_llvm():
continue
with pytest.raises(TypeError) as e:
dt0(select=~f[0], engine=engine)
assert str(e.value) == ("Operator `~` cannot be applied to a `%s` "
"column" % dt0.stypes[0].name)
def test_filter_on_view3():
df0 = dt.Frame({"A": range(20)})
df1 = df0[::5, :]
df2 = df1(f.A <= 10, engine="eager")
df2.internal.check()
assert df2.topython() == [[0, 5, 10]]
if has_llvm():
df3 = df1(f.A <= 10, engine="llvm")
df3.internal.check()
assert df3.topython() == [[0, 5, 10]]
def test_rows_unary_minus(df1):
dt1 = df1(-f.A < -3, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[4, 5, 6, 7, 9], [4, 0, 2, None, 9]]
if has_llvm():
dt2 = df1(-f.A < -3, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_compare_to_scalar_lt(df1):
dt1 = df1(f.A < 3, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[0, 1, 2], [3, 2, 1]]
if has_llvm():
dt2 = df1(f.A < 3, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_compare_to_scalar_eq(df1):
dt1 = df1(f.A == None, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[None, None], [None, 8]]
if has_llvm():
dt2 = df1(f.A == None, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_less_than_or_equal(df1):
dt1 = df1(f.A <= f.B, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[0, 1, 3, 4, None, 9], [3, 2, 3, 4, None, 9]]
if has_llvm():
dt2 = df1(f.A <= f.B, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_compare_to_scalar_gt(df1):
dt1 = df1(f.A > 3, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[4, 5, 6, 7, 9], [4, 0, 2, None, 9]]
if has_llvm():
dt2 = df1(f.A > 3, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_greater_than(df1):
dt1 = df1(f.A > f.B, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[2, 5, 6], [1, 0, 2]]
if has_llvm():
dt2 = df1(f.A > f.B, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_less_than(df1):
dt1 = df1(f.A < f.B, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[0, 1], [3, 2]]
if has_llvm():
dt2 = df1(f.A < f.B, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_dt_invert(src):
dt0 = dt.Frame(src)
df2 = dt0(select=~f[0], engine="eager")
df2.internal.check()
assert df2.stypes == dt0.stypes
assert df2.to_list() == [[inv(x) for x in src]]
if has_llvm():
df1 = dt0(select=~f[0], engine="llvm")
df1.internal.check()
assert df1.stypes == dt0.stypes
assert df1.to_list() == [[inv(x) for x in src]]
def test_rows_isna(df1):
from datatable import isna
dt1 = df1(isna(f.A), engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[None, None], [None, 8]]
if has_llvm():
dt2 = df1(isna(f.A), engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_greater_than_or_equal(df1):
dt1 = df1(f.A >= f.B, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[2, 3, 4, 5, 6, None, 9],
[1, 3, 4, 0, 2, None, 9]]
if has_llvm():
dt2 = df1(f.A >= f.B, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_rows_not_equal(df1):
dt1 = df1(f.A != f.B, engine="eager")
dt1.internal.check()
assert dt1.names == df1.names
assert dt1.topython() == [[0, 1, 2, 5, 6, 7, None],
[3, 2, 1, 0, 2, None, 8]]
if has_llvm():
dt2 = df1(f.A != f.B, engine="llvm")
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.topython() == dt2.topython()
def test_dt_isna(src):
dt0 = dt.Frame(src)
dt1 = dt0(select=lambda f: dt.isna(f[0]), engine="eager")
dt1.internal.check()
assert dt1.stypes == (stype.bool8, )
pyans = [x is None for x in src]
assert dt1.topython()[0] == pyans
if has_llvm():
dt2 = dt0(select=lambda f: dt.isna(f[0]), engine="llvm")
dt2.internal.check()
assert dt2.stypes == (stype.bool8, )
assert dt2.topython()[0] == pyans
def test_dt_isna(src):
from datatable import isna
dt0 = dt.Frame(src)
dt1 = dt0[:, isna(f[0])]
dt1.internal.check()
assert dt1.stypes == (stype.bool8, )
pyans = [x is None for x in src]
assert dt1.to_list()[0] == pyans
if has_llvm():
dt2 = dt0(select=lambda f: dt.isna(f[0]), engine="llvm")
dt2.internal.check()
assert dt2.stypes == (stype.bool8, )
assert dt2.to_list()[0] == pyans
def test_cols_expression2():
# In Py3.6+ we could have used a regular dictionary here...
from collections import OrderedDict
selector = OrderedDict([("foo", f.A), ("bar", -f.A)])
f0 = dt.Frame({"A": range(10)})
f1 = f0(select=selector, engine="eager")
f1.internal.check()
assert f1.names == ("foo", "bar")
assert f1.stypes == f0.stypes * 2
assert f1.topython() == [list(range(10)), list(range(0, -10, -1))]
if has_llvm():
f2 = f0(select=selector, engine="llvm")
f2.internal.check()
assert f2.stypes == f1.stypes
assert f2.names == f1.names
assert f2.topython() == f1.topython()
