def test_unstack(data, is_multi_idx, is_multi_col):
modin_df, pandas_df = create_test_dfs(data)
if is_multi_idx:
index = generate_multiindex(len(pandas_df),
nlevels=4,
is_tree_like=True)
else:
index = pandas_df.index
if is_multi_col:
columns = generate_multiindex(len(pandas_df.columns),
nlevels=3,
is_tree_like=True)
else:
columns = pandas_df.columns
pandas_df.columns = modin_df.columns = columns
pandas_df.index = modin_df.index = index
df_equals(modin_df.unstack(), pandas_df.unstack())
df_equals(modin_df.unstack(level=1), pandas_df.unstack(level=1))
if is_multi_idx:
df_equals(modin_df.unstack(level=[0, 1]),
pandas_df.unstack(level=[0, 1]))
df_equals(modin_df.unstack(level=[0, 1, 2]),
pandas_df.unstack(level=[0, 1, 2]))
df_equals(modin_df.unstack(level=[0, 1, 2, 3]),
pandas_df.unstack(level=[0, 1, 2, 3]))
def get_new_index(index, cond):
if cond == "col_multi_tree" or cond == "idx_multi_tree":
return generate_multiindex(len(index), nlevels=3, is_tree_like=True)
elif cond == "col_multi_not_tree" or cond == "idx_multi_not_tree":
return generate_multiindex(len(index), nlevels=3)
else:
return index
def test_sort_values(data, by, axis, ascending, inplace, kind, na_position,
ignore_index, key):
if ascending is None:
pytest.skip("None is not a valid value for ascending.")
if (axis == 1 or axis == "columns") and ignore_index:
pytest.skip("Pandas bug #39426 which is fixed in Pandas 1.3")
if ascending is None and key is not None:
pytest.skip("Pandas bug #41318")
if "multiindex" in by:
index = generate_multiindex(len(data[list(data.keys())[0]]), nlevels=2)
columns = generate_multiindex(len(data.keys()), nlevels=2)
data = {columns[ind]: data[key] for ind, key in enumerate(data)}
else:
index = None
columns = None
modin_df = pd.DataFrame(data, index=index, columns=columns)
pandas_df = pandas.DataFrame(data, index=index, columns=columns)
index = modin_df.index if axis == 1 or axis == "columns" else modin_df.columns
# Parse "by" spec
by_list = []
for b in by.split(","):
if b == "first":
by_list.append(index[0])
elif b == "last":
by_list.append(index[-1])
elif b == "middle":
by_list.append(index[len(index) // 2])
elif b.startswith("multiindex_level"):
by_list.append(index.names[int(b[len("multiindex_level"):])])
else:
raise Exception('Unknown "by" specifier:' + b)
# Create "ascending" list
if ascending in ["list_first_True", "list_first_False"]:
start = 0 if ascending == "list_first_False" else 1
ascending = [i & 1 > 0 for i in range(start, len(by_list) + start)]
eval_general(
modin_df,
pandas_df,
lambda df: df.sort_values(
by_list,
axis=axis,
ascending=ascending,
inplace=inplace,
kind=kind,
na_position=na_position,
ignore_index=ignore_index,
key=key,
),
__inplace__=inplace,
)
def test_all_any_level(data, axis, level, method):
modin_df, pandas_df = pd.DataFrame(data), pandas.DataFrame(data)
if axis == 0:
new_idx = generate_multiindex(len(modin_df.index))
modin_df.index = new_idx
pandas_df.index = new_idx
else:
new_col = generate_multiindex(len(modin_df.columns))
modin_df.columns = new_col
pandas_df.columns = new_col
eval_general(
modin_df,
pandas_df,
lambda df: getattr(df, method)(axis=axis, level=level),
)
def test_count_level(data, axis, level):
modin_df, pandas_df = pd.DataFrame(data), pandas.DataFrame(data)
if axis == 0:
new_idx = generate_multiindex(len(modin_df.index))
modin_df.index = new_idx
pandas_df.index = new_idx
else:
new_col = generate_multiindex(len(modin_df.columns))
modin_df.columns = new_col
pandas_df.columns = new_col
eval_general(
modin_df,
pandas_df,
lambda df: df.count(axis=axis, level=level),
)
def test_mad_level(level):
data = test_data_values[0]
modin_df, pandas_df = pd.DataFrame(data), pandas.DataFrame(data)
index = generate_multiindex(len(data.keys()))
modin_df.columns = index
pandas_df.columns = index
eval_general(
modin_df,
pandas_df,
lambda df: df.mad(axis=1, level=level),
)
def test_kurt_kurtosis_level(level):
data = test_data["int_data"]
df_modin, df_pandas = pd.DataFrame(data), pandas.DataFrame(data)
index = generate_multiindex(len(data.keys()))
df_modin.columns = index
df_pandas.columns = index
eval_general(
df_modin,
df_pandas,
lambda df: df.kurtosis(axis=1, level=level),
)
def test_sort_multiindex(sort_remaining):
data = test_data["int_data"]
modin_df, pandas_df = pd.DataFrame(data), pandas.DataFrame(data)
for index in ["index", "columns"]:
new_index = generate_multiindex(len(getattr(modin_df, index)))
for df in [modin_df, pandas_df]:
setattr(df, index, new_index)
for kwargs in [{"level": 0}, {"axis": 0}, {"axis": 1}]:
with pytest.warns(UserWarning):
df_equals(
modin_df.sort_index(sort_remaining=sort_remaining, **kwargs),
pandas_df.sort_index(sort_remaining=sort_remaining, **kwargs),
)
def applier(df1, df2, **kwargs):
df2.index = generate_multiindex(len(df2))
return df1.add(df2, level=1)
