def test_equals():
pandas_df1 = pd.DataFrame({'col1': [2.9, 3, 3, 3], 'col2': [2, 3, 4, 1]})
ray_df1 = rdf.from_pandas(pandas_df1, 2)
ray_df2 = rdf.from_pandas(pandas_df1, 3)
assert ray_df1.equals(ray_df2)
pandas_df2 = pd.DataFrame({'col1': [2.9, 3, 3, 3], 'col2': [2, 3, 5, 1]})
ray_df3 = rdf.from_pandas(pandas_df2, 4)
assert not ray_df3.equals(ray_df1)
assert not ray_df3.equals(ray_df2)
def test_float_dataframe():
pandas_df = pd.DataFrame({
'col1': [0.0, 1.0, 2.0, 3.0],
'col2': [4.0, 5.0, 6.0, 7.0],
'col3': [8.0, 9.0, 10.0, 11.0],
'col4': [12.0, 13.0, 14.0, 15.0]
})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [
lambda x: x + 1, lambda x: str(x), lambda x: x * x, lambda x: x,
lambda x: False
]
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_abs(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
def test_float_dataframe():
pandas_df = pd.DataFrame({
'col1': [0.0, 1.0, 2.0, 3.0],
'col2': [4.0, 5.0, 6.0, 7.0],
'col3': [8.0, 9.0, 10.0, 11.0],
'col4': [12.0, 13.0, 14.0, 15.0],
'col5': [0.0, 0.0, 0.0, 0.0]
})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [
lambda x: x + 1, lambda x: str(x), lambda x: x * x, lambda x: x,
lambda x: False
]
keys = ['col1', 'col2', 'col3', 'col4']
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_abs(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
test_round(ray_df, pandas_df)
test_all(ray_df, pandas_df)
test_any(ray_df, pandas_df)
test___getitem__(ray_df, pandas_df)
test___delitem__(ray_df, pandas_df)
test___copy__(ray_df, pandas_df)
test___deepcopy__(ray_df, pandas_df)
test_bool(ray_df, pandas_df)
test_count(ray_df, pandas_df)
test_head(ray_df, pandas_df)
test_tail(ray_df, pandas_df)
test_idxmax(ray_df, pandas_df)
test_idxmin(ray_df, pandas_df)
test_pop(ray_df, pandas_df)
for key in keys:
test_get(ray_df, pandas_df, key)
test_get_dtype_counts(ray_df, pandas_df)
test_get_ftype_counts(ray_df, pandas_df)
def create_test_dataframe():
df = pd.DataFrame({'col1': [0, 1, 2, 3],
'col2': [4, 5, 6, 7],
'col3': [8, 9, 10, 11],
'col4': [12, 13, 14, 15],
'col5': [0, 0, 0, 0]})
return rdf.from_pandas(df, 2)
def test_bool(ray_df, pd_df):
with pytest.raises(ValueError):
ray_df.bool()
pd_df.bool()
single_bool_pd_df = pd.DataFrame([True])
single_bool_ray_df = rdf.from_pandas(single_bool_pd_df, 1)
assert single_bool_pd_df.bool() == single_bool_ray_df.bool()
def test_mixed_dtype_dataframe():
pandas_df = pd.DataFrame({
'col1': [1, 2, 3, 4],
'col2': [4, 5, 6, 7],
'col3': [8.0, 9.4, 10.1, 11.3],
'col4': ['a', 'b', 'c', 'd']})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [lambda x: x + x,
lambda x: str(x),
lambda x: x,
lambda x: False]
keys = ['col1',
'col2',
'col3',
'col4']
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
for key in keys:
test_get(ray_df, pandas_df, key)
test_get_dtype_counts(ray_df, pandas_df)
test_get_ftype_counts(ray_df, pandas_df)
test_items(ray_df, pandas_df)
test_iterrows(ray_df, pandas_df)
test_items(ray_df, pandas_df)
test_iteritems(ray_df, pandas_df)
test_itertuples(ray_df, pandas_df)
test_max(ray_df, pandas_df)
test_min(ray_df, pandas_df)
test_notna(ray_df, pandas_df)
test_notnull(ray_df, pandas_df)
def test_nan_dataframe():
pandas_df = pd.DataFrame({
'col1': [1, 2, 3, np.nan],
'col2': [4, 5, np.nan, 7],
'col3': [8, np.nan, 10, 11],
'col4': [np.nan, 13, 14, 15]})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [lambda x: x + x,
lambda x: str(x),
lambda x: x,
lambda x: False]
keys = ['col1',
'col2',
'col3',
'col4']
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
for key in keys:
test_get(ray_df, pandas_df, key)
test_get_dtype_counts(ray_df, pandas_df)
test_get_ftype_counts(ray_df, pandas_df)
def test_int_dataframe():
ray.init()
pandas_df = pd.DataFrame({
'col1': [0, 1, 2, 3],
'col2': [4, 5, 6, 7],
'col3': [8, 9, 10, 11],
'col4': [12, 13, 14, 15]
})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [
lambda x: x + 1, lambda x: str(x), lambda x: x * x, lambda x: x,
lambda x: False
]
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_abs(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
def test_int_dataframe():
ray.init()
pandas_df = pd.DataFrame({'col1': [0, 1, 2, 3],
'col2': [4, 5, 6, 7],
'col3': [8, 9, 10, 11],
'col4': [12, 13, 14, 15],
'col5': [0, 0, 0, 0]})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [lambda x: x + 1,
lambda x: str(x),
lambda x: x * x,
lambda x: x,
lambda x: False]
keys = ['col1',
'col2',
'col3',
'col4']
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_abs(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
test_round(ray_df, pandas_df)
test_all(ray_df, pandas_df)
test_any(ray_df, pandas_df)
test___getitem__(ray_df, pandas_df)
test___delitem__(ray_df, pandas_df)
test___copy__(ray_df, pandas_df)
test___deepcopy__(ray_df, pandas_df)
test_bool(ray_df, pandas_df)
test_count(ray_df, pandas_df)
test_head(ray_df, pandas_df)
test_tail(ray_df, pandas_df)
test_idxmax(ray_df, pandas_df)
test_idxmin(ray_df, pandas_df)
test_pop(ray_df, pandas_df)
for key in keys:
test_get(ray_df, pandas_df, key)
test_get_dtype_counts(ray_df, pandas_df)
test_get_ftype_counts(ray_df, pandas_df)
test_iterrows(ray_df, pandas_df)
test_items(ray_df, pandas_df)
test_iteritems(ray_df, pandas_df)
test_itertuples(ray_df, pandas_df)
test_max(ray_df, pandas_df)
test_min(ray_df, pandas_df)
test_notna(ray_df, pandas_df)
test_notnull(ray_df, pandas_df)
def test_nan_dataframe():
pandas_df = pd.DataFrame({
'col1': [1, 2, 3, np.nan],
'col2': [4, 5, np.nan, 7],
'col3': [8, np.nan, 10, 11],
'col4': [np.nan, 13, 14, 15]
})
ray_df = rdf.from_pandas(pandas_df, 2)
testfuncs = [
lambda x: x + x, lambda x: str(x), lambda x: x, lambda x: False
]
query_funcs = [
'col1 < col2', 'col3 > col4', 'col1 == col2',
'(col2 > col1) and (col1 < col3)'
]
keys = ['col1', 'col2', 'col3', 'col4']
test_roundtrip(ray_df, pandas_df)
test_index(ray_df, pandas_df)
test_size(ray_df, pandas_df)
test_ndim(ray_df, pandas_df)
test_ftypes(ray_df, pandas_df)
test_values(ray_df, pandas_df)
test_axes(ray_df, pandas_df)
test_shape(ray_df, pandas_df)
test_add_prefix(ray_df, pandas_df)
test_add_suffix(ray_df, pandas_df)
for testfunc in testfuncs:
test_applymap(ray_df, pandas_df, testfunc)
test_copy(ray_df)
test_sum(ray_df, pandas_df)
test_abs(ray_df, pandas_df)
test_keys(ray_df, pandas_df)
test_transpose(ray_df, pandas_df)
test_round(ray_df, pandas_df)
test_query(ray_df, pandas_df, query_funcs)
test_all(ray_df, pandas_df)
test_any(ray_df, pandas_df)
test___getitem__(ray_df, pandas_df)
test___neg__(ray_df, pandas_df)
test___iter__(ray_df, pandas_df)
test___abs__(ray_df, pandas_df)
test___delitem__(ray_df, pandas_df)
test___copy__(ray_df, pandas_df)
test___deepcopy__(ray_df, pandas_df)
test_bool(ray_df, pandas_df)
test_count(ray_df, pandas_df)
test_head(ray_df, pandas_df, 2)
test_head(ray_df, pandas_df)
test_tail(ray_df, pandas_df)
test_idxmax(ray_df, pandas_df)
test_idxmin(ray_df, pandas_df)
test_pop(ray_df, pandas_df)
test_max(ray_df, pandas_df)
test_min(ray_df, pandas_df)
test_notna(ray_df, pandas_df)
test_notnull(ray_df, pandas_df)
for key in keys:
test_get(ray_df, pandas_df, key)
test_get_dtype_counts(ray_df, pandas_df)
test_get_ftype_counts(ray_df, pandas_df)
test_iterrows(ray_df, pandas_df)
test_items(ray_df, pandas_df)
test_iteritems(ray_df, pandas_df)
test_itertuples(ray_df, pandas_df)
test_loc(ray_df, pandas_df)
test_iloc(ray_df, pandas_df)
labels = ['a', 'b', 'c', 'd']
test_set_axis(ray_df, pandas_df, labels, 0)
test_set_axis(ray_df, pandas_df, labels, 'rows')
test_set_axis(ray_df, pandas_df, labels, 1)
test_set_axis(ray_df, pandas_df, labels, 'columns')
for key in keys:
test_set_index(ray_df, pandas_df, key)
test_set_index(ray_df, pandas_df, key, inplace=True)
test_reset_index(ray_df, pandas_df)
test_reset_index(ray_df, pandas_df, inplace=True)
for key in keys:
test_insert(ray_df, pandas_df, 0, "New Column", ray_df[key])
test_insert(ray_df, pandas_df, 0, "New Column", pandas_df[key])
test_insert(ray_df, pandas_df, 1, "New Column", ray_df[key])
test_insert(ray_df, pandas_df, 4, "New Column", ray_df[key])
