|
| 1 | +# |
| 2 | +# Copyright (c) 2012-2025 Snowflake Computing Inc. All rights reserved. |
| 3 | +# |
| 4 | + |
| 5 | +import modin.pandas as pd |
| 6 | +import pandas as native_pd |
| 7 | +import pytest |
| 8 | + |
| 9 | +import snowflake.snowpark.modin.plugin # noqa: F401 |
| 10 | +from snowflake.snowpark.exceptions import SnowparkSQLException |
| 11 | +from snowflake.snowpark.session import Session |
| 12 | +from tests.integ.modin.utils import BASIC_TYPE_DATA1, BASIC_TYPE_DATA2 |
| 13 | +from tests.integ.utils.sql_counter import sql_count_checker |
| 14 | +from tests.utils import Utils |
| 15 | + |
| 16 | + |
| 17 | +@pytest.fixture(scope="function") |
| 18 | +def native_pandas_df_basic(): |
| 19 | + native_df = native_pd.DataFrame( |
| 20 | + { |
| 21 | + "ID": [1, 2, 3], |
| 22 | + "FOOT_SIZE": [32.0, 27.0, 40.0], |
| 23 | + "SHOE_MODEL": ["medium", "small", "large"], |
| 24 | + } |
| 25 | + ) |
| 26 | + native_df = native_df.set_index("ID") |
| 27 | + return native_df |
| 28 | + |
| 29 | + |
| 30 | +@sql_count_checker(query_count=2) |
| 31 | +def test_to_view_basic(session, native_pandas_df_basic) -> None: |
| 32 | + view_name = Utils.random_view_name() |
| 33 | + try: |
| 34 | + snow_dataframe = pd.DataFrame(native_pandas_df_basic) |
| 35 | + |
| 36 | + assert ( |
| 37 | + "successfully created" |
| 38 | + in snow_dataframe.to_view(name=view_name)[0]["status"] |
| 39 | + ) |
| 40 | + |
| 41 | + finally: |
| 42 | + Utils.drop_view(session, view_name) |
| 43 | + |
| 44 | + |
| 45 | +@sql_count_checker(query_count=6) |
| 46 | +def test_to_view_multiple_sessions_no_relaxed_ordering_raises( |
| 47 | + session, |
| 48 | + db_parameters, |
| 49 | +) -> None: |
| 50 | + try: |
| 51 | + # create table |
| 52 | + table_name = Utils.random_table_name() |
| 53 | + session.create_dataframe( |
| 54 | + [BASIC_TYPE_DATA1, BASIC_TYPE_DATA2] |
| 55 | + ).write.save_as_table(table_name) |
| 56 | + |
| 57 | + # create dataframe with relaxed_ordering disabled |
| 58 | + snow_dataframe = pd.read_snowflake( |
| 59 | + f"(((SELECT * FROM {table_name})))", relaxed_ordering=False |
| 60 | + ) |
| 61 | + |
| 62 | + # create view |
| 63 | + view_name = Utils.random_view_name() |
| 64 | + assert ( |
| 65 | + "successfully created" |
| 66 | + in snow_dataframe.to_view(name=view_name)[0]["status"] |
| 67 | + ) |
| 68 | + |
| 69 | + # another session |
| 70 | + new_session = Session.builder.configs(db_parameters).create() |
| 71 | + pd.session = session |
| 72 | + |
| 73 | + # accessing the created view in another session fails when relaxed_ordering is disabled |
| 74 | + with pytest.raises( |
| 75 | + SnowparkSQLException, |
| 76 | + match="Object 'VIEW_NAME' does not exist or not authorized", |
| 77 | + ): |
| 78 | + new_session.sql("select * from view_name").collect() |
| 79 | + new_session.close() |
| 80 | + |
| 81 | + finally: |
| 82 | + # cleanup |
| 83 | + Utils.drop_view(session, view_name) |
| 84 | + Utils.drop_table(session, table_name) |
| 85 | + pd.session = session |
| 86 | + |
| 87 | + |
| 88 | +@sql_count_checker(query_count=4) |
| 89 | +def test_to_view_multiple_sessions_relaxed_ordering( |
| 90 | + session, |
| 91 | + db_parameters, |
| 92 | +) -> None: |
| 93 | + try: |
| 94 | + # create table |
| 95 | + table_name = Utils.random_table_name() |
| 96 | + session.create_dataframe( |
| 97 | + [BASIC_TYPE_DATA1, BASIC_TYPE_DATA2] |
| 98 | + ).write.save_as_table(table_name) |
| 99 | + |
| 100 | + # create dataframe with relaxed_ordering enabled |
| 101 | + snow_dataframe = pd.read_snowflake( |
| 102 | + f"(((SELECT * FROM {table_name})))", relaxed_ordering=True |
| 103 | + ) |
| 104 | + |
| 105 | + # create view |
| 106 | + view_name = Utils.random_view_name() |
| 107 | + assert ( |
| 108 | + "successfully created" |
| 109 | + in snow_dataframe.to_view(name=view_name)[0]["status"] |
| 110 | + ) |
| 111 | + |
| 112 | + # another session |
| 113 | + new_session = Session.builder.configs(db_parameters).create() |
| 114 | + pd.session = new_session |
| 115 | + |
| 116 | + # accessing the created view in another session succeeds when relaxed_ordering is enabled |
| 117 | + res = new_session.sql(f"select * from {view_name}").collect() |
| 118 | + assert len(res) == 2 |
| 119 | + new_session.close() |
| 120 | + |
| 121 | + finally: |
| 122 | + # cleanup |
| 123 | + Utils.drop_view(session, view_name) |
| 124 | + Utils.drop_table(session, table_name) |
| 125 | + pd.session = session |
| 126 | + |
| 127 | + |
| 128 | +@pytest.mark.parametrize("index", [True, False]) |
| 129 | +@pytest.mark.parametrize("index_labels", [None, ["my_index"]]) |
| 130 | +@sql_count_checker(query_count=6) |
| 131 | +def test_to_view_index(session, index, index_labels): |
| 132 | + try: |
| 133 | + # create table |
| 134 | + table_name = Utils.random_table_name() |
| 135 | + session.create_dataframe( |
| 136 | + [BASIC_TYPE_DATA1, BASIC_TYPE_DATA2] |
| 137 | + ).write.save_as_table(table_name) |
| 138 | + |
| 139 | + # create dataframe with relaxed_ordering enabled |
| 140 | + snow_dataframe = pd.read_snowflake( |
| 141 | + f"(((SELECT * FROM {table_name})))", relaxed_ordering=True |
| 142 | + ) |
| 143 | + |
| 144 | + view_name = Utils.random_view_name() |
| 145 | + snow_dataframe.to_view(name=view_name, index=index, index_label=index_labels) |
| 146 | + expected_columns = [] |
| 147 | + if index: |
| 148 | + # if index is retained in the result, add it as the first expected column |
| 149 | + expected_index = ["index"] |
| 150 | + if index_labels: |
| 151 | + expected_index = index_labels |
| 152 | + expected_columns = expected_columns + expected_index |
| 153 | + # add the expected data columns |
| 154 | + expected_columns = expected_columns + ["_1", "_2", "_3", "_4", "_5", "_6", "_7"] |
| 155 | + |
| 156 | + # verify columns |
| 157 | + actual = pd.read_snowflake(view_name).columns |
| 158 | + assert actual.tolist() == expected_columns |
| 159 | + finally: |
| 160 | + # cleanup |
| 161 | + Utils.drop_view(session, view_name) |
| 162 | + Utils.drop_table(session, table_name) |
| 163 | + |
| 164 | + |
| 165 | +@sql_count_checker(query_count=6) |
| 166 | +def test_to_view_multiindex(session): |
| 167 | + try: |
| 168 | + # create table |
| 169 | + table_name = Utils.random_table_name() |
| 170 | + session.create_dataframe( |
| 171 | + [BASIC_TYPE_DATA1, BASIC_TYPE_DATA2] |
| 172 | + ).write.save_as_table(table_name) |
| 173 | + |
| 174 | + # create dataframe with relaxed_ordering enabled |
| 175 | + snow_dataframe = pd.read_snowflake( |
| 176 | + f"(((SELECT * FROM {table_name})))", relaxed_ordering=True |
| 177 | + ) |
| 178 | + |
| 179 | + # make sure dataframe has a multi-index |
| 180 | + snow_dataframe = snow_dataframe.set_index(["_1", "_2"]) |
| 181 | + |
| 182 | + view_name = Utils.random_view_name() |
| 183 | + snow_dataframe.to_view( |
| 184 | + name=view_name, |
| 185 | + index=True, |
| 186 | + ) |
| 187 | + |
| 188 | + # verify columns |
| 189 | + actual = pd.read_snowflake(view_name).columns |
| 190 | + assert actual.tolist() == ["_1", "_2", "_3", "_4", "_5", "_6", "_7"] |
| 191 | + |
| 192 | + with pytest.raises( |
| 193 | + ValueError, match="Length of 'index_label' should match number of levels" |
| 194 | + ): |
| 195 | + snow_dataframe.to_view(name=view_name, index=True, index_label=["a"]) |
| 196 | + finally: |
| 197 | + # cleanup |
| 198 | + Utils.drop_view(session, view_name) |
| 199 | + Utils.drop_table(session, table_name) |
0 commit comments