From 2272764864555f26095563937e06a3389d42d789 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sat, 19 Nov 2022 15:50:35 +0100 Subject: Adding upstream version 10.0.8. Signed-off-by: Daniel Baumann --- tests/dataframe/unit/test_dataframe_writer.py | 36 ++++++++++++++------------- 1 file changed, 19 insertions(+), 17 deletions(-) (limited to 'tests/dataframe/unit/test_dataframe_writer.py') diff --git a/tests/dataframe/unit/test_dataframe_writer.py b/tests/dataframe/unit/test_dataframe_writer.py index 14b4a0a..7c646f5 100644 --- a/tests/dataframe/unit/test_dataframe_writer.py +++ b/tests/dataframe/unit/test_dataframe_writer.py @@ -6,39 +6,41 @@ from tests.dataframe.unit.dataframe_sql_validator import DataFrameSQLValidator class TestDataFrameWriter(DataFrameSQLValidator): + maxDiff = None + def test_insertInto_full_path(self): df = self.df_employee.write.insertInto("catalog.db.table_name") - expected = "INSERT INTO catalog.db.table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT INTO catalog.db.table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_insertInto_db_table(self): df = self.df_employee.write.insertInto("db.table_name") - expected = "INSERT INTO db.table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT INTO db.table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_insertInto_table(self): df = self.df_employee.write.insertInto("table_name") - expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_insertInto_overwrite(self): df = self.df_employee.write.insertInto("table_name", overwrite=True) - expected = "INSERT OVERWRITE TABLE table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT OVERWRITE TABLE table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) @mock.patch("sqlglot.schema", MappingSchema()) def test_insertInto_byName(self): sqlglot.schema.add_table("table_name", {"employee_id": "INT"}) df = self.df_employee.write.byName.insertInto("table_name") - expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_insertInto_cache(self): df = self.df_employee.cache().write.insertInto("table_name") expected_statements = [ - "DROP VIEW IF EXISTS t35612", - "CACHE LAZY TABLE t35612 OPTIONS('storageLevel' = 'MEMORY_AND_DISK') AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)", - "INSERT INTO table_name SELECT `t35612`.`employee_id` AS `employee_id`, `t35612`.`fname` AS `fname`, `t35612`.`lname` AS `lname`, `t35612`.`age` AS `age`, `t35612`.`store_id` AS `store_id` FROM `t35612` AS `t35612`", + "DROP VIEW IF EXISTS t37164", + "CACHE LAZY TABLE t37164 OPTIONS('storageLevel' = 'MEMORY_AND_DISK') AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)", + "INSERT INTO table_name SELECT `t37164`.`employee_id` AS `employee_id`, `t37164`.`fname` AS `fname`, `t37164`.`lname` AS `lname`, `t37164`.`age` AS `age`, `t37164`.`store_id` AS `store_id` FROM `t37164` AS `t37164`", ] self.compare_sql(df, expected_statements) @@ -48,39 +50,39 @@ class TestDataFrameWriter(DataFrameSQLValidator): def test_saveAsTable_append(self): df = self.df_employee.write.saveAsTable("table_name", mode="append") - expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "INSERT INTO table_name SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_saveAsTable_overwrite(self): df = self.df_employee.write.saveAsTable("table_name", mode="overwrite") - expected = "CREATE OR REPLACE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "CREATE OR REPLACE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_saveAsTable_error(self): df = self.df_employee.write.saveAsTable("table_name", mode="error") - expected = "CREATE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "CREATE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_saveAsTable_ignore(self): df = self.df_employee.write.saveAsTable("table_name", mode="ignore") - expected = "CREATE TABLE IF NOT EXISTS table_name AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "CREATE TABLE IF NOT EXISTS table_name AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_mode_standalone(self): df = self.df_employee.write.mode("ignore").saveAsTable("table_name") - expected = "CREATE TABLE IF NOT EXISTS table_name AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "CREATE TABLE IF NOT EXISTS table_name AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_mode_override(self): df = self.df_employee.write.mode("ignore").saveAsTable("table_name", mode="overwrite") - expected = "CREATE OR REPLACE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" + expected = "CREATE OR REPLACE TABLE table_name AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)" self.compare_sql(df, expected) def test_saveAsTable_cache(self): df = self.df_employee.cache().write.saveAsTable("table_name") expected_statements = [ - "DROP VIEW IF EXISTS t35612", - "CACHE LAZY TABLE t35612 OPTIONS('storageLevel' = 'MEMORY_AND_DISK') AS SELECT CAST(`a1`.`employee_id` AS int) AS `employee_id`, CAST(`a1`.`fname` AS string) AS `fname`, CAST(`a1`.`lname` AS string) AS `lname`, CAST(`a1`.`age` AS int) AS `age`, CAST(`a1`.`store_id` AS int) AS `store_id` FROM (VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100)) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)", - "CREATE TABLE table_name AS SELECT `t35612`.`employee_id` AS `employee_id`, `t35612`.`fname` AS `fname`, `t35612`.`lname` AS `lname`, `t35612`.`age` AS `age`, `t35612`.`store_id` AS `store_id` FROM `t35612` AS `t35612`", + "DROP VIEW IF EXISTS t37164", + "CACHE LAZY TABLE t37164 OPTIONS('storageLevel' = 'MEMORY_AND_DISK') AS SELECT CAST(`a1`.`employee_id` AS INT) AS `employee_id`, CAST(`a1`.`fname` AS STRING) AS `fname`, CAST(`a1`.`lname` AS STRING) AS `lname`, CAST(`a1`.`age` AS INT) AS `age`, CAST(`a1`.`store_id` AS INT) AS `store_id` FROM VALUES (1, 'Jack', 'Shephard', 37, 1), (2, 'John', 'Locke', 65, 1), (3, 'Kate', 'Austen', 37, 2), (4, 'Claire', 'Littleton', 27, 2), (5, 'Hugo', 'Reyes', 29, 100) AS `a1`(`employee_id`, `fname`, `lname`, `age`, `store_id`)", + "CREATE TABLE table_name AS SELECT `t37164`.`employee_id` AS `employee_id`, `t37164`.`fname` AS `fname`, `t37164`.`lname` AS `lname`, `t37164`.`age` AS `age`, `t37164`.`store_id` AS `store_id` FROM `t37164` AS `t37164`", ] self.compare_sql(df, expected_statements) -- cgit v1.2.3