diff options
Diffstat (limited to 'tests/dialects/test_spark.py')
| -rw-r--r-- | tests/dialects/test_spark.py | 44 |
1 files changed, 28 insertions, 16 deletions
diff --git a/tests/dialects/test_spark.py b/tests/dialects/test_spark.py index 24f68f8..56a573a 100644 --- a/tests/dialects/test_spark.py +++ b/tests/dialects/test_spark.py @@ -237,6 +237,10 @@ TBLPROPERTIES ( "REFRESH TABLE t", ) + self.validate_identity("DESCRIBE EXTENDED db.table") + self.validate_identity("SELECT * FROM test TABLESAMPLE (50 PERCENT)") + self.validate_identity("SELECT * FROM test TABLESAMPLE (5 ROWS)") + self.validate_identity("SELECT * FROM test TABLESAMPLE (BUCKET 4 OUT OF 10)") self.validate_identity("REFRESH 'hdfs://path/to/table'") self.validate_identity("REFRESH TABLE tempDB.view1") self.validate_identity("SELECT CASE WHEN a = NULL THEN 1 ELSE 2 END") @@ -244,7 +248,7 @@ TBLPROPERTIES ( self.validate_identity("SELECT TRANSFORM(ARRAY(1, 2, 3), x -> x + 1)") self.validate_identity("SELECT TRANSFORM(ARRAY(1, 2, 3), (x, i) -> x + i)") self.validate_identity("REFRESH TABLE a.b.c") - self.validate_identity("INTERVAL -86 days") + self.validate_identity("INTERVAL -86 DAYS") self.validate_identity("SELECT UNIX_TIMESTAMP()") self.validate_identity("TRIM(' SparkSQL ')") self.validate_identity("TRIM(BOTH 'SL' FROM 'SSparkSQLS')") @@ -252,6 +256,14 @@ TBLPROPERTIES ( self.validate_identity("TRIM(TRAILING 'SL' FROM 'SSparkSQLS')") self.validate_identity("SPLIT(str, pattern, lim)") self.validate_identity( + "SELECT INTERVAL '5' HOURS '30' MINUTES '5' SECONDS '6' MILLISECONDS '7' MICROSECONDS", + "SELECT INTERVAL '5' HOURS + INTERVAL '30' MINUTES + INTERVAL '5' SECONDS + INTERVAL '6' MILLISECONDS + INTERVAL '7' MICROSECONDS", + ) + self.validate_identity( + "SELECT INTERVAL 5 HOURS 30 MINUTES 5 SECONDS 6 MILLISECONDS 7 MICROSECONDS", + "SELECT INTERVAL '5' HOURS + INTERVAL '30' MINUTES + INTERVAL '5' SECONDS + INTERVAL '6' MILLISECONDS + INTERVAL '7' MICROSECONDS", + ) + self.validate_identity( "SELECT REGEXP_REPLACE('100-200', r'([^0-9])', '')", "SELECT REGEXP_REPLACE('100-200', '([^0-9])', '')", ) @@ -299,25 +311,25 @@ TBLPROPERTIES ( }, ) self.validate_all( - "SELECT DATEDIFF(month, CAST('1996-10-30' AS TIMESTAMP), CAST('1997-02-28 10:30:00' AS TIMESTAMP))", + "SELECT DATEDIFF(MONTH, CAST('1996-10-30' AS TIMESTAMP), CAST('1997-02-28 10:30:00' AS TIMESTAMP))", read={ "duckdb": "SELECT DATEDIFF('month', CAST('1996-10-30' AS TIMESTAMP), CAST('1997-02-28 10:30:00' AS TIMESTAMP))", }, write={ - "spark": "SELECT DATEDIFF(month, TO_DATE(CAST('1996-10-30' AS TIMESTAMP)), TO_DATE(CAST('1997-02-28 10:30:00' AS TIMESTAMP)))", + "spark": "SELECT DATEDIFF(MONTH, TO_DATE(CAST('1996-10-30' AS TIMESTAMP)), TO_DATE(CAST('1997-02-28 10:30:00' AS TIMESTAMP)))", "spark2": "SELECT CAST(MONTHS_BETWEEN(TO_DATE(CAST('1997-02-28 10:30:00' AS TIMESTAMP)), TO_DATE(CAST('1996-10-30' AS TIMESTAMP))) AS INT)", }, ) self.validate_all( "SELECT DATEDIFF(week, '2020-01-01', '2020-12-31')", write={ - "bigquery": "SELECT DATE_DIFF(CAST('2020-12-31' AS DATE), CAST('2020-01-01' AS DATE), week)", - "duckdb": "SELECT DATE_DIFF('week', CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", + "bigquery": "SELECT DATE_DIFF(CAST('2020-12-31' AS DATE), CAST('2020-01-01' AS DATE), WEEK)", + "duckdb": "SELECT DATE_DIFF('WEEK', CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", "hive": "SELECT CAST(DATEDIFF(TO_DATE('2020-12-31'), TO_DATE('2020-01-01')) / 7 AS INT)", "postgres": "SELECT CAST(EXTRACT(days FROM (CAST(CAST('2020-12-31' AS DATE) AS TIMESTAMP) - CAST(CAST('2020-01-01' AS DATE) AS TIMESTAMP))) / 7 AS BIGINT)", - "redshift": "SELECT DATEDIFF(week, CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", - "snowflake": "SELECT DATEDIFF(week, CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", - "spark": "SELECT DATEDIFF(week, TO_DATE('2020-01-01'), TO_DATE('2020-12-31'))", + "redshift": "SELECT DATEDIFF(WEEK, CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", + "snowflake": "SELECT DATEDIFF(WEEK, CAST('2020-01-01' AS DATE), CAST('2020-12-31' AS DATE))", + "spark": "SELECT DATEDIFF(WEEK, TO_DATE('2020-01-01'), TO_DATE('2020-12-31'))", }, ) self.validate_all( @@ -624,23 +636,23 @@ TBLPROPERTIES ( self.validate_all( "SELECT EXPLODE(x) FROM tbl", write={ - "bigquery": "SELECT IF(pos = pos_2, col, NULL) AS col FROM tbl, UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH(x)) - 1)) AS pos CROSS JOIN UNNEST(x) AS col WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH(x) - 1) AND pos_2 = (ARRAY_LENGTH(x) - 1))", - "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM tbl, UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(x)))) AS _u(pos) CROSS JOIN UNNEST(x) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(x) AND _u_2.pos_2 = CARDINALITY(x))", + "bigquery": "SELECT IF(pos = pos_2, col, NULL) AS col FROM tbl CROSS JOIN UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH(x)) - 1)) AS pos CROSS JOIN UNNEST(x) AS col WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH(x) - 1) AND pos_2 = (ARRAY_LENGTH(x) - 1))", + "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM tbl CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(x)))) AS _u(pos) CROSS JOIN UNNEST(x) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(x) AND _u_2.pos_2 = CARDINALITY(x))", "spark": "SELECT EXPLODE(x) FROM tbl", }, ) self.validate_all( "SELECT EXPLODE(col) FROM _u", write={ - "bigquery": "SELECT IF(pos = pos_2, col_2, NULL) AS col_2 FROM _u, UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH(col)) - 1)) AS pos CROSS JOIN UNNEST(col) AS col_2 WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH(col) - 1) AND pos_2 = (ARRAY_LENGTH(col) - 1))", - "presto": "SELECT IF(_u_2.pos = _u_3.pos_2, _u_3.col_2) AS col_2 FROM _u, UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(col)))) AS _u_2(pos) CROSS JOIN UNNEST(col) WITH ORDINALITY AS _u_3(col_2, pos_2) WHERE _u_2.pos = _u_3.pos_2 OR (_u_2.pos > CARDINALITY(col) AND _u_3.pos_2 = CARDINALITY(col))", + "bigquery": "SELECT IF(pos = pos_2, col_2, NULL) AS col_2 FROM _u CROSS JOIN UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH(col)) - 1)) AS pos CROSS JOIN UNNEST(col) AS col_2 WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH(col) - 1) AND pos_2 = (ARRAY_LENGTH(col) - 1))", + "presto": "SELECT IF(_u_2.pos = _u_3.pos_2, _u_3.col_2) AS col_2 FROM _u CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(col)))) AS _u_2(pos) CROSS JOIN UNNEST(col) WITH ORDINALITY AS _u_3(col_2, pos_2) WHERE _u_2.pos = _u_3.pos_2 OR (_u_2.pos > CARDINALITY(col) AND _u_3.pos_2 = CARDINALITY(col))", "spark": "SELECT EXPLODE(col) FROM _u", }, ) self.validate_all( "SELECT EXPLODE(col) AS exploded FROM schema.tbl", write={ - "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.exploded) AS exploded FROM schema.tbl, UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(col)))) AS _u(pos) CROSS JOIN UNNEST(col) WITH ORDINALITY AS _u_2(exploded, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(col) AND _u_2.pos_2 = CARDINALITY(col))", + "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.exploded) AS exploded FROM schema.tbl CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(col)))) AS _u(pos) CROSS JOIN UNNEST(col) WITH ORDINALITY AS _u_2(exploded, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(col) AND _u_2.pos_2 = CARDINALITY(col))", }, ) self.validate_all( @@ -666,13 +678,13 @@ TBLPROPERTIES ( self.validate_all( "SELECT POSEXPLODE(ARRAY(2, 3)), EXPLODE(ARRAY(4, 5, 6)) FROM tbl", write={ - "bigquery": "SELECT IF(pos = pos_2, col, NULL) AS col, IF(pos = pos_2, pos_2, NULL) AS pos_2, IF(pos = pos_3, col_2, NULL) AS col_2 FROM tbl, UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH([2, 3]), ARRAY_LENGTH([4, 5, 6])) - 1)) AS pos CROSS JOIN UNNEST([2, 3]) AS col WITH OFFSET AS pos_2 CROSS JOIN UNNEST([4, 5, 6]) AS col_2 WITH OFFSET AS pos_3 WHERE (pos = pos_2 OR (pos > (ARRAY_LENGTH([2, 3]) - 1) AND pos_2 = (ARRAY_LENGTH([2, 3]) - 1))) AND (pos = pos_3 OR (pos > (ARRAY_LENGTH([4, 5, 6]) - 1) AND pos_3 = (ARRAY_LENGTH([4, 5, 6]) - 1)))", - "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col, IF(_u.pos = _u_2.pos_2, _u_2.pos_2) AS pos_2, IF(_u.pos = _u_3.pos_3, _u_3.col_2) AS col_2 FROM tbl, UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[2, 3]), CARDINALITY(ARRAY[4, 5, 6])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[2, 3]) WITH ORDINALITY AS _u_2(col, pos_2) CROSS JOIN UNNEST(ARRAY[4, 5, 6]) WITH ORDINALITY AS _u_3(col_2, pos_3) WHERE (_u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(ARRAY[2, 3]) AND _u_2.pos_2 = CARDINALITY(ARRAY[2, 3]))) AND (_u.pos = _u_3.pos_3 OR (_u.pos > CARDINALITY(ARRAY[4, 5, 6]) AND _u_3.pos_3 = CARDINALITY(ARRAY[4, 5, 6])))", + "bigquery": "SELECT IF(pos = pos_2, col, NULL) AS col, IF(pos = pos_2, pos_2, NULL) AS pos_2, IF(pos = pos_3, col_2, NULL) AS col_2 FROM tbl CROSS JOIN UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH([2, 3]), ARRAY_LENGTH([4, 5, 6])) - 1)) AS pos CROSS JOIN UNNEST([2, 3]) AS col WITH OFFSET AS pos_2 CROSS JOIN UNNEST([4, 5, 6]) AS col_2 WITH OFFSET AS pos_3 WHERE (pos = pos_2 OR (pos > (ARRAY_LENGTH([2, 3]) - 1) AND pos_2 = (ARRAY_LENGTH([2, 3]) - 1))) AND (pos = pos_3 OR (pos > (ARRAY_LENGTH([4, 5, 6]) - 1) AND pos_3 = (ARRAY_LENGTH([4, 5, 6]) - 1)))", + "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col, IF(_u.pos = _u_2.pos_2, _u_2.pos_2) AS pos_2, IF(_u.pos = _u_3.pos_3, _u_3.col_2) AS col_2 FROM tbl CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[2, 3]), CARDINALITY(ARRAY[4, 5, 6])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[2, 3]) WITH ORDINALITY AS _u_2(col, pos_2) CROSS JOIN UNNEST(ARRAY[4, 5, 6]) WITH ORDINALITY AS _u_3(col_2, pos_3) WHERE (_u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(ARRAY[2, 3]) AND _u_2.pos_2 = CARDINALITY(ARRAY[2, 3]))) AND (_u.pos = _u_3.pos_3 OR (_u.pos > CARDINALITY(ARRAY[4, 5, 6]) AND _u_3.pos_3 = CARDINALITY(ARRAY[4, 5, 6])))", }, ) self.validate_all( "SELECT col, pos, POSEXPLODE(ARRAY(2, 3)) FROM _u", write={ - "presto": "SELECT col, pos, IF(_u_2.pos_2 = _u_3.pos_3, _u_3.col_2) AS col_2, IF(_u_2.pos_2 = _u_3.pos_3, _u_3.pos_3) AS pos_3 FROM _u, UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[2, 3])))) AS _u_2(pos_2) CROSS JOIN UNNEST(ARRAY[2, 3]) WITH ORDINALITY AS _u_3(col_2, pos_3) WHERE _u_2.pos_2 = _u_3.pos_3 OR (_u_2.pos_2 > CARDINALITY(ARRAY[2, 3]) AND _u_3.pos_3 = CARDINALITY(ARRAY[2, 3]))", + "presto": "SELECT col, pos, IF(_u_2.pos_2 = _u_3.pos_3, _u_3.col_2) AS col_2, IF(_u_2.pos_2 = _u_3.pos_3, _u_3.pos_3) AS pos_3 FROM _u CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[2, 3])))) AS _u_2(pos_2) CROSS JOIN UNNEST(ARRAY[2, 3]) WITH ORDINALITY AS _u_3(col_2, pos_3) WHERE _u_2.pos_2 = _u_3.pos_3 OR (_u_2.pos_2 > CARDINALITY(ARRAY[2, 3]) AND _u_3.pos_3 = CARDINALITY(ARRAY[2, 3]))", }, ) |