diff options
Diffstat (limited to 'tests/dialects/test_spark.py')
| -rw-r--r-- | tests/dialects/test_spark.py | 72 |
1 files changed, 65 insertions, 7 deletions
diff --git a/tests/dialects/test_spark.py b/tests/dialects/test_spark.py index becb66a..2e43ba5 100644 --- a/tests/dialects/test_spark.py +++ b/tests/dialects/test_spark.py @@ -229,6 +229,7 @@ TBLPROPERTIES ( self.assertIsInstance(expr.args.get("ignore_nulls"), exp.Boolean) self.assertEqual(expr.sql(dialect="spark"), "ANY_VALUE(col, TRUE)") + self.validate_identity("SELECT * FROM t1 SEMI JOIN t2 ON t1.x = t2.x") 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") @@ -460,13 +461,13 @@ TBLPROPERTIES ( self.validate_all( "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC NULLS LAST, lname", write={ - "clickhouse": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname, lname NULLS FIRST", - "duckdb": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname, lname NULLS FIRST", - "postgres": "SELECT fname, lname, age FROM person ORDER BY age DESC, fname, lname NULLS FIRST", - "presto": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname, lname NULLS FIRST", - "hive": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname NULLS LAST, lname", - "spark": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname NULLS LAST, lname", - "snowflake": "SELECT fname, lname, age FROM person ORDER BY age DESC, fname, lname NULLS FIRST", + "clickhouse": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC, lname NULLS FIRST", + "duckdb": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC, lname NULLS FIRST", + "postgres": "SELECT fname, lname, age FROM person ORDER BY age DESC, fname ASC, lname NULLS FIRST", + "presto": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC, lname NULLS FIRST", + "hive": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC NULLS LAST, lname", + "spark": "SELECT fname, lname, age FROM person ORDER BY age DESC NULLS FIRST, fname ASC NULLS LAST, lname", + "snowflake": "SELECT fname, lname, age FROM person ORDER BY age DESC, fname ASC, lname NULLS FIRST", }, ) self.validate_all( @@ -583,3 +584,60 @@ TBLPROPERTIES ( "databricks": "WITH cte AS (SELECT cola FROM other_table) INSERT OVERWRITE TABLE table SELECT cola FROM cte", }, ) + + def test_explode_to_unnest(self): + 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(pos = pos_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 pos = pos_2 OR (pos > CARDINALITY(x) AND 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(pos = pos_2, 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 pos = pos_2 OR (pos > CARDINALITY(col) AND 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(pos = pos_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 pos = pos_2 OR (pos > CARDINALITY(col) AND pos_2 = CARDINALITY(col))", + }, + ) + self.validate_all( + "SELECT EXPLODE(ARRAY(1, 2))", + write={ + "bigquery": "SELECT IF(pos = pos_2, col, NULL) AS col FROM UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH([1, 2])) - 1)) AS pos CROSS JOIN UNNEST([1, 2]) AS col WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH([1, 2]) - 1) AND pos_2 = (ARRAY_LENGTH([1, 2]) - 1))", + "presto": "SELECT IF(pos = pos_2, col) AS col FROM UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[1, 2])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[1, 2]) WITH ORDINALITY AS _u_2(col, pos_2) WHERE pos = pos_2 OR (pos > CARDINALITY(ARRAY[1, 2]) AND pos_2 = CARDINALITY(ARRAY[1, 2]))", + }, + ) + self.validate_all( + "SELECT POSEXPLODE(ARRAY(2, 3)) AS x", + write={ + "bigquery": "SELECT IF(pos = pos_2, x, NULL) AS x, IF(pos = pos_2, pos_2, NULL) AS pos_2 FROM UNNEST(GENERATE_ARRAY(0, GREATEST(ARRAY_LENGTH([2, 3])) - 1)) AS pos CROSS JOIN UNNEST([2, 3]) AS x WITH OFFSET AS pos_2 WHERE pos = pos_2 OR (pos > (ARRAY_LENGTH([2, 3]) - 1) AND pos_2 = (ARRAY_LENGTH([2, 3]) - 1))", + "presto": "SELECT IF(pos = pos_2, x) AS x, IF(pos = pos_2, pos_2) AS pos_2 FROM UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[2, 3])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[2, 3]) WITH ORDINALITY AS _u_2(x, pos_2) WHERE pos = pos_2 OR (pos > CARDINALITY(ARRAY[2, 3]) AND pos_2 = CARDINALITY(ARRAY[2, 3]))", + }, + ) + self.validate_all( + "SELECT POSEXPLODE(x) AS (a, b)", + write={ + "presto": "SELECT IF(pos = a, b) AS b, IF(pos = a, a) AS a FROM UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(x)))) AS _u(pos) CROSS JOIN UNNEST(x) WITH ORDINALITY AS _u_2(b, a) WHERE pos = a OR (pos > CARDINALITY(x) AND a = CARDINALITY(x))", + }, + ) + 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(pos = pos_2, col) AS col, IF(pos = pos_2, pos_2) AS pos_2, IF(pos = pos_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 (pos = pos_2 OR (pos > CARDINALITY(ARRAY[2, 3]) AND pos_2 = CARDINALITY(ARRAY[2, 3]))) AND (pos = pos_3 OR (pos > CARDINALITY(ARRAY[4, 5, 6]) AND 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(pos_2 = pos_3, col_2) AS col_2, IF(pos_2 = pos_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 pos_2 = pos_3 OR (pos_2 > CARDINALITY(ARRAY[2, 3]) AND pos_3 = CARDINALITY(ARRAY[2, 3]))", + }, + ) |