summaryrefslogtreecommitdiffstats
path: root/tests/dialects
diff options
context:
space:
mode:
Diffstat (limited to 'tests/dialects')
-rw-r--r--tests/dialects/test_bigquery.py44
-rw-r--r--tests/dialects/test_clickhouse.py20
-rw-r--r--tests/dialects/test_dialect.py11
-rw-r--r--tests/dialects/test_duckdb.py4
-rw-r--r--tests/dialects/test_hive.py8
-rw-r--r--tests/dialects/test_mysql.py2
-rw-r--r--tests/dialects/test_oracle.py1
-rw-r--r--tests/dialects/test_postgres.py450
-rw-r--r--tests/dialects/test_redshift.py6
-rw-r--r--tests/dialects/test_snowflake.py46
-rw-r--r--tests/dialects/test_spark.py20
-rw-r--r--tests/dialects/test_teradata.py12
-rw-r--r--tests/dialects/test_tsql.py30
13 files changed, 399 insertions, 255 deletions
diff --git a/tests/dialects/test_bigquery.py b/tests/dialects/test_bigquery.py
index 48480f9..2c8ac7b 100644
--- a/tests/dialects/test_bigquery.py
+++ b/tests/dialects/test_bigquery.py
@@ -9,6 +9,7 @@ from sqlglot import (
transpile,
)
from sqlglot.helper import logger as helper_logger
+from sqlglot.parser import logger as parser_logger
from tests.dialects.test_dialect import Validator
@@ -17,6 +18,29 @@ class TestBigQuery(Validator):
maxDiff = None
def test_bigquery(self):
+ self.validate_identity(
+ "create or replace view test (tenant_id OPTIONS(description='Test description on table creation')) select 1 as tenant_id, 1 as customer_id;",
+ "CREATE OR REPLACE VIEW test (tenant_id OPTIONS (description='Test description on table creation')) AS SELECT 1 AS tenant_id, 1 AS customer_id",
+ )
+
+ with self.assertLogs(helper_logger) as cm:
+ statements = parse(
+ """
+ BEGIN
+ DECLARE 1;
+ IF from_date IS NULL THEN SET x = 1;
+ END IF;
+ END
+ """,
+ read="bigquery",
+ )
+ self.assertIn("unsupported syntax", cm.output[0])
+
+ for actual, expected in zip(
+ statements, ("BEGIN DECLARE 1", "IF from_date IS NULL THEN SET x = 1", "END IF", "END")
+ ):
+ self.assertEqual(actual.sql(dialect="bigquery"), expected)
+
with self.assertLogs(helper_logger) as cm:
self.validate_identity(
"SELECT * FROM t AS t(c1, c2)",
@@ -77,14 +101,16 @@ class TestBigQuery(Validator):
with self.assertRaises(ParseError):
transpile("DATE_ADD(x, day)", read="bigquery")
- for_in_stmts = parse(
- "FOR record IN (SELECT word FROM shakespeare) DO SELECT record.word; END FOR;",
- read="bigquery",
- )
- self.assertEqual(
- [s.sql(dialect="bigquery") for s in for_in_stmts],
- ["FOR record IN (SELECT word FROM shakespeare) DO SELECT record.word", "END FOR"],
- )
+ with self.assertLogs(parser_logger) as cm:
+ for_in_stmts = parse(
+ "FOR record IN (SELECT word FROM shakespeare) DO SELECT record.word; END FOR;",
+ read="bigquery",
+ )
+ self.assertEqual(
+ [s.sql(dialect="bigquery") for s in for_in_stmts],
+ ["FOR record IN (SELECT word FROM shakespeare) DO SELECT record.word", "END FOR"],
+ )
+ assert "'END FOR'" in cm.output[0]
self.validate_identity("SELECT * FROM dataset.my_table TABLESAMPLE SYSTEM (10 PERCENT)")
self.validate_identity("TIME('2008-12-25 15:30:00+08')")
@@ -135,7 +161,7 @@ class TestBigQuery(Validator):
self.validate_identity("""CREATE TABLE x (a STRUCT<b STRING OPTIONS (description='b')>)""")
self.validate_identity("CAST(x AS TIMESTAMP)")
self.validate_identity("REGEXP_EXTRACT(`foo`, 'bar: (.+?)', 1, 1)")
- self.validate_identity("BEGIN A B C D E F")
+ self.validate_identity("BEGIN DECLARE y INT64", check_command_warning=True)
self.validate_identity("BEGIN TRANSACTION")
self.validate_identity("COMMIT TRANSACTION")
self.validate_identity("ROLLBACK TRANSACTION")
diff --git a/tests/dialects/test_clickhouse.py b/tests/dialects/test_clickhouse.py
index 84903aa..f36af41 100644
--- a/tests/dialects/test_clickhouse.py
+++ b/tests/dialects/test_clickhouse.py
@@ -42,7 +42,6 @@ class TestClickhouse(Validator):
self.validate_identity("SELECT isNaN(1.0)")
self.validate_identity("SELECT startsWith('Spider-Man', 'Spi')")
self.validate_identity("SELECT xor(TRUE, FALSE)")
- self.validate_identity("ATTACH DATABASE DEFAULT ENGINE = ORDINARY")
self.validate_identity("CAST(['hello'], 'Array(Enum8(''hello'' = 1))')")
self.validate_identity("SELECT x, COUNT() FROM y GROUP BY x WITH TOTALS")
self.validate_identity("SELECT INTERVAL t.days DAY")
@@ -76,6 +75,9 @@ class TestClickhouse(Validator):
self.validate_identity("CAST(x as MEDIUMINT)", "CAST(x AS Int32)")
self.validate_identity("SELECT arrayJoin([1, 2, 3] AS src) AS dst, 'Hello', src")
self.validate_identity(
+ "ATTACH DATABASE DEFAULT ENGINE = ORDINARY", check_command_warning=True
+ )
+ self.validate_identity(
"SELECT n, source FROM (SELECT toFloat32(number % 10) AS n, 'original' AS source FROM numbers(10) WHERE number % 3 = 1) ORDER BY n WITH FILL"
)
self.validate_identity(
@@ -728,3 +730,19 @@ LIFETIME(MIN 0 MAX 0)""",
)
self.validate_identity("""CREATE TABLE ip_data (ip4 IPv4, ip6 IPv6) ENGINE=TinyLog()""")
self.validate_identity("""CREATE TABLE dates (dt1 Date32) ENGINE=TinyLog()""")
+ self.validate_all(
+ """
+ CREATE TABLE t (
+ a AggregateFunction(quantiles(0.5, 0.9), UInt64),
+ b AggregateFunction(quantiles, UInt64),
+ c SimpleAggregateFunction(sum, Float64)
+ )""",
+ write={
+ "clickhouse": """CREATE TABLE t (
+ a AggregateFunction(quantiles(0.5, 0.9), UInt64),
+ b AggregateFunction(quantiles, UInt64),
+ c SimpleAggregateFunction(sum, Float64)
+)"""
+ },
+ pretty=True,
+ )
diff --git a/tests/dialects/test_dialect.py b/tests/dialects/test_dialect.py
index 3cf4ddc..22e7d49 100644
--- a/tests/dialects/test_dialect.py
+++ b/tests/dialects/test_dialect.py
@@ -11,6 +11,7 @@ from sqlglot import (
parse_one,
)
from sqlglot.dialects import BigQuery, Hive, Snowflake
+from sqlglot.parser import logger as parser_logger
class Validator(unittest.TestCase):
@@ -19,8 +20,14 @@ class Validator(unittest.TestCase):
def parse_one(self, sql):
return parse_one(sql, read=self.dialect)
- def validate_identity(self, sql, write_sql=None, pretty=False):
- expression = self.parse_one(sql)
+ def validate_identity(self, sql, write_sql=None, pretty=False, check_command_warning=False):
+ if check_command_warning:
+ with self.assertLogs(parser_logger) as cm:
+ expression = self.parse_one(sql)
+ assert f"'{sql[:100]}' contains unsupported syntax" in cm.output[0]
+ else:
+ expression = self.parse_one(sql)
+
self.assertEqual(write_sql or sql, expression.sql(dialect=self.dialect, pretty=pretty))
return expression
diff --git a/tests/dialects/test_duckdb.py b/tests/dialects/test_duckdb.py
index e5f7e0c..f3b41b4 100644
--- a/tests/dialects/test_duckdb.py
+++ b/tests/dialects/test_duckdb.py
@@ -196,11 +196,13 @@ class TestDuckDB(Validator):
self.validate_identity("SELECT ROW(x, x + 1, y) FROM (SELECT 1 AS x, 'a' AS y)")
self.validate_identity("SELECT (x, x + 1, y) FROM (SELECT 1 AS x, 'a' AS y)")
self.validate_identity("SELECT a.x FROM (SELECT {'x': 1, 'y': 2, 'z': 3} AS a)")
- self.validate_identity("ATTACH DATABASE ':memory:' AS new_database")
self.validate_identity("FROM x SELECT x UNION SELECT 1", "SELECT x FROM x UNION SELECT 1")
self.validate_identity("FROM (FROM tbl)", "SELECT * FROM (SELECT * FROM tbl)")
self.validate_identity("FROM tbl", "SELECT * FROM tbl")
self.validate_identity(
+ "ATTACH DATABASE ':memory:' AS new_database", check_command_warning=True
+ )
+ self.validate_identity(
"SELECT {'yes': 'duck', 'maybe': 'goose', 'huh': NULL, 'no': 'heron'}"
)
self.validate_identity(
diff --git a/tests/dialects/test_hive.py b/tests/dialects/test_hive.py
index 8b5a945..d1b7589 100644
--- a/tests/dialects/test_hive.py
+++ b/tests/dialects/test_hive.py
@@ -399,6 +399,7 @@ class TestHive(Validator):
)
def test_hive(self):
+ self.validate_identity("SET hiveconf:some_var = 5", check_command_warning=True)
self.validate_identity("(VALUES (1 AS a, 2 AS b, 3))")
self.validate_identity("SELECT * FROM my_table TIMESTAMP AS OF DATE_ADD(CURRENT_DATE, -1)")
self.validate_identity("SELECT * FROM my_table VERSION AS OF DATE_ADD(CURRENT_DATE, -1)")
@@ -441,13 +442,6 @@ class TestHive(Validator):
)
self.validate_all(
- "SET hiveconf:some_var = 5",
- write={
- "hive": "SET hiveconf:some_var = 5",
- "spark": "SET hiveconf:some_var = 5",
- },
- )
- self.validate_all(
"SELECT ${hiveconf:some_var}",
write={
"hive": "SELECT ${hiveconf:some_var}",
diff --git a/tests/dialects/test_mysql.py b/tests/dialects/test_mysql.py
index 85bf261..3a3e49e 100644
--- a/tests/dialects/test_mysql.py
+++ b/tests/dialects/test_mysql.py
@@ -126,7 +126,7 @@ class TestMySQL(Validator):
self.validate_identity("SELECT DATE_FORMAT(NOW(), '%Y-%m-%d %H:%i:00.0000')")
self.validate_identity("SELECT @var1 := 1, @var2")
self.validate_identity("UNLOCK TABLES")
- self.validate_identity("LOCK TABLES `app_fields` WRITE")
+ self.validate_identity("LOCK TABLES `app_fields` WRITE", check_command_warning=True)
self.validate_identity("SELECT 1 XOR 0")
self.validate_identity("SELECT 1 && 0", "SELECT 1 AND 0")
self.validate_identity("SELECT /*+ BKA(t1) NO_BKA(t2) */ * FROM t1 INNER JOIN t2")
diff --git a/tests/dialects/test_oracle.py b/tests/dialects/test_oracle.py
index fce714e..bc8f8bb 100644
--- a/tests/dialects/test_oracle.py
+++ b/tests/dialects/test_oracle.py
@@ -12,6 +12,7 @@ class TestOracle(Validator):
exp.AlterTable
)
+ self.validate_identity("TIMESTAMP(3) WITH TIME ZONE")
self.validate_identity("CURRENT_TIMESTAMP(precision)")
self.validate_identity("ALTER TABLE tbl_name DROP FOREIGN KEY fk_symbol")
self.validate_identity("ALTER TABLE Payments ADD Stock NUMBER NOT NULL")
diff --git a/tests/dialects/test_postgres.py b/tests/dialects/test_postgres.py
index f46eeba..dc00c85 100644
--- a/tests/dialects/test_postgres.py
+++ b/tests/dialects/test_postgres.py
@@ -7,214 +7,10 @@ class TestPostgres(Validator):
maxDiff = None
dialect = "postgres"
- def test_ddl(self):
- expr = parse_one("CREATE TABLE t (x INTERVAL day)", read="postgres")
- cdef = expr.find(exp.ColumnDef)
- cdef.args["kind"].assert_is(exp.DataType)
- self.assertEqual(expr.sql(dialect="postgres"), "CREATE TABLE t (x INTERVAL DAY)")
-
- self.validate_identity("CREATE INDEX idx_x ON x USING BTREE(x, y) WHERE (NOT y IS NULL)")
- self.validate_identity("CREATE TABLE test (elems JSONB[])")
- self.validate_identity("CREATE TABLE public.y (x TSTZRANGE NOT NULL)")
- self.validate_identity("CREATE TABLE test (foo HSTORE)")
- self.validate_identity("CREATE TABLE test (foo JSONB)")
- self.validate_identity("CREATE TABLE test (foo VARCHAR(64)[])")
- self.validate_identity("CREATE TABLE test (foo INT) PARTITION BY HASH(foo)")
- self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING a")
- self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING a, b")
- self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING *")
- self.validate_identity("UPDATE tbl_name SET foo = 123 RETURNING a")
- self.validate_identity("CREATE TABLE cities_partdef PARTITION OF cities DEFAULT")
- self.validate_identity(
- "CREATE CONSTRAINT TRIGGER my_trigger AFTER INSERT OR DELETE OR UPDATE OF col_a, col_b ON public.my_table DEFERRABLE INITIALLY DEFERRED FOR EACH ROW EXECUTE FUNCTION do_sth()"
- )
- self.validate_identity(
- "CREATE TABLE cust_part3 PARTITION OF customers FOR VALUES WITH (MODULUS 3, REMAINDER 2)"
- )
- self.validate_identity(
- "CREATE TABLE measurement_y2016m07 PARTITION OF measurement (unitsales DEFAULT 0) FOR VALUES FROM ('2016-07-01') TO ('2016-08-01')"
- )
- self.validate_identity(
- "CREATE TABLE measurement_ym_older PARTITION OF measurement_year_month FOR VALUES FROM (MINVALUE, MINVALUE) TO (2016, 11)"
- )
- self.validate_identity(
- "CREATE TABLE measurement_ym_y2016m11 PARTITION OF measurement_year_month FOR VALUES FROM (2016, 11) TO (2016, 12)"
- )
- self.validate_identity(
- "CREATE TABLE cities_ab PARTITION OF cities (CONSTRAINT city_id_nonzero CHECK (city_id <> 0)) FOR VALUES IN ('a', 'b')"
- )
- self.validate_identity(
- "CREATE TABLE cities_ab PARTITION OF cities (CONSTRAINT city_id_nonzero CHECK (city_id <> 0)) FOR VALUES IN ('a', 'b') PARTITION BY RANGE(population)"
- )
- self.validate_identity(
- "CREATE INDEX foo ON bar.baz USING btree(col1 varchar_pattern_ops ASC, col2)"
- )
- self.validate_identity(
- "CREATE INDEX index_issues_on_title_trigram ON public.issues USING gin(title public.gin_trgm_ops)"
- )
- self.validate_identity(
- "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO NOTHING RETURNING *"
- )
- self.validate_identity(
- "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO UPDATE SET x.id = 1 RETURNING *"
- )
- self.validate_identity(
- "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO UPDATE SET x.id = excluded.id RETURNING *"
- )
- self.validate_identity(
- "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT ON CONSTRAINT pkey DO NOTHING RETURNING *"
- )
- self.validate_identity(
- "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT ON CONSTRAINT pkey DO UPDATE SET x.id = 1 RETURNING *"
- )
- self.validate_identity(
- "DELETE FROM event USING sales AS s WHERE event.eventid = s.eventid RETURNING a"
- )
- self.validate_identity(
- "CREATE TABLE test (x TIMESTAMP WITHOUT TIME ZONE[][])",
- "CREATE TABLE test (x TIMESTAMP[][])",
- )
- self.validate_identity(
- "CREATE UNLOGGED TABLE foo AS WITH t(c) AS (SELECT 1) SELECT * FROM (SELECT c AS c FROM t) AS temp"
- )
- self.validate_identity(
- "WITH t(c) AS (SELECT 1) SELECT * INTO UNLOGGED foo FROM (SELECT c AS c FROM t) AS temp"
- )
-
- self.validate_all(
- "CREATE OR REPLACE FUNCTION function_name (input_a character varying DEFAULT NULL::character varying)",
- write={
- "postgres": "CREATE OR REPLACE FUNCTION function_name(input_a VARCHAR DEFAULT CAST(NULL AS VARCHAR))",
- },
- )
- self.validate_all(
- "CREATE TABLE products (product_no INT UNIQUE, name TEXT, price DECIMAL)",
- write={
- "postgres": "CREATE TABLE products (product_no INT UNIQUE, name TEXT, price DECIMAL)"
- },
- )
- self.validate_all(
- "CREATE TABLE products (product_no INT CONSTRAINT must_be_different UNIQUE, name TEXT CONSTRAINT present NOT NULL, price DECIMAL)",
- write={
- "postgres": "CREATE TABLE products (product_no INT CONSTRAINT must_be_different UNIQUE, name TEXT CONSTRAINT present NOT NULL, price DECIMAL)"
- },
- )
- self.validate_all(
- "CREATE TABLE products (product_no INT, name TEXT, price DECIMAL, UNIQUE (product_no, name))",
- write={
- "postgres": "CREATE TABLE products (product_no INT, name TEXT, price DECIMAL, UNIQUE (product_no, name))"
- },
- )
- self.validate_all(
- "CREATE TABLE products ("
- "product_no INT UNIQUE,"
- " name TEXT,"
- " price DECIMAL CHECK (price > 0),"
- " discounted_price DECIMAL CONSTRAINT positive_discount CHECK (discounted_price > 0),"
- " CHECK (product_no > 1),"
- " CONSTRAINT valid_discount CHECK (price > discounted_price))",
- write={
- "postgres": "CREATE TABLE products ("
- "product_no INT UNIQUE,"
- " name TEXT,"
- " price DECIMAL CHECK (price > 0),"
- " discounted_price DECIMAL CONSTRAINT positive_discount CHECK (discounted_price > 0),"
- " CHECK (product_no > 1),"
- " CONSTRAINT valid_discount CHECK (price > discounted_price))"
- },
- )
- self.validate_identity(
- """
- CREATE INDEX index_ci_builds_on_commit_id_and_artifacts_expireatandidpartial
- ON public.ci_builds
- USING btree (commit_id, artifacts_expire_at, id)
- WHERE (
- ((type)::text = 'Ci::Build'::text)
- AND ((retried = false) OR (retried IS NULL))
- AND ((name)::text = ANY (ARRAY[
- ('sast'::character varying)::text,
- ('dependency_scanning'::character varying)::text,
- ('sast:container'::character varying)::text,
- ('container_scanning'::character varying)::text,
- ('dast'::character varying)::text
- ]))
- )
- """,
- "CREATE INDEX index_ci_builds_on_commit_id_and_artifacts_expireatandidpartial ON public.ci_builds USING btree(commit_id, artifacts_expire_at, id) WHERE ((CAST((type) AS TEXT) = CAST('Ci::Build' AS TEXT)) AND ((retried = FALSE) OR (retried IS NULL)) AND (CAST((name) AS TEXT) = ANY (ARRAY[CAST((CAST('sast' AS VARCHAR)) AS TEXT), CAST((CAST('dependency_scanning' AS VARCHAR)) AS TEXT), CAST((CAST('sast:container' AS VARCHAR)) AS TEXT), CAST((CAST('container_scanning' AS VARCHAR)) AS TEXT), CAST((CAST('dast' AS VARCHAR)) AS TEXT)])))",
- )
- self.validate_identity(
- "CREATE INDEX index_ci_pipelines_on_project_idandrefandiddesc ON public.ci_pipelines USING btree(project_id, ref, id DESC)"
- )
-
- with self.assertRaises(ParseError):
- transpile("CREATE TABLE products (price DECIMAL CHECK price > 0)", read="postgres")
- with self.assertRaises(ParseError):
- transpile(
- "CREATE TABLE products (price DECIMAL, CHECK price > 1)",
- read="postgres",
- )
-
- def test_unnest(self):
- self.validate_identity(
- "SELECT * FROM UNNEST(ARRAY[1, 2], ARRAY['foo', 'bar', 'baz']) AS x(a, b)"
- )
-
- self.validate_all(
- "SELECT UNNEST(c) FROM t",
- write={
- "hive": "SELECT EXPLODE(c) FROM t",
- "postgres": "SELECT UNNEST(c) FROM t",
- "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM t CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(c)))) AS _u(pos) CROSS JOIN UNNEST(c) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(c) AND _u_2.pos_2 = CARDINALITY(c))",
- },
- )
- self.validate_all(
- "SELECT UNNEST(ARRAY[1])",
- write={
- "hive": "SELECT EXPLODE(ARRAY(1))",
- "postgres": "SELECT UNNEST(ARRAY[1])",
- "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[1])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[1]) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(ARRAY[1]) AND _u_2.pos_2 = CARDINALITY(ARRAY[1]))",
- },
- )
-
- def test_array_offset(self):
- with self.assertLogs(helper_logger) as cm:
- self.validate_all(
- "SELECT col[1]",
- write={
- "bigquery": "SELECT col[0]",
- "duckdb": "SELECT col[1]",
- "hive": "SELECT col[0]",
- "postgres": "SELECT col[1]",
- "presto": "SELECT col[1]",
- },
- )
-
- self.assertEqual(
- cm.output,
- [
- "WARNING:sqlglot:Applying array index offset (-1)",
- "WARNING:sqlglot:Applying array index offset (1)",
- "WARNING:sqlglot:Applying array index offset (1)",
- "WARNING:sqlglot:Applying array index offset (1)",
- ],
- )
-
- def test_operator(self):
- expr = parse_one("1 OPERATOR(+) 2 OPERATOR(*) 3", read="postgres")
-
- expr.left.assert_is(exp.Operator)
- expr.left.left.assert_is(exp.Literal)
- expr.left.right.assert_is(exp.Literal)
- expr.right.assert_is(exp.Literal)
- self.assertEqual(expr.sql(dialect="postgres"), "1 OPERATOR(+) 2 OPERATOR(*) 3")
-
- self.validate_identity("SELECT operator FROM t")
- self.validate_identity("SELECT 1 OPERATOR(+) 2")
- self.validate_identity("SELECT 1 OPERATOR(+) /* foo */ 2")
- self.validate_identity("SELECT 1 OPERATOR(pg_catalog.+) 2")
-
def test_postgres(self):
- self.validate_identity("EXEC AS myfunc @id = 123")
+ self.validate_identity("SELECT CURRENT_USER")
+ self.validate_identity("CAST(1 AS DECIMAL) / CAST(2 AS DECIMAL) * -100")
+ self.validate_identity("EXEC AS myfunc @id = 123", check_command_warning=True)
expr = parse_one(
"SELECT * FROM r CROSS JOIN LATERAL UNNEST(ARRAY[1]) AS s(location)", read="postgres"
@@ -782,6 +578,246 @@ class TestPostgres(Validator):
self.assertIsInstance(parse_one("id::UUID", read="postgres"), exp.Cast)
+ def test_ddl(self):
+ expr = parse_one("CREATE TABLE t (x INTERVAL day)", read="postgres")
+ cdef = expr.find(exp.ColumnDef)
+ cdef.args["kind"].assert_is(exp.DataType)
+ self.assertEqual(expr.sql(dialect="postgres"), "CREATE TABLE t (x INTERVAL DAY)")
+
+ self.validate_identity("CREATE INDEX et_vid_idx ON et(vid) INCLUDE (fid)")
+ self.validate_identity("CREATE INDEX idx_x ON x USING BTREE(x, y) WHERE (NOT y IS NULL)")
+ self.validate_identity("CREATE TABLE test (elems JSONB[])")
+ self.validate_identity("CREATE TABLE public.y (x TSTZRANGE NOT NULL)")
+ self.validate_identity("CREATE TABLE test (foo HSTORE)")
+ self.validate_identity("CREATE TABLE test (foo JSONB)")
+ self.validate_identity("CREATE TABLE test (foo VARCHAR(64)[])")
+ self.validate_identity("CREATE TABLE test (foo INT) PARTITION BY HASH(foo)")
+ self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING a")
+ self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING a, b")
+ self.validate_identity("INSERT INTO x VALUES (1, 'a', 2.0) RETURNING *")
+ self.validate_identity("UPDATE tbl_name SET foo = 123 RETURNING a")
+ self.validate_identity("CREATE TABLE cities_partdef PARTITION OF cities DEFAULT")
+ self.validate_identity("CREATE TABLE t (c CHAR(2) UNIQUE NOT NULL) INHERITS (t1)")
+ self.validate_identity("CREATE TABLE s.t (c CHAR(2) UNIQUE NOT NULL) INHERITS (s.t1, s.t2)")
+ self.validate_identity("CREATE FUNCTION x(INT) RETURNS INT SET search_path = 'public'")
+ self.validate_identity(
+ "CREATE TABLE cust_part3 PARTITION OF customers FOR VALUES WITH (MODULUS 3, REMAINDER 2)"
+ )
+ self.validate_identity(
+ "CREATE TABLE measurement_y2016m07 PARTITION OF measurement (unitsales DEFAULT 0) FOR VALUES FROM ('2016-07-01') TO ('2016-08-01')"
+ )
+ self.validate_identity(
+ "CREATE TABLE measurement_ym_older PARTITION OF measurement_year_month FOR VALUES FROM (MINVALUE, MINVALUE) TO (2016, 11)"
+ )
+ self.validate_identity(
+ "CREATE TABLE measurement_ym_y2016m11 PARTITION OF measurement_year_month FOR VALUES FROM (2016, 11) TO (2016, 12)"
+ )
+ self.validate_identity(
+ "CREATE TABLE cities_ab PARTITION OF cities (CONSTRAINT city_id_nonzero CHECK (city_id <> 0)) FOR VALUES IN ('a', 'b')"
+ )
+ self.validate_identity(
+ "CREATE TABLE cities_ab PARTITION OF cities (CONSTRAINT city_id_nonzero CHECK (city_id <> 0)) FOR VALUES IN ('a', 'b') PARTITION BY RANGE(population)"
+ )
+ self.validate_identity(
+ "CREATE INDEX foo ON bar.baz USING btree(col1 varchar_pattern_ops ASC, col2)"
+ )
+ self.validate_identity(
+ "CREATE INDEX index_issues_on_title_trigram ON public.issues USING gin(title public.gin_trgm_ops)"
+ )
+ self.validate_identity(
+ "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO NOTHING RETURNING *"
+ )
+ self.validate_identity(
+ "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO UPDATE SET x.id = 1 RETURNING *"
+ )
+ self.validate_identity(
+ "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT (id) DO UPDATE SET x.id = excluded.id RETURNING *"
+ )
+ self.validate_identity(
+ "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT ON CONSTRAINT pkey DO NOTHING RETURNING *"
+ )
+ self.validate_identity(
+ "INSERT INTO x VALUES (1, 'a', 2.0) ON CONFLICT ON CONSTRAINT pkey DO UPDATE SET x.id = 1 RETURNING *"
+ )
+ self.validate_identity(
+ "DELETE FROM event USING sales AS s WHERE event.eventid = s.eventid RETURNING a"
+ )
+ self.validate_identity(
+ "WITH t(c) AS (SELECT 1) SELECT * INTO UNLOGGED foo FROM (SELECT c AS c FROM t) AS temp"
+ )
+ self.validate_identity(
+ "CREATE TABLE test (x TIMESTAMP WITHOUT TIME ZONE[][])",
+ "CREATE TABLE test (x TIMESTAMP[][])",
+ )
+ self.validate_identity(
+ "CREATE FUNCTION add(INT, INT) RETURNS INT SET search_path TO 'public' AS 'select $1 + $2;' LANGUAGE SQL IMMUTABLE",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE FUNCTION x(INT) RETURNS INT SET foo FROM CURRENT",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE FUNCTION add(integer, integer) RETURNS integer AS 'select $1 + $2;' LANGUAGE SQL IMMUTABLE RETURNS NULL ON NULL INPUT",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE FUNCTION add(integer, integer) RETURNS integer AS 'select $1 + $2;' LANGUAGE SQL IMMUTABLE CALLED ON NULL INPUT",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE FUNCTION add(integer, integer) RETURNS integer AS 'select $1 + $2;' LANGUAGE SQL IMMUTABLE STRICT",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE CONSTRAINT TRIGGER my_trigger AFTER INSERT OR DELETE OR UPDATE OF col_a, col_b ON public.my_table DEFERRABLE INITIALLY DEFERRED FOR EACH ROW EXECUTE FUNCTION do_sth()",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE UNLOGGED TABLE foo AS WITH t(c) AS (SELECT 1) SELECT * FROM (SELECT c AS c FROM t) AS temp",
+ check_command_warning=True,
+ )
+ self.validate_identity(
+ "CREATE FUNCTION x(INT) RETURNS INT SET search_path TO 'public'",
+ "CREATE FUNCTION x(INT) RETURNS INT SET search_path = 'public'",
+ )
+ self.validate_identity(
+ "CREATE TABLE test (x TIMESTAMP WITHOUT TIME ZONE[][])",
+ "CREATE TABLE test (x TIMESTAMP[][])",
+ )
+
+ self.validate_all(
+ "CREATE OR REPLACE FUNCTION function_name (input_a character varying DEFAULT NULL::character varying)",
+ write={
+ "postgres": "CREATE OR REPLACE FUNCTION function_name(input_a VARCHAR DEFAULT CAST(NULL AS VARCHAR))",
+ },
+ )
+ self.validate_all(
+ "CREATE TABLE products (product_no INT UNIQUE, name TEXT, price DECIMAL)",
+ write={
+ "postgres": "CREATE TABLE products (product_no INT UNIQUE, name TEXT, price DECIMAL)"
+ },
+ )
+ self.validate_all(
+ "CREATE TABLE products (product_no INT CONSTRAINT must_be_different UNIQUE, name TEXT CONSTRAINT present NOT NULL, price DECIMAL)",
+ write={
+ "postgres": "CREATE TABLE products (product_no INT CONSTRAINT must_be_different UNIQUE, name TEXT CONSTRAINT present NOT NULL, price DECIMAL)"
+ },
+ )
+ self.validate_all(
+ "CREATE TABLE products (product_no INT, name TEXT, price DECIMAL, UNIQUE (product_no, name))",
+ write={
+ "postgres": "CREATE TABLE products (product_no INT, name TEXT, price DECIMAL, UNIQUE (product_no, name))"
+ },
+ )
+ self.validate_all(
+ "CREATE TABLE products ("
+ "product_no INT UNIQUE,"
+ " name TEXT,"
+ " price DECIMAL CHECK (price > 0),"
+ " discounted_price DECIMAL CONSTRAINT positive_discount CHECK (discounted_price > 0),"
+ " CHECK (product_no > 1),"
+ " CONSTRAINT valid_discount CHECK (price > discounted_price))",
+ write={
+ "postgres": "CREATE TABLE products ("
+ "product_no INT UNIQUE,"
+ " name TEXT,"
+ " price DECIMAL CHECK (price > 0),"
+ " discounted_price DECIMAL CONSTRAINT positive_discount CHECK (discounted_price > 0),"
+ " CHECK (product_no > 1),"
+ " CONSTRAINT valid_discount CHECK (price > discounted_price))"
+ },
+ )
+ self.validate_identity(
+ """
+ CREATE INDEX index_ci_builds_on_commit_id_and_artifacts_expireatandidpartial
+ ON public.ci_builds
+ USING btree (commit_id, artifacts_expire_at, id)
+ WHERE (
+ ((type)::text = 'Ci::Build'::text)
+ AND ((retried = false) OR (retried IS NULL))
+ AND ((name)::text = ANY (ARRAY[
+ ('sast'::character varying)::text,
+ ('dependency_scanning'::character varying)::text,
+ ('sast:container'::character varying)::text,
+ ('container_scanning'::character varying)::text,
+ ('dast'::character varying)::text
+ ]))
+ )
+ """,
+ "CREATE INDEX index_ci_builds_on_commit_id_and_artifacts_expireatandidpartial ON public.ci_builds USING btree(commit_id, artifacts_expire_at, id) WHERE ((CAST((type) AS TEXT) = CAST('Ci::Build' AS TEXT)) AND ((retried = FALSE) OR (retried IS NULL)) AND (CAST((name) AS TEXT) = ANY (ARRAY[CAST((CAST('sast' AS VARCHAR)) AS TEXT), CAST((CAST('dependency_scanning' AS VARCHAR)) AS TEXT), CAST((CAST('sast:container' AS VARCHAR)) AS TEXT), CAST((CAST('container_scanning' AS VARCHAR)) AS TEXT), CAST((CAST('dast' AS VARCHAR)) AS TEXT)])))",
+ )
+ self.validate_identity(
+ "CREATE INDEX index_ci_pipelines_on_project_idandrefandiddesc ON public.ci_pipelines USING btree(project_id, ref, id DESC)"
+ )
+
+ with self.assertRaises(ParseError):
+ transpile("CREATE TABLE products (price DECIMAL CHECK price > 0)", read="postgres")
+ with self.assertRaises(ParseError):
+ transpile(
+ "CREATE TABLE products (price DECIMAL, CHECK price > 1)",
+ read="postgres",
+ )
+
+ def test_unnest(self):
+ self.validate_identity(
+ "SELECT * FROM UNNEST(ARRAY[1, 2], ARRAY['foo', 'bar', 'baz']) AS x(a, b)"
+ )
+
+ self.validate_all(
+ "SELECT UNNEST(c) FROM t",
+ write={
+ "hive": "SELECT EXPLODE(c) FROM t",
+ "postgres": "SELECT UNNEST(c) FROM t",
+ "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM t CROSS JOIN UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(c)))) AS _u(pos) CROSS JOIN UNNEST(c) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(c) AND _u_2.pos_2 = CARDINALITY(c))",
+ },
+ )
+ self.validate_all(
+ "SELECT UNNEST(ARRAY[1])",
+ write={
+ "hive": "SELECT EXPLODE(ARRAY(1))",
+ "postgres": "SELECT UNNEST(ARRAY[1])",
+ "presto": "SELECT IF(_u.pos = _u_2.pos_2, _u_2.col) AS col FROM UNNEST(SEQUENCE(1, GREATEST(CARDINALITY(ARRAY[1])))) AS _u(pos) CROSS JOIN UNNEST(ARRAY[1]) WITH ORDINALITY AS _u_2(col, pos_2) WHERE _u.pos = _u_2.pos_2 OR (_u.pos > CARDINALITY(ARRAY[1]) AND _u_2.pos_2 = CARDINALITY(ARRAY[1]))",
+ },
+ )
+
+ def test_array_offset(self):
+ with self.assertLogs(helper_logger) as cm:
+ self.validate_all(
+ "SELECT col[1]",
+ write={
+ "bigquery": "SELECT col[0]",
+ "duckdb": "SELECT col[1]",
+ "hive": "SELECT col[0]",
+ "postgres": "SELECT col[1]",
+ "presto": "SELECT col[1]",
+ },
+ )
+
+ self.assertEqual(
+ cm.output,
+ [
+ "WARNING:sqlglot:Applying array index offset (-1)",
+ "WARNING:sqlglot:Applying array index offset (1)",
+ "WARNING:sqlglot:Applying array index offset (1)",
+ "WARNING:sqlglot:Applying array index offset (1)",
+ ],
+ )
+
+ def test_operator(self):
+ expr = parse_one("1 OPERATOR(+) 2 OPERATOR(*) 3", read="postgres")
+
+ expr.left.assert_is(exp.Operator)
+ expr.left.left.assert_is(exp.Literal)
+ expr.left.right.assert_is(exp.Literal)
+ expr.right.assert_is(exp.Literal)
+ self.assertEqual(expr.sql(dialect="postgres"), "1 OPERATOR(+) 2 OPERATOR(*) 3")
+
+ self.validate_identity("SELECT operator FROM t")
+ self.validate_identity("SELECT 1 OPERATOR(+) 2")
+ self.validate_identity("SELECT 1 OPERATOR(+) /* foo */ 2")
+ self.validate_identity("SELECT 1 OPERATOR(pg_catalog.+) 2")
+
def test_bool_or(self):
self.validate_all(
"SELECT a, LOGICAL_OR(b) FROM table GROUP BY a",
diff --git a/tests/dialects/test_redshift.py b/tests/dialects/test_redshift.py
index 88fef67..9ccd955 100644
--- a/tests/dialects/test_redshift.py
+++ b/tests/dialects/test_redshift.py
@@ -332,10 +332,12 @@ class TestRedshift(Validator):
"CREATE TABLE sales (salesid INTEGER NOT NULL) DISTKEY(listid) COMPOUND SORTKEY(listid, sellerid) DISTSTYLE AUTO"
)
self.validate_identity(
- "COPY customer FROM 's3://mybucket/customer' IAM_ROLE 'arn:aws:iam::0123456789012:role/MyRedshiftRole'"
+ "COPY customer FROM 's3://mybucket/customer' IAM_ROLE 'arn:aws:iam::0123456789012:role/MyRedshiftRole'",
+ check_command_warning=True,
)
self.validate_identity(
- "UNLOAD ('select * from venue') TO 's3://mybucket/unload/' IAM_ROLE 'arn:aws:iam::0123456789012:role/MyRedshiftRole'"
+ "UNLOAD ('select * from venue') TO 's3://mybucket/unload/' IAM_ROLE 'arn:aws:iam::0123456789012:role/MyRedshiftRole'",
+ check_command_warning=True,
)
self.validate_identity(
"CREATE TABLE SOUP (SOUP1 VARCHAR(50) NOT NULL ENCODE ZSTD, SOUP2 VARCHAR(70) NULL ENCODE DELTA)"
diff --git a/tests/dialects/test_snowflake.py b/tests/dialects/test_snowflake.py
index 0882290..7e41fd4 100644
--- a/tests/dialects/test_snowflake.py
+++ b/tests/dialects/test_snowflake.py
@@ -10,6 +10,7 @@ class TestSnowflake(Validator):
dialect = "snowflake"
def test_snowflake(self):
+ self.validate_identity("ALTER TABLE table1 CLUSTER BY (name DESC)")
self.validate_identity(
"INSERT OVERWRITE TABLE t SELECT 1", "INSERT OVERWRITE INTO t SELECT 1"
)
@@ -39,8 +40,8 @@ WHERE
)""",
)
- self.validate_identity("RM @parquet_stage")
- self.validate_identity("REMOVE @parquet_stage")
+ self.validate_identity("RM @parquet_stage", check_command_warning=True)
+ self.validate_identity("REMOVE @parquet_stage", check_command_warning=True)
self.validate_identity("SELECT TIMESTAMP_FROM_PARTS(d, t)")
self.validate_identity("SELECT GET_PATH(v, 'attr[0].name') FROM vartab")
self.validate_identity("SELECT TO_ARRAY(CAST(x AS ARRAY))")
@@ -84,6 +85,10 @@ WHERE
"SELECT a FROM test PIVOT(SUM(x) FOR y IN ('z', 'q')) AS x TABLESAMPLE (0.1)"
)
self.validate_identity(
+ "SELECT p FROM t WHERE p:val NOT IN ('2')",
+ "SELECT p FROM t WHERE NOT GET_PATH(p, 'val') IN ('2')",
+ )
+ self.validate_identity(
"""SELECT PARSE_JSON('{"x": "hello"}'):x LIKE 'hello'""",
"""SELECT GET_PATH(PARSE_JSON('{"x": "hello"}'), 'x') LIKE 'hello'""",
)
@@ -777,9 +782,10 @@ WHERE
self.validate_identity("SELECT * FROM @namespace.mystage/path/to/file.json.gz")
self.validate_identity("SELECT * FROM @namespace.%table_name/path/to/file.json.gz")
self.validate_identity("SELECT * FROM '@external/location' (FILE_FORMAT => 'path.to.csv')")
- self.validate_identity("PUT file:///dir/tmp.csv @%table")
+ self.validate_identity("PUT file:///dir/tmp.csv @%table", check_command_warning=True)
self.validate_identity(
- 'COPY INTO NEW_TABLE ("foo", "bar") FROM (SELECT $1, $2, $3, $4 FROM @%old_table)'
+ 'COPY INTO NEW_TABLE ("foo", "bar") FROM (SELECT $1, $2, $3, $4 FROM @%old_table)',
+ check_command_warning=True,
)
self.validate_identity(
"SELECT * FROM @foo/bar (FILE_FORMAT => ds_sandbox.test.my_csv_format, PATTERN => 'test') AS bla"
@@ -1095,7 +1101,7 @@ WHERE
)
def test_stored_procedures(self):
- self.validate_identity("CALL a.b.c(x, y)")
+ self.validate_identity("CALL a.b.c(x, y)", check_command_warning=True)
self.validate_identity(
"CREATE PROCEDURE a.b.c(x INT, y VARIANT) RETURNS OBJECT EXECUTE AS CALLER AS 'BEGIN SELECT 1; END;'"
)
@@ -1449,10 +1455,10 @@ MATCH_RECOGNIZE (
def test_show(self):
# Parsed as Command
- self.validate_identity("SHOW TABLES LIKE 'line%' IN tpch.public")
-
- ast = parse_one("SHOW TABLES HISTORY IN tpch.public", read="snowflake")
- self.assertIsInstance(ast, exp.Command)
+ self.validate_identity(
+ "SHOW TABLES LIKE 'line%' IN tpch.public", check_command_warning=True
+ )
+ self.validate_identity("SHOW TABLES HISTORY IN tpch.public", check_command_warning=True)
# Parsed as Show
self.validate_identity("SHOW PRIMARY KEYS")
@@ -1469,6 +1475,18 @@ MATCH_RECOGNIZE (
'SHOW TERSE PRIMARY KEYS IN "TEST"."PUBLIC"."customers"',
'SHOW PRIMARY KEYS IN TABLE "TEST"."PUBLIC"."customers"',
)
+ self.validate_identity(
+ "show terse schemas in database db1 starts with 'a' limit 10 from 'b'",
+ "SHOW TERSE SCHEMAS IN DATABASE db1 STARTS WITH 'a' LIMIT 10 FROM 'b'",
+ )
+ self.validate_identity(
+ "show terse objects in schema db1.schema1 starts with 'a' limit 10 from 'b'",
+ "SHOW TERSE OBJECTS IN SCHEMA db1.schema1 STARTS WITH 'a' LIMIT 10 FROM 'b'",
+ )
+ self.validate_identity(
+ "show terse objects in db1.schema1 starts with 'a' limit 10 from 'b'",
+ "SHOW TERSE OBJECTS IN SCHEMA db1.schema1 STARTS WITH 'a' LIMIT 10 FROM 'b'",
+ )
ast = parse_one('SHOW PRIMARY KEYS IN "TEST"."PUBLIC"."customers"', read="snowflake")
table = ast.find(exp.Table)
@@ -1489,6 +1507,16 @@ MATCH_RECOGNIZE (
self.assertEqual(literal.sql(dialect="snowflake"), "'_testing%'")
+ ast = parse_one("SHOW SCHEMAS IN DATABASE db1", read="snowflake")
+ self.assertEqual(ast.args.get("scope_kind"), "DATABASE")
+ table = ast.find(exp.Table)
+ self.assertEqual(table.sql(dialect="snowflake"), "db1")
+
+ ast = parse_one("SHOW OBJECTS IN db1.schema1", read="snowflake")
+ self.assertEqual(ast.args.get("scope_kind"), "SCHEMA")
+ table = ast.find(exp.Table)
+ self.assertEqual(table.sql(dialect="snowflake"), "db1.schema1")
+
def test_swap(self):
ast = parse_one("ALTER TABLE a SWAP WITH b", read="snowflake")
assert isinstance(ast, exp.AlterTable)
diff --git a/tests/dialects/test_spark.py b/tests/dialects/test_spark.py
index 56a573a..6044037 100644
--- a/tests/dialects/test_spark.py
+++ b/tests/dialects/test_spark.py
@@ -227,7 +227,6 @@ TBLPROPERTIES (
)
def test_spark(self):
- self.validate_identity("FROM_UTC_TIMESTAMP(CAST(x AS TIMESTAMP), 'utc')")
expr = parse_one("any_value(col, true)", read="spark")
self.assertIsInstance(expr.args.get("ignore_nulls"), exp.Boolean)
self.assertEqual(expr.sql(dialect="spark"), "ANY_VALUE(col, TRUE)")
@@ -277,6 +276,25 @@ TBLPROPERTIES (
)
self.validate_all(
+ "SELECT TO_UTC_TIMESTAMP('2016-08-31', 'Asia/Seoul')",
+ write={
+ "bigquery": "SELECT DATETIME(TIMESTAMP(CAST('2016-08-31' AS DATETIME), 'Asia/Seoul'), 'UTC')",
+ "duckdb": "SELECT CAST('2016-08-31' AS TIMESTAMP) AT TIME ZONE 'Asia/Seoul' AT TIME ZONE 'UTC'",
+ "postgres": "SELECT CAST('2016-08-31' AS TIMESTAMP) AT TIME ZONE 'Asia/Seoul' AT TIME ZONE 'UTC'",
+ "presto": "SELECT WITH_TIMEZONE(CAST('2016-08-31' AS TIMESTAMP), 'Asia/Seoul') AT TIME ZONE 'UTC'",
+ "redshift": "SELECT CAST('2016-08-31' AS TIMESTAMP) AT TIME ZONE 'Asia/Seoul' AT TIME ZONE 'UTC'",
+ "snowflake": "SELECT CONVERT_TIMEZONE('Asia/Seoul', 'UTC', CAST('2016-08-31' AS TIMESTAMPNTZ))",
+ "spark": "SELECT TO_UTC_TIMESTAMP(CAST('2016-08-31' AS TIMESTAMP), 'Asia/Seoul')",
+ },
+ )
+ self.validate_all(
+ "SELECT FROM_UTC_TIMESTAMP('2016-08-31', 'Asia/Seoul')",
+ write={
+ "presto": "SELECT CAST('2016-08-31' AS TIMESTAMP) AT TIME ZONE 'Asia/Seoul'",
+ "spark": "SELECT FROM_UTC_TIMESTAMP(CAST('2016-08-31' AS TIMESTAMP), 'Asia/Seoul')",
+ },
+ )
+ self.validate_all(
"foo.bar",
read={
"": "STRUCT_EXTRACT(foo, bar)",
diff --git a/tests/dialects/test_teradata.py b/tests/dialects/test_teradata.py
index 85d4ebf..f3894fd 100644
--- a/tests/dialects/test_teradata.py
+++ b/tests/dialects/test_teradata.py
@@ -42,11 +42,13 @@ class TestTeradata(Validator):
)
def test_statistics(self):
- self.validate_identity("COLLECT STATISTICS ON tbl INDEX(col)")
- self.validate_identity("COLLECT STATS ON tbl COLUMNS(col)")
- self.validate_identity("COLLECT STATS COLUMNS(col) ON tbl")
- self.validate_identity("HELP STATISTICS personel.employee")
- self.validate_identity("HELP STATISTICS personnel.employee FROM my_qcd")
+ self.validate_identity("COLLECT STATISTICS ON tbl INDEX(col)", check_command_warning=True)
+ self.validate_identity("COLLECT STATS ON tbl COLUMNS(col)", check_command_warning=True)
+ self.validate_identity("COLLECT STATS COLUMNS(col) ON tbl", check_command_warning=True)
+ self.validate_identity("HELP STATISTICS personel.employee", check_command_warning=True)
+ self.validate_identity(
+ "HELP STATISTICS personnel.employee FROM my_qcd", check_command_warning=True
+ )
def test_create(self):
self.validate_identity(
diff --git a/tests/dialects/test_tsql.py b/tests/dialects/test_tsql.py
index 7cf9971..101d356 100644
--- a/tests/dialects/test_tsql.py
+++ b/tests/dialects/test_tsql.py
@@ -1,4 +1,5 @@
from sqlglot import exp, parse, parse_one
+from sqlglot.parser import logger as parser_logger
from tests.dialects.test_dialect import Validator
@@ -7,7 +8,7 @@ class TestTSQL(Validator):
def test_tsql(self):
self.validate_identity("ROUND(x, 1, 0)")
- self.validate_identity("EXEC MyProc @id=7, @name='Lochristi'")
+ self.validate_identity("EXEC MyProc @id=7, @name='Lochristi'", check_command_warning=True)
# https://learn.microsoft.com/en-us/previous-versions/sql/sql-server-2008-r2/ms187879(v=sql.105)?redirectedfrom=MSDN
# tsql allows .. which means use the default schema
self.validate_identity("SELECT * FROM a..b")
@@ -225,7 +226,7 @@ class TestTSQL(Validator):
"MERGE INTO mytable WITH (HOLDLOCK) AS T USING mytable_merge AS S "
"ON (T.user_id = S.user_id) WHEN NOT MATCHED THEN INSERT (c1, c2) VALUES (S.c1, S.c2)"
)
- self.validate_identity("UPDATE STATISTICS x")
+ self.validate_identity("UPDATE STATISTICS x", check_command_warning=True)
self.validate_identity("UPDATE x SET y = 1 OUTPUT x.a, x.b INTO @y FROM y")
self.validate_identity("UPDATE x SET y = 1 OUTPUT x.a, x.b FROM y")
self.validate_identity("INSERT INTO x (y) OUTPUT x.a, x.b INTO l SELECT * FROM z")
@@ -238,14 +239,16 @@ class TestTSQL(Validator):
self.validate_identity("END")
self.validate_identity("@x")
self.validate_identity("#x")
- self.validate_identity("DECLARE @TestVariable AS VARCHAR(100)='Save Our Planet'")
- self.validate_identity("PRINT @TestVariable")
+ self.validate_identity("PRINT @TestVariable", check_command_warning=True)
self.validate_identity("SELECT Employee_ID, Department_ID FROM @MyTableVar")
self.validate_identity("INSERT INTO @TestTable VALUES (1, 'Value1', 12, 20)")
self.validate_identity("SELECT * FROM #foo")
self.validate_identity("SELECT * FROM ##foo")
self.validate_identity("SELECT a = 1", "SELECT 1 AS a")
self.validate_identity(
+ "DECLARE @TestVariable AS VARCHAR(100)='Save Our Planet'", check_command_warning=True
+ )
+ self.validate_identity(
"SELECT a = 1 UNION ALL SELECT a = b", "SELECT 1 AS a UNION ALL SELECT b AS a"
)
self.validate_identity(
@@ -789,7 +792,8 @@ class TestTSQL(Validator):
def test_udf(self):
self.validate_identity(
- "DECLARE @DWH_DateCreated DATETIME = CONVERT(DATETIME, getdate(), 104)"
+ "DECLARE @DWH_DateCreated DATETIME = CONVERT(DATETIME, getdate(), 104)",
+ check_command_warning=True,
)
self.validate_identity(
"CREATE PROCEDURE foo @a INTEGER, @b INTEGER AS SELECT @a = SUM(bla) FROM baz AS bar"
@@ -882,8 +886,9 @@ WHERE
"END",
]
- for expr, expected_sql in zip(parse(sql, read="tsql"), expected_sqls):
- self.assertEqual(expr.sql(dialect="tsql"), expected_sql)
+ with self.assertLogs(parser_logger) as cm:
+ for expr, expected_sql in zip(parse(sql, read="tsql"), expected_sqls):
+ self.assertEqual(expr.sql(dialect="tsql"), expected_sql)
sql = """
CREATE PROC [dbo].[transform_proc] AS
@@ -902,8 +907,9 @@ WHERE
"CREATE TABLE [target_schema].[target_table] (a INTEGER) WITH (DISTRIBUTION=REPLICATE, HEAP)",
]
- for expr, expected_sql in zip(parse(sql, read="tsql"), expected_sqls):
- self.assertEqual(expr.sql(dialect="tsql"), expected_sql)
+ with self.assertLogs(parser_logger) as cm:
+ for expr, expected_sql in zip(parse(sql, read="tsql"), expected_sqls):
+ self.assertEqual(expr.sql(dialect="tsql"), expected_sql)
def test_charindex(self):
self.validate_all(
@@ -932,7 +938,11 @@ WHERE
)
def test_len(self):
- self.validate_all("LEN(x)", read={"": "LENGTH(x)"}, write={"spark": "LENGTH(x)"})
+ self.validate_all(
+ "LEN(x)", read={"": "LENGTH(x)"}, write={"spark": "LENGTH(CAST(x AS STRING))"}
+ )
+ self.validate_all("LEN(1)", write={"tsql": "LEN(1)", "spark": "LENGTH(CAST(1 AS STRING))"})
+ self.validate_all("LEN('x')", write={"tsql": "LEN('x')", "spark": "LENGTH('x')"})
def test_replicate(self):
self.validate_all("REPLICATE('x', 2)", write={"spark": "REPEAT('x', 2)"})