Fix!: don't parse SEMI, ANTI as table aliases, fix join side issue (#…

…2247)

sqlglot/dialects/bigquery.py

@@ -176,6 +176,7 @@ def _parse_to_hex(args: t.List) -> exp.Hex | exp.MD5:
 class BigQuery(Dialect):
     UNNEST_COLUMN_ONLY = True
     SUPPORTS_USER_DEFINED_TYPES = False
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     # https://cloud.google.com/bigquery/docs/reference/standard-sql/lexical#case_sensitivity
     RESOLVES_IDENTIFIERS_AS_UPPERCASE = None

sqlglot/dialects/clickhouse.py

@@ -113,15 +113,11 @@ class Parser(parser.Parser):
             *parser.Parser.JOIN_KINDS,
             TokenType.ANY,
             TokenType.ASOF,
-            TokenType.ANTI,
-            TokenType.SEMI,
             TokenType.ARRAY,
         }
 
         TABLE_ALIAS_TOKENS = parser.Parser.TABLE_ALIAS_TOKENS - {
             TokenType.ANY,
-            TokenType.SEMI,
-            TokenType.ANTI,
             TokenType.SETTINGS,
             TokenType.FORMAT,
             TokenType.ARRAY,

sqlglot/dialects/dialect.py

@@ -125,6 +125,12 @@ def get_start_end(token_type: TokenType) -> t.Tuple[t.Optional[str], t.Optional[
         if not klass.STRICT_STRING_CONCAT and klass.DPIPE_IS_STRING_CONCAT:
             klass.parser_class.BITWISE[TokenType.DPIPE] = exp.SafeDPipe
 
+        if not klass.SUPPORTS_SEMI_ANTI_JOIN:
+            klass.parser_class.TABLE_ALIAS_TOKENS = klass.parser_class.TABLE_ALIAS_TOKENS | {
+                TokenType.ANTI,
+                TokenType.SEMI,
+            }
+
         klass.generator_class.can_identify = klass.can_identify
 
         return klass
@@ -156,6 +162,9 @@ class Dialect(metaclass=_Dialect):
     # Determines whether or not user-defined data types are supported
     SUPPORTS_USER_DEFINED_TYPES = True
 
+    # Determines whether or not SEMI/ANTI JOINs are supported
+    SUPPORTS_SEMI_ANTI_JOIN = True
+
     # Determines how function names are going to be normalized
     NORMALIZE_FUNCTIONS: bool | str = "upper"
 

sqlglot/dialects/drill.py

@@ -40,6 +40,7 @@ class Drill(Dialect):
     DATEINT_FORMAT = "'yyyyMMdd'"
     TIME_FORMAT = "'yyyy-MM-dd HH:mm:ss'"
     SUPPORTS_USER_DEFINED_TYPES = False
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     TIME_MAPPING = {
         "y": "%Y",

sqlglot/dialects/duckdb.py

@@ -230,6 +230,7 @@ class Generator(generator.Generator):
         STRUCT_DELIMITER = ("(", ")")
         RENAME_TABLE_WITH_DB = False
         NVL2_SUPPORTED = False
+        SEMI_ANTI_JOIN_WITH_SIDE = False
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,

sqlglot/dialects/mysql.py

@@ -102,6 +102,7 @@ class MySQL(Dialect):
     TIME_FORMAT = "'%Y-%m-%d %T'"
     DPIPE_IS_STRING_CONCAT = False
     SUPPORTS_USER_DEFINED_TYPES = False
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     # https://prestodb.io/docs/current/functions/datetime.html#mysql-date-functions
     TIME_MAPPING = {

sqlglot/dialects/presto.py

@@ -174,6 +174,7 @@ class Presto(Dialect):
     TIME_FORMAT = MySQL.TIME_FORMAT
     TIME_MAPPING = MySQL.TIME_MAPPING
     STRICT_STRING_CONCAT = True
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     # https://github.com/trinodb/trino/issues/17
     # https://github.com/trinodb/trino/issues/12289

sqlglot/dialects/snowflake.py

@@ -204,6 +204,7 @@ class Snowflake(Dialect):
     NULL_ORDERING = "nulls_are_large"
     TIME_FORMAT = "'YYYY-MM-DD HH24:MI:SS'"
     SUPPORTS_USER_DEFINED_TYPES = False
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     TIME_MAPPING = {
         "YYYY": "%Y",

sqlglot/dialects/sqlite.py

@@ -64,6 +64,7 @@ def _transform_create(expression: exp.Expression) -> exp.Expression:
 class SQLite(Dialect):
     # https://sqlite.org/forum/forumpost/5e575586ac5c711b?raw
     RESOLVES_IDENTIFIERS_AS_UPPERCASE = None
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     class Tokenizer(tokens.Tokenizer):
         IDENTIFIERS = ['"', ("[", "]"), "`"]

sqlglot/dialects/teradata.py

@@ -8,6 +8,8 @@
 
 
 class Teradata(Dialect):
+    SUPPORTS_SEMI_ANTI_JOIN = False
+
     TIME_MAPPING = {
         "Y": "%Y",
         "YYYY": "%Y",

sqlglot/dialects/tsql.py

@@ -206,6 +206,7 @@ class TSQL(Dialect):
     RESOLVES_IDENTIFIERS_AS_UPPERCASE = None
     NULL_ORDERING = "nulls_are_small"
     TIME_FORMAT = "'yyyy-mm-dd hh:mm:ss'"
+    SUPPORTS_SEMI_ANTI_JOIN = False
 
     TIME_MAPPING = {
         "year": "%Y",

sqlglot/generator.py

@@ -194,6 +194,9 @@ class Generator:
     # Whether or not FILTER (WHERE cond) can be used for conditional aggregation
     AGGREGATE_FILTER_SUPPORTED = True
 
+    # Whether or not JOIN sides (LEFT, RIGHT) are supported in conjunction with SEMI/ANTI join kinds
+    SEMI_ANTI_JOIN_WITH_SIDE = True
+
     TYPE_MAPPING = {
         exp.DataType.Type.NCHAR: "CHAR",
         exp.DataType.Type.NVARCHAR: "VARCHAR",
@@ -1491,12 +1494,17 @@ def prior_sql(self, expression: exp.Prior) -> str:
         return f"PRIOR {self.sql(expression, 'this')}"
 
     def join_sql(self, expression: exp.Join) -> str:
+        if not self.SEMI_ANTI_JOIN_WITH_SIDE and expression.kind in ("SEMI", "ANTI"):
+            side = None
+        else:
+            side = expression.side
+
         op_sql = " ".join(
             op
             for op in (
                 expression.method,
                 "GLOBAL" if expression.args.get("global") else None,
-                expression.side,
+                side,
                 expression.kind,
                 expression.hint if self.JOIN_HINTS else None,
             )

sqlglot/parser.py

@@ -316,6 +316,7 @@ class Parser(metaclass=_Parser):
     INTERVAL_VARS = ID_VAR_TOKENS - {TokenType.END}
 
     TABLE_ALIAS_TOKENS = ID_VAR_TOKENS - {
+        TokenType.ANTI,
         TokenType.APPLY,
         TokenType.ASOF,
         TokenType.FULL,
@@ -324,6 +325,7 @@ class Parser(metaclass=_Parser):
         TokenType.NATURAL,
         TokenType.OFFSET,
         TokenType.RIGHT,
+        TokenType.SEMI,
         TokenType.WINDOW,
     }
 

tests/dialects/test_duckdb.py

@@ -8,6 +8,7 @@ class TestDuckDB(Validator):
     def test_duckdb(self):
         for join_type in ("SEMI", "ANTI"):
             exists = "EXISTS" if join_type == "SEMI" else "NOT EXISTS"
+
             self.validate_all(
                 f"SELECT * FROM t1 {join_type} JOIN t2 ON t1.x = t2.x",
                 write={
@@ -32,6 +33,13 @@ def test_duckdb(self):
                     "tsql": f"SELECT * FROM t1 WHERE {exists}(SELECT 1 FROM t2 WHERE t1.x = t2.x)",
                 },
             )
+            self.validate_all(
+                f"SELECT * FROM t1 {join_type} JOIN t2 ON t1.x = t2.x",
+                read={
+                    "duckdb": f"SELECT * FROM t1 {join_type} JOIN t2 ON t1.x = t2.x",
+                    "spark": f"SELECT * FROM t1 LEFT {join_type} JOIN t2 ON t1.x = t2.x",
+                },
+            )
 
         self.validate_all(
             "WITH cte(x) AS (SELECT 1 UNION ALL SELECT 2 UNION ALL SELECT 3) SELECT AVG(x) FILTER (WHERE x > 1) FROM cte",

tests/dialects/test_spark.py

@@ -229,6 +229,7 @@ def test_spark(self):
         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")

tests/fixtures/identity.sql

@@ -401,6 +401,7 @@ SELECT 1 FROM a INNER JOIN b ON a.x = b.x
 SELECT 1 FROM a LEFT JOIN b ON a.x = b.x
 SELECT 1 FROM a RIGHT JOIN b ON a.x = b.x
 SELECT 1 FROM a CROSS JOIN b ON a.x = b.x
+SELECT 1 FROM a SEMI JOIN b ON a.x = b.x
 SELECT 1 FROM a LEFT SEMI JOIN b ON a.x = b.x
 SELECT 1 FROM a LEFT ANTI JOIN b ON a.x = b.x
 SELECT 1 FROM a RIGHT SEMI JOIN b ON a.x = b.x