#2149

opteryx/operators/cross_join_node.py

@@ -271,6 +271,8 @@ class CrossJoinNode(JoinNode):
     Implements a SQL CROSS JOIN
     """
 
+    join_type = "cross"
+
     def __init__(self, properties: QueryProperties, **parameters):
         JoinNode.__init__(self, properties=properties, **parameters)
 

opteryx/operators/filter_join_node.py

@@ -141,7 +141,9 @@ def execute(self, morsel: pyarrow.Table, join_leg: str) -> pyarrow.Table:
             if morsel == EOS:
                 right_relation = pyarrow.concat_tables(self.right_buffer, promote_options="none")
                 self.right_buffer.clear()
-                non_null_right_values = right_relation.select(self.right_columns).drop_null().itercolumns()
+                non_null_right_values = (
+                    right_relation.select(self.right_columns).drop_null().itercolumns()
+                )
                 self.right_hash_set = set(map(hash, zip(*non_null_right_values)))
             else:
                 self.right_buffer.append(morsel)

opteryx/operators/inner_join_node.py

@@ -76,7 +76,6 @@ def inner_join_with_preprocessed_left_side(left_relation, right_relation, join_c
 
 
 class InnerJoinNode(JoinNode):
-
     join_type = "inner"
 
     def __init__(self, properties: QueryProperties, **parameters):

opteryx/operators/inner_join_node_single.py

@@ -159,7 +159,6 @@ def inner_join_with_preprocessed_left_side(left_relation, right_relation, join_c
 
 
 class InnerJoinSingleNode(JoinNode):
-
     join_type = "inner"
 
     def __init__(self, properties: QueryProperties, **parameters):

opteryx/operators/outer_join_node.py

@@ -207,7 +207,6 @@ def config(self) -> str:  # pragma: no cover
         return f"{self.join_type.upper()}"
 
     def execute(self, morsel: pyarrow.Table, join_leg: str) -> pyarrow.Table:
-        print("OuterJoinNode.execute", join_leg, type(morsel))
         if join_leg == "left":
             if morsel == EOS:
                 self.left_relation = pyarrow.concat_tables(self.left_buffer, promote_options="none")

opteryx/utils/formatter.py

@@ -78,6 +78,7 @@ def _replacer(match):
             "OUTER",
             "RIGHT",
             "SELECT",
+            "SEMI",
             "SET",
             "SHOW",
             "SINCE",

opteryx/utils/sql.py

@@ -121,7 +121,7 @@ def split_sql_statements(sql: str) -> List[str]:
 def regex_match_any(
     arr: numpy.ndarray,
     patterns: List[str],
-    flags: int = re.NOFLAG,
+    flags: int = 0,
     invert: bool = False,
 ) -> numpy.ndarray:
     """

tests/fuzzing/test_sql_fuzzer_join.py

@@ -51,7 +51,7 @@ def generate_condition(table, columns):
     return f"{table}.{where_column.name} {where_operator} {where_value}"
 
 def generate_random_sql_join(columns1, table1, columns2, table2) -> str:
-    join_type = random.choice(["JOIN", "INNER JOIN", "LEFT JOIN", "RIGHT JOIN", "FULL OUTER JOIN"])
+    join_type = random.choice(["JOIN", "INNER JOIN", "LEFT JOIN", "LEFT OUTER JOIN", "RIGHT JOIN", "FULL OUTER JOIN", "LEFT ANTI JOIN", "LEFT SEMI JOIN"])
 
     last_value = -1
     this_value = random.random()
@@ -70,7 +70,11 @@ def generate_random_sql_join(columns1, table1, columns2, table2) -> str:
         conditions.append(condition)
 
     join_condition = " AND ".join(conditions)
-    selected_columns = [f"{table1}.{col.name}" for col in columns1 if random.random() < 0.2] + [f"{table2}.{col.name}" for col in columns2 if random.random() < 0.2]
+
+    if join_type in ("LEFT ANTI JOIN", "LEFT SEMI JOIN"):
+        selected_columns = [f"{table1}.{col.name}" for col in columns1 if random.random() < 0.2]
+    else:
+        selected_columns = [f"{table1}.{col.name}" for col in columns1 if random.random() < 0.2] + [f"{table2}.{col.name}" for col in columns2 if random.random() < 0.2]
     if len(selected_columns) == 0:
         selected_columns = ["*"]
     select_clause = "SELECT " + ", ".join(selected_columns)
@@ -86,7 +90,7 @@ def generate_random_sql_join(columns1, table1, columns2, table2) -> str:
             linking_condition = random.choice(["AND", "OR", "AND NOT"])
             where_clause += f" {linking_condition} {generate_condition(table1, columns1)}"
 
-    if random.random() < 0.3:
+    if join_type not in ("LEFT ANTI JOIN", "LEFT SEMI JOIN") and random.random() < 0.3:
         if where_clause == "--":
             where_clause = " WHERE "
         else: