Merge pull request #2137 from mabel-dev/#2129

#2129

opteryx/__version__.py

@@ -1,4 +1,4 @@
-__build__ = 905
+__build__ = 906
 
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

opteryx/exceptions.py

@@ -248,9 +248,10 @@ def __init__(self, dataset: str):
 class UnexpectedDatasetReferenceError(SqlError):
     """Exception raised for unexpected dataset references."""
 
-    def __init__(self, dataset: str):
+    def __init__(self, dataset: str, message: Optional[str] = None):
         self.dataset = dataset
-        message = f"Dataset '{dataset}' is referenced in query but it doesn't appear in a FROM or JOIN clause."
+        if not message:
+            message = f"Dataset '{dataset}' is referenced in query but it doesn't appear in a FROM or JOIN clause."
         super().__init__(message)
 
 

opteryx/operators/outer_join_node.py

@@ -23,11 +23,13 @@
 popular SEMI and ANTI joins we leave to PyArrow for now.
 """
 
+from collections import deque
 from typing import List
 
 import pyarrow
 
 from opteryx import EOS
+from opteryx.compiled.structures import HashSet
 from opteryx.compiled.structures import HashTable
 from opteryx.models import QueryProperties
 from opteryx.utils.arrow import align_tables
@@ -52,8 +54,6 @@ def left_join(left_relation, right_relation, left_columns: List[str], right_colu
     Returns:
         A pyarrow.Table containing the result of the LEFT JOIN operation.
     """
-    from collections import deque
-
     from opteryx.compiled.structures.hash_table import hash_join_map
 
     left_indexes: deque = deque()
@@ -190,16 +190,15 @@ def left_anti_join(
     Returns:
         A pyarrow.Table containing the result of the LEFT ANTI JOIN operation.
     """
-    hash_table = HashTable()
     non_null_right_values = right_relation.select(right_columns).itercolumns()
-    for i, value_tuple in enumerate(zip(*non_null_right_values)):
-        hash_table.insert(hash(value_tuple), i)
+    right_hash_set = set(zip(*non_null_right_values))
 
     left_indexes = []
     left_values = left_relation.select(left_columns).itercolumns()
     for i, value_tuple in enumerate(zip(*left_values)):
-        rows = hash_table.get(hash(value_tuple))
-        if not rows:  # Only include left rows that have no match in the right table
+        if (
+            value_tuple not in right_hash_set
+        ):  # Only include left rows that have no match in the right table
             left_indexes.append(i)
 
     # Filter the left_chunk based on the anti join condition

opteryx/planner/binder/binder.py

@@ -138,7 +138,13 @@ def create_variable_node(node: Node, context: BindingContext) -> Node:
 
     # if there are no candidates, we probably don't know the relation
     if not candidate_schemas:
-        raise UnexpectedDatasetReferenceError(dataset=node.source)
+        if node.source in context.relations:
+            raise UnexpectedDatasetReferenceError(
+                dataset=node.source,
+                message=f"Dataset `{node.source}` is not available after being used on the right side of a ANTI or SEMI JOIN",
+            )
+        else:
+            raise UnexpectedDatasetReferenceError(dataset=node.source)
 
     # look up the column in the candidate schemas
     column, found_source_relation = locate_identifier_in_loaded_schemas(

tests/sql_battery/test_shapes_and_errors_battery.py

@@ -1535,8 +1535,8 @@
         ("SELECT * FROM $planets AS P LEFT ANTI JOIN $satellites AS S ON S.id = P.id WHERE P.id > 5;", 0, 20, None),
         ("SELECT * FROM $planets AS P LEFT ANTI JOIN (SELECT id FROM $satellites WHERE name LIKE 'Moon%') AS S ON S.id = P.id;", 8, 20, None),
         ("SELECT * FROM GENERATE_SERIES(1, 10) AS C LEFT ANTI JOIN $satellites AS S ON S.id = C;", 0, 1, None),
-#        ("SELECT * FROM $planets AS P LEFT ANTI JOIN $satellites AS S ON S.id = P.id WHERE S.size > 1000;", 0, 20, None),
-#        ("SELECT * FROM $planets AS P LEFT SEMI JOIN $satellites AS S ON S.id = P.id WHERE S.name LIKE 'Moon%';", 0, 20, None),
+        ("SELECT * FROM $planets AS P LEFT SEMI JOIN $satellites AS S ON S.id = P.id WHERE P.name LIKE 'E%';", 1, 20, None),
+        ("SELECT * FROM $planets AS P LEFT SEMI JOIN $satellites AS S ON S.id = P.id WHERE S.name LIKE 'E%';", 1, 20, UnexpectedDatasetReferenceError),
         ("SELECT * FROM $planets AS P LEFT SEMI JOIN (SELECT id FROM $satellites WHERE name != 'Moon') AS S ON S.id = P.id;", 8, 20, None),
         ("SELECT * FROM $planets AS P LEFT SEMI JOIN $satellites AS S ON S.id = P.id WHERE P.name != 'Earth';", 8, 20, None),
         ("SELECT * FROM GENERATE_SERIES(1, 10) AS G LEFT SEMI JOIN $satellites AS S ON S.id = G;", 10, 1, None),