From b76779797af4a7684a973ed9b3b4b3c4acb27ba0 Mon Sep 17 00:00:00 2001
From: vdimir <vdimir@clickhouse.com>
Date: Wed, 18 Jan 2023 12:17:30 +0000
Subject: [PATCH] Do not move to prewhere in select with joins

---
 src/Interpreters/InterpreterSelectQuery.cpp   |  5 ++-
 .../02534_join_prewhere_bug_44062.reference   | 38 ++++++++++++++++
 .../02534_join_prewhere_bug_44062.sql         | 43 +++++++++++++++++++
 3 files changed, 85 insertions(+), 1 deletion(-)
 create mode 100644 tests/queries/0_stateless/02534_join_prewhere_bug_44062.reference
 create mode 100644 tests/queries/0_stateless/02534_join_prewhere_bug_44062.sql

diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp
index 48326afda45..6477188ec59 100644
--- a/src/Interpreters/InterpreterSelectQuery.cpp
+++ b/src/Interpreters/InterpreterSelectQuery.cpp
@@ -560,7 +560,10 @@ InterpreterSelectQuery::InterpreterSelectQuery(
             view = nullptr;
         }
 
-        if (try_move_to_prewhere && storage && storage->canMoveConditionsToPrewhere() && query.where() && !query.prewhere())
+        if (try_move_to_prewhere
+            && storage && storage->canMoveConditionsToPrewhere()
+            && query.where() && !query.prewhere()
+            && !query.hasJoin()) /// Join may produce rows with nulls or default values, it's difficult to analyze if they affected or not.
         {
             /// PREWHERE optimization: transfer some condition from WHERE to PREWHERE if enabled and viable
             if (const auto & column_sizes = storage->getColumnSizes(); !column_sizes.empty())
diff --git a/tests/queries/0_stateless/02534_join_prewhere_bug_44062.reference b/tests/queries/0_stateless/02534_join_prewhere_bug_44062.reference
new file mode 100644
index 00000000000..aaef17371d8
--- /dev/null
+++ b/tests/queries/0_stateless/02534_join_prewhere_bug_44062.reference
@@ -0,0 +1,38 @@
+-- { echoOn }
+
+SELECT * FROM test1 LEFT JOIN test2 ON test1.col1 = test2.col1
+WHERE test2.col1 IS NULL
+ORDER BY test2.col1
+;
+12321	-30	\N	\N
+SELECT * FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NULL
+ORDER BY test2.col1
+;
+\N	\N	12321	-30
+SELECT * FROM test1 LEFT JOIN test2 ON test1.col1 = test2.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+123	123	123	5600
+321	-32	321	5601
+SELECT * FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+123	5600	123	123
+321	5601	321	-32
+SELECT test2.col1, test1.* FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+123	123	123
+321	321	-32
+SELECT test2.col3, test1.* FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+5600	123	123
+5601	321	-32
+DROP TABLE IF EXISTS test1;
+DROP TABLE IF EXISTS test2;
diff --git a/tests/queries/0_stateless/02534_join_prewhere_bug_44062.sql b/tests/queries/0_stateless/02534_join_prewhere_bug_44062.sql
new file mode 100644
index 00000000000..d68800b0f75
--- /dev/null
+++ b/tests/queries/0_stateless/02534_join_prewhere_bug_44062.sql
@@ -0,0 +1,43 @@
+
+CREATE OR REPLACE TABLE test1 ( `col1` UInt64, `col2` Int8 ) ENGINE = MergeTree ORDER BY col1;
+CREATE OR REPLACE TABLE test2 ( `col1` UInt64, `col3` Int16 ) ENGINE = MergeTree ORDER BY col1;
+
+INSERT INTO test1 VALUES (123, 123), (12321, -30), (321, -32);
+INSERT INTO test2 VALUES (123, 5600), (321, 5601);
+
+SET join_use_nulls = 1;
+
+-- { echoOn }
+
+SELECT * FROM test1 LEFT JOIN test2 ON test1.col1 = test2.col1
+WHERE test2.col1 IS NULL
+ORDER BY test2.col1
+;
+
+SELECT * FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NULL
+ORDER BY test2.col1
+;
+
+SELECT * FROM test1 LEFT JOIN test2 ON test1.col1 = test2.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+
+SELECT * FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+
+SELECT test2.col1, test1.* FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+
+SELECT test2.col3, test1.* FROM test2 RIGHT JOIN test1 ON test2.col1 = test1.col1
+WHERE test2.col1 IS NOT NULL
+ORDER BY test2.col1
+;
+
+DROP TABLE IF EXISTS test1;
+DROP TABLE IF EXISTS test2;