Merge pull request #66725 from ClickHouse/vdimir/join_on_null_fix_followup

Support one more case in JOIN ON ... IS NULL

src/Analyzer/Passes/LogicalExpressionOptimizerPass.cpp

@@ -68,6 +68,41 @@ QueryTreeNodePtr findEqualsFunction(const QueryTreeNodes & nodes)
     return nullptr;
 }
 
+/// Checks if the node is combination of isNull and notEquals functions of two the same arguments
+bool matchIsNullOfTwoArgs(const QueryTreeNodes & nodes, QueryTreeNodePtr & lhs, QueryTreeNodePtr & rhs)
+{
+    QueryTreeNodePtrWithHashSet all_arguments;
+    for (const auto & node : nodes)
+    {
+        const auto * func_node = node->as<FunctionNode>();
+        if (!func_node)
+            return false;
+
+        const auto & arguments = func_node->getArguments().getNodes();
+        if (func_node->getFunctionName() == "isNull" && arguments.size() == 1)
+            all_arguments.insert(QueryTreeNodePtrWithHash(arguments[0]));
+        else if (func_node->getFunctionName() == "notEquals" && arguments.size() == 2)
+        {
+            if (arguments[0]->isEqual(*arguments[1]))
+                return false;
+            all_arguments.insert(QueryTreeNodePtrWithHash(arguments[0]));
+            all_arguments.insert(QueryTreeNodePtrWithHash(arguments[1]));
+        }
+        else
+            return false;
+
+        if (all_arguments.size() > 2)
+            return false;
+    }
+
+    if (all_arguments.size() != 2)
+        return false;
+
+    lhs = all_arguments.begin()->node;
+    rhs = std::next(all_arguments.begin())->node;
+    return true;
+}
+
 bool isBooleanConstant(const QueryTreeNodePtr & node, bool expected_value)
 {
     const auto * constant_node = node->as<ConstantNode>();
@@ -213,11 +248,14 @@ private:
             else if (func_name == "and")
             {
                 const auto & and_arguments = argument_function->getArguments().getNodes();
-                bool all_are_is_null = and_arguments.size() == 2 && isNodeFunction(and_arguments[0], "isNull") && isNodeFunction(and_arguments[1], "isNull");
+
-                if (all_are_is_null)
+                QueryTreeNodePtr is_null_lhs_arg;
+                QueryTreeNodePtr is_null_rhs_arg;
+                if (matchIsNullOfTwoArgs(and_arguments, is_null_lhs_arg, is_null_rhs_arg))
                 {
-                    is_null_argument_to_indices[getFunctionArgument(and_arguments.front(), 0)].push_back(or_operands.size() - 1);
+                    is_null_argument_to_indices[is_null_lhs_arg].push_back(or_operands.size() - 1);
-                    is_null_argument_to_indices[getFunctionArgument(and_arguments.back(), 0)].push_back(or_operands.size() - 1);
+                    is_null_argument_to_indices[is_null_rhs_arg].push_back(or_operands.size() - 1);
+                    continue;
                 }
 
                 /// Expression `a = b AND (a IS NOT NULL) AND true AND (b IS NOT NULL)` we can be replaced with `a = b`

tests/queries/0_stateless/02911_join_on_nullsafe_optimization.reference

@@ -35,6 +35,10 @@ SELECT * FROM t1 JOIN t2 ON (t1.x == t2.x AND ((t2.x IS NOT NULL) AND (t1.x IS N
 2	2	2	2
 3	3	3	33
 \N	\N	\N	\N
+SELECT * FROM t1 JOIN t2 ON (t1.x == t2.x AND ((t2.x IS NOT NULL) AND (t1.x IS NOT NULL)) ) OR ( t2.x <> t1.x AND (t2.x IS NULL) AND (t1.x IS NULL) ) ORDER BY t1.x NULLS LAST;
+2	2	2	2
+3	3	3	33
+\N	\N	\N	\N
 -- aliases defined in the join condition are valid
 -- FIXME(@vdimir) broken query formatting for the following queries:
 -- SELECT *, e, e2 FROM t1 FULL JOIN t2 ON ( ( ((t1.x == t2.x) AS e) AND ((t2.x IS NOT NULL) AND (t1.x IS NOT NULL)) ) OR ( (t2.x IS NULL) AND (t1.x IS NULL) ) AS e2 ) ORDER BY t1.x NULLS LAST, t2.x NULLS LAST;

tests/queries/0_stateless/02911_join_on_nullsafe_optimization.sql

@@ -34,6 +34,7 @@ SELECT * FROM t1 JOIN t2 ON (t1.x <=> t2.x OR (t1.x IS NULL AND t2.x IS NULL)) A
 SELECT x = y OR (x IS NULL AND y IS NULL) FROM t1 ORDER BY x NULLS LAST;
 
 SELECT * FROM t1 JOIN t2 ON (t1.x == t2.x AND ((t2.x IS NOT NULL) AND (t1.x IS NOT NULL)) ) OR ( (t2.x IS NULL) AND (t1.x IS NULL) ) ORDER BY t1.x NULLS LAST;
+SELECT * FROM t1 JOIN t2 ON (t1.x == t2.x AND ((t2.x IS NOT NULL) AND (t1.x IS NOT NULL)) ) OR ( t2.x <> t1.x AND (t2.x IS NULL) AND (t1.x IS NULL) ) ORDER BY t1.x NULLS LAST;
 
 -- aliases defined in the join condition are valid
 -- FIXME(@vdimir) broken query formatting for the following queries: