[fix](inverted index) Incorrect handling of NULL literal

be/src/olap/rowset/segment_v2/segment_iterator.cpp

@@ -2849,6 +2849,11 @@ void SegmentIterator::_calculate_pred_in_remaining_conjunct_root(
     } else if (_is_literal_node(node_type)) {
         auto v_literal_expr = static_cast<const doris::vectorized::VLiteral*>(expr.get());
         _column_predicate_info->query_values.insert(v_literal_expr->value());
+    } else if (node_type == TExprNodeType::NULL_LITERAL) {
+        if (!_column_predicate_info->column_name.empty()) {
+            auto v_literal_expr = static_cast<const doris::vectorized::VLiteral*>(expr.get());
+            _column_predicate_info->query_values.insert(v_literal_expr->value());
+        }
     } else {
         if (node_type == TExprNodeType::MATCH_PRED) {
             _column_predicate_info->query_op = "match";

be/src/pipeline/exec/scan_operator.cpp

@@ -1100,18 +1100,12 @@ Status ScanLocalState<Derived>::_normalize_in_and_not_in_compound_predicate(
         auto hybrid_set = expr->get_set_func();
 
         if (hybrid_set != nullptr) {
-            if (hybrid_set->size() <=
-                _parent->cast<typename Derived::Parent>()._max_pushdown_conditions_per_column) {
-                iter = hybrid_set->begin();
-            } else {
-                _filter_predicates.in_filters.emplace_back(slot->col_name(), expr->get_set_func());
-                *pdt = PushDownType::ACCEPTABLE;
-                return Status::OK();
-            }
+            *pdt = PushDownType::UNACCEPTABLE;
+            return Status::OK();
         } else {
-            vectorized::VInPredicate* pred = static_cast<vectorized::VInPredicate*>(expr);
+            auto* pred = static_cast<vectorized::VInPredicate*>(expr);
 
-            vectorized::InState* state = reinterpret_cast<vectorized::InState*>(
+            auto* state = reinterpret_cast<vectorized::InState*>(
                     expr_ctx->fn_context(pred->fn_context_index())
                             ->get_function_state(FunctionContext::FRAGMENT_LOCAL));
 
@@ -1120,6 +1114,11 @@ Status ScanLocalState<Derived>::_normalize_in_and_not_in_compound_predicate(
             }
 
             iter = state->hybrid_set->begin();
+
+            if (state->hybrid_set->contain_null()) {
+                *pdt = PushDownType::UNACCEPTABLE;
+                return Status::OK();
+            }
         }
 
         while (iter->has_next()) {

regression-test/data/fault_injection_p0/test_index_inlist_fault_injection.out

@@ -17,3 +17,51 @@
 -- !sql --
 208
 
+-- !sql --
+30
+
+-- !sql --
+0
+
+-- !sql --
+0
+
+-- !sql --
+34
+
+-- !sql --
+2
+
+-- !sql --
+2
+
+-- !sql --
+3
+
+-- !sql --
+2
+
+-- !sql --
+30
+
+-- !sql --
+0
+
+-- !sql --
+0
+
+-- !sql --
+34
+
+-- !sql --
+2
+
+-- !sql --
+2
+
+-- !sql --
+3
+
+-- !sql --
+2
+

regression-test/data/inverted_index_p0/test_index_rqg_bug4.out

@@ -0,0 +1,18 @@
+-- This file is automatically generated. You should know what you did if you want to edit this
+-- !sql --
+\N
+a
+b
+f
+h
+i
+j
+k
+l
+o
+p
+q
+v
+y
+z
+

regression-test/suites/fault_injection_p0/test_index_inlist_fault_injection.groovy

@@ -90,6 +90,35 @@ suite("test_index_inlist_fault_injection", "nonConcurrent") {
       } finally {
         GetDebugPoint().disableDebugPointForAllBEs("segment_iterator._rowid_result_for_index")
       }
+
+      try {
+        sql """ set enable_common_expr_pushdown = true; """
+
+        qt_sql """ select count() from ${indexTbName} where (clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0')); """
+        qt_sql """ select count() from ${indexTbName} where (clientip not in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0')); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' and clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') and clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' or clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') or clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' and clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') or clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' or clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') and clientip match '120.1.0.0'); """
+
+        qt_sql """ select count() from ${indexTbName} where (clientip = '2.1.0.0' and clientip = NULL or clientip = '40.135.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip = '2.1.0.0' or clientip = NULL and clientip = '40.135.0.0'); """
+
+        sql """ set enable_common_expr_pushdown = false; """
+
+        qt_sql """ select count() from ${indexTbName} where (clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0')); """
+        qt_sql """ select count() from ${indexTbName} where (clientip not in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0')); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' and clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') and clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' or clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') or clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' and clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') or clientip match '120.1.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip match '2.1.0.0' or clientip in ('40.135.0.0', '232.0.0.0', NULL, '26.1.0.0', '247.37.0.0') and clientip match '120.1.0.0'); """
+
+        qt_sql """ select count() from ${indexTbName} where (clientip = '2.1.0.0' and clientip = NULL or clientip = '40.135.0.0'); """
+        qt_sql """ select count() from ${indexTbName} where (clientip = '2.1.0.0' or clientip = NULL and clientip = '40.135.0.0'); """
+
+        sql """ set enable_common_expr_pushdown = true; """
+      } finally {
+      }
     } finally {
     }
 }

regression-test/suites/inverted_index_p0/test_index_rqg_bug2.groovy

@@ -14,7 +14,7 @@
 // KIND, either express or implied.  See the License for the
 // specific language governing permissions and limitations
 // under the License.
-suite("test_index_rqg_bug2", "test_index_rqg_bug2"){
+suite("test_index_rqg_bug2", "test_index_rqg_bug"){
     def table1 = "test_index_rqg_bug2"
 
     sql "drop table if exists ${table1}"