branch-4.1: [fix](search) Replace ExcludeScorer with null-bitmap-aware version for MUST_NOT to handle NULL rows #61200

be/src/olap/rowset/segment_v2/inverted_index/query_v2/boolean_query/boolean_query_builder.h

@@ -31,20 +31,26 @@ class OccurBooleanQueryBuilder {
     OccurBooleanQueryBuilder() = default;
     ~OccurBooleanQueryBuilder() = default;
 
-    void add(const QueryPtr& query, Occur occur) { _sub_queries.emplace_back(occur, query); }
+    void add(const QueryPtr& query, Occur occur, std::string binding_key = {}) {
+        _sub_queries.emplace_back(occur, query);
+        _binding_keys.emplace_back(std::move(binding_key));
+    }
 
     void set_minimum_number_should_match(size_t value) { _minimum_number_should_match = value; }
 
     QueryPtr build() {
         if (_minimum_number_should_match.has_value()) {
             return std::make_shared<OccurBooleanQuery>(std::move(_sub_queries),
+                                                       std::move(_binding_keys),
                                                        _minimum_number_should_match.value());
         }
-        return std::make_shared<OccurBooleanQuery>(std::move(_sub_queries));
+        return std::make_shared<OccurBooleanQuery>(std::move(_sub_queries),
+                                                   std::move(_binding_keys));
     }
 
 private:
     std::vector<std::pair<Occur, QueryPtr>> _sub_queries;
+    std::vector<std::string> _binding_keys;
     std::optional<size_t> _minimum_number_should_match;
 };
 

be/src/olap/rowset/segment_v2/inverted_index/query_v2/boolean_query/occur_boolean_query.h

@@ -29,13 +29,16 @@ using OccurBooleanQueryPtr = std::shared_ptr<OccurBooleanQuery>;
 
 class OccurBooleanQuery : public Query {
 public:
-    explicit OccurBooleanQuery(std::vector<std::pair<Occur, QueryPtr>> clauses)
+    explicit OccurBooleanQuery(std::vector<std::pair<Occur, QueryPtr>> clauses,
+                               std::vector<std::string> binding_keys = {})
             : _sub_queries(std::move(clauses)),
+              _binding_keys(std::move(binding_keys)),
               _minimum_number_should_match(compute_default_minimum_should_match(_sub_queries)) {}
 
     OccurBooleanQuery(std::vector<std::pair<Occur, QueryPtr>> clauses,
-                      size_t minimum_number_should_match)
+                      std::vector<std::string> binding_keys, size_t minimum_number_should_match)
             : _sub_queries(std::move(clauses)),
+              _binding_keys(std::move(binding_keys)),
               _minimum_number_should_match(minimum_number_should_match) {}
 
     ~OccurBooleanQuery() override = default;
@@ -47,8 +50,8 @@ class OccurBooleanQuery : public Query {
             sub_weights.emplace_back(occur, query->weight(enable_scoring));
         }
         return std::make_shared<OccurBooleanWeight<SumCombinerPtr>>(
-                std::move(sub_weights), _minimum_number_should_match, enable_scoring,
-                std::make_shared<SumCombiner>());
+                std::move(sub_weights), std::move(_binding_keys), _minimum_number_should_match,
+                enable_scoring, std::make_shared<SumCombiner>());
     }
 
     const std::vector<std::pair<Occur, QueryPtr>>& clauses() const { return _sub_queries; }
@@ -69,6 +72,7 @@ class OccurBooleanQuery : public Query {
     }
 
     std::vector<std::pair<Occur, QueryPtr>> _sub_queries;
+    std::vector<std::string> _binding_keys;
     size_t _minimum_number_should_match = 0;
 };
 

be/src/olap/rowset/segment_v2/inverted_index/query_v2/boolean_query/occur_boolean_weight.cpp

@@ -21,19 +21,27 @@
 #include "olap/rowset/segment_v2/inverted_index/query_v2/disjunction_scorer.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/exclude_scorer.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/intersection.h"
+#include "olap/rowset/segment_v2/inverted_index/query_v2/intersection_scorer.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/reqopt_scorer.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/union/buffered_union.h"
 
 namespace doris::segment_v2::inverted_index::query_v2 {
 
 template <typename ScoreCombinerPtrT>
 OccurBooleanWeight<ScoreCombinerPtrT>::OccurBooleanWeight(
-        std::vector<std::pair<Occur, WeightPtr>> sub_weights, size_t minimum_number_should_match,
-        bool enable_scoring, ScoreCombinerPtrT score_combiner)
+        std::vector<std::pair<Occur, WeightPtr>> sub_weights, std::vector<std::string> binding_keys,
+        size_t minimum_number_should_match, bool enable_scoring, ScoreCombinerPtrT score_combiner)
         : _sub_weights(std::move(sub_weights)),
+          _binding_keys(std::move(binding_keys)),
           _minimum_number_should_match(minimum_number_should_match),
           _enable_scoring(enable_scoring),
-          _score_combiner(std::move(score_combiner)) {}
+          _score_combiner(std::move(score_combiner)) {
+    DCHECK(_binding_keys.empty() || _binding_keys.size() == _sub_weights.size())
+            << "binding_keys size (" << _binding_keys.size() << ") must match sub_weights size ("
+            << _sub_weights.size() << ") when non-empty";
+    // Ensure binding_keys has the same size as sub_weights (pads with empty strings if needed).
+    _binding_keys.resize(_sub_weights.size());
+}
 
 template <typename ScoreCombinerPtrT>
 ScorerPtr OccurBooleanWeight<ScoreCombinerPtrT>::scorer(const QueryExecutionContext& context) {
@@ -62,8 +70,10 @@ template <typename ScoreCombinerPtrT>
 std::unordered_map<Occur, std::vector<ScorerPtr>>
 OccurBooleanWeight<ScoreCombinerPtrT>::per_occur_scorers(const QueryExecutionContext& context) {
     std::unordered_map<Occur, std::vector<ScorerPtr>> result;
-    for (const auto& [occur, weight] : _sub_weights) {
-        auto sub_scorer = weight->scorer(context);
+    for (size_t i = 0; i < _sub_weights.size(); ++i) {
+        const auto& [occur, weight] = _sub_weights[i];
+        const auto& binding_key = _binding_keys[i];
+        auto sub_scorer = weight->scorer(context, binding_key);
         if (sub_scorer) {
             result[occur].push_back(std::move(sub_scorer));
         }
@@ -122,17 +132,6 @@ std::optional<CombinationMethod> OccurBooleanWeight<ScoreCombinerPtrT>::build_sh
     }
 }
 
-template <typename ScoreCombinerPtrT>
-ScorerPtr OccurBooleanWeight<ScoreCombinerPtrT>::build_exclude_opt(
-        std::vector<ScorerPtr> must_not_scorers) {
-    if (must_not_scorers.empty()) {
-        return nullptr;
-    }
-    auto do_nothing = std::make_shared<DoNothingCombiner>();
-    auto specialized_scorer = scorer_union(std::move(must_not_scorers), do_nothing);
-    return into_box_scorer(std::move(specialized_scorer), do_nothing);
-}
-
 template <typename ScoreCombinerPtrT>
 ScorerPtr OccurBooleanWeight<ScoreCombinerPtrT>::effective_must_scorer(
         std::vector<ScorerPtr> must_scorers, size_t must_num_all_scorers) {
@@ -242,13 +241,24 @@ SpecializedScorer OccurBooleanWeight<ScoreCombinerPtrT>::complex_scorer(
         return std::make_shared<EmptyScorer>();
     }
 
-    ScorerPtr exclude_opt = build_exclude_opt(std::move(must_not_scorers));
+    // Collect null bitmaps from MUST_NOT scorers (read from index, no iteration needed)
+    // and union the scorers into one for lazy exclusion.
+    roaring::Roaring exclude_null;
+    ScorerPtr exclude_opt =
+            build_exclude_opt(std::move(must_not_scorers), context.null_resolver, exclude_null);
+
     SpecializedScorer positive_opt =
             build_positive_opt(*should_opt, std::move(must_scorers), combiner, must_special_counts,
                                should_special_counts);
+    // Use null-bitmap-aware ExcludeScorer for MUST_NOT clauses.
+    // ExcludeScorer keeps lazy TRUE exclusion via seek-based iteration and adds
+    // O(1) null bitmap checks so that NOT(NULL) = NULL (SQL three-valued logic).
+    // Documents where the excluded field is NULL are placed in the null bitmap
+    // rather than being incorrectly included in the true result set.
     if (exclude_opt) {
         ScorerPtr positive_boxed = into_box_scorer(std::move(positive_opt), combiner);
-        return make_exclude(std::move(positive_boxed), std::move(exclude_opt));
+        return make_exclude(std::move(positive_boxed), std::move(exclude_opt),
+                            std::move(exclude_null), context.null_resolver);
     }
     return positive_opt;
 }
@@ -321,6 +331,30 @@ ScorerPtr OccurBooleanWeight<ScoreCombinerPtrT>::into_box_scorer(SpecializedScor
             std::move(specialized));
 }
 
+template <typename ScoreCombinerPtrT>
+ScorerPtr OccurBooleanWeight<ScoreCombinerPtrT>::build_exclude_opt(
+        std::vector<ScorerPtr> must_not_scorers, const NullBitmapResolver* resolver,
+        roaring::Roaring& exclude_null_out) {
+    if (must_not_scorers.empty()) {
+        return nullptr;
+    }
+
+    // Collect null bitmaps before union (read from index, no iteration needed).
+    for (auto& s : must_not_scorers) {
+        if (resolver != nullptr && s && s->has_null_bitmap(resolver)) {
+            const auto* nb = s->get_null_bitmap(resolver);
+            if (nb != nullptr) {
+                exclude_null_out |= *nb;
+            }
+        }
+    }
+
+    // Union all MUST_NOT scorers into one for lazy seek-based exclusion.
+    auto do_nothing = std::make_shared<DoNothingCombiner>();
+    auto specialized = scorer_union(std::move(must_not_scorers), do_nothing);
+    return into_box_scorer(std::move(specialized), do_nothing);
+}
+
 template class OccurBooleanWeight<SumCombinerPtr>;
 template class OccurBooleanWeight<DoNothingCombinerPtr>;
 

be/src/olap/rowset/segment_v2/inverted_index/query_v2/boolean_query/occur_boolean_weight.h

@@ -17,6 +17,8 @@
 
 #pragma once
 
+#include <roaring/roaring.hh>
+
 #include "olap/rowset/segment_v2/inverted_index/query_v2/boolean_query/occur.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/scorer.h"
 #include "olap/rowset/segment_v2/inverted_index/query_v2/term_query/term_scorer.h"
@@ -44,8 +46,8 @@ template <typename ScoreCombinerPtrT>
 class OccurBooleanWeight : public Weight {
 public:
     OccurBooleanWeight(std::vector<std::pair<Occur, WeightPtr>> sub_weights,
-                       size_t minimum_number_should_match, bool enable_scoring,
-                       ScoreCombinerPtrT score_combiner);
+                       std::vector<std::string> binding_keys, size_t minimum_number_should_match,
+                       bool enable_scoring, ScoreCombinerPtrT score_combiner);
     ~OccurBooleanWeight() override = default;
 
     ScorerPtr scorer(const QueryExecutionContext& context) override;
@@ -62,8 +64,6 @@ class OccurBooleanWeight : public Weight {
     std::optional<CombinationMethod> build_should_opt(std::vector<ScorerPtr>& must_scorers,
                                                       std::vector<ScorerPtr> should_scorers,
                                                       CombinerT combiner, size_t num_all_scorers);
-    ScorerPtr build_exclude_opt(std::vector<ScorerPtr> must_not_scorers);
-
     ScorerPtr effective_must_scorer(std::vector<ScorerPtr> must_scorers,
                                     size_t must_num_all_scorers);
 
@@ -87,7 +87,12 @@ class OccurBooleanWeight : public Weight {
     template <typename CombinerT>
     ScorerPtr into_box_scorer(SpecializedScorer&& specialized, CombinerT combiner);
 
+    ScorerPtr build_exclude_opt(std::vector<ScorerPtr> must_not_scorers,
+                                const NullBitmapResolver* resolver,
+                                roaring::Roaring& exclude_null_out);
+
     std::vector<std::pair<Occur, WeightPtr>> _sub_weights;
+    std::vector<std::string> _binding_keys;
     size_t _minimum_number_should_match = 1;
     bool _enable_scoring = false;
     ScoreCombinerPtrT _score_combiner;

be/src/olap/rowset/segment_v2/inverted_index/query_v2/exclude_scorer.cpp

@@ -21,11 +21,29 @@ namespace doris::segment_v2::inverted_index::query_v2 {
 
 template <typename TDocSet, typename TDocSetExclude>
 Exclude<TDocSet, TDocSetExclude>::Exclude(TDocSet underlying_docset,
-                                          TDocSetExclude excluding_docset)
+                                          TDocSetExclude excluding_docset,
+                                          roaring::Roaring exclude_null,
+                                          const NullBitmapResolver* resolver)
         : _underlying_docset(std::move(underlying_docset)),
-          _excluding_docset(std::move(excluding_docset)) {
+          _excluding_docset(std::move(excluding_docset)),
+          _exclude_null(std::move(exclude_null)) {
+    // Inherit the include scorer's null bitmap (e.g. from parent AND operations).
+    if (resolver != nullptr && _underlying_docset->has_null_bitmap(resolver)) {
+        const auto* nb = _underlying_docset->get_null_bitmap(resolver);
+        if (nb != nullptr) {
+            _null_bitmap |= *nb;
+        }
+    }
+
     while (_underlying_docset->doc() != TERMINATED) {
         uint32_t target = _underlying_docset->doc();
+        // O(1) null bitmap check (pre-computed from index, cheap).
+        if (!_exclude_null.isEmpty() && _exclude_null.contains(target)) {
+            _null_bitmap.add(target);
+            _underlying_docset->advance();
+            continue;
+        }
+        // Original lazy seek (unchanged).
         if (!is_within(_excluding_docset, target)) {
             break;
         }
@@ -40,6 +58,12 @@ uint32_t Exclude<TDocSet, TDocSetExclude>::advance() {
         if (candidate == TERMINATED) {
             return TERMINATED;
         }
+        // O(1) null bitmap check (pre-computed from index, cheap).
+        if (!_exclude_null.isEmpty() && _exclude_null.contains(candidate)) {
+            _null_bitmap.add(candidate);
+            continue;
+        }
+        // Original lazy seek (unchanged).
         if (!is_within(_excluding_docset, candidate)) {
             return candidate;
         }
@@ -52,6 +76,11 @@ uint32_t Exclude<TDocSet, TDocSetExclude>::seek(uint32_t target) {
     if (candidate == TERMINATED) {
         return TERMINATED;
     }
+    // O(1) null bitmap check (pre-computed from index, cheap).
+    if (!_exclude_null.isEmpty() && _exclude_null.contains(candidate)) {
+        _null_bitmap.add(candidate);
+        return advance();
+    }
     if (!is_within(_excluding_docset, candidate)) {
         return candidate;
     }
@@ -76,11 +105,23 @@ float Exclude<TDocSet, TDocSetExclude>::score() {
     return 0.0F;
 }
 
-ScorerPtr make_exclude(ScorerPtr underlying, ScorerPtr excluding) {
-    return std::make_shared<Exclude<ScorerPtr, ScorerPtr>>(std::move(underlying),
-                                                           std::move(excluding));
+template <typename TDocSet, typename TDocSetExclude>
+bool Exclude<TDocSet, TDocSetExclude>::has_null_bitmap(const NullBitmapResolver* /*resolver*/) {
+    return !_null_bitmap.isEmpty();
+}
+
+template <typename TDocSet, typename TDocSetExclude>
+const roaring::Roaring* Exclude<TDocSet, TDocSetExclude>::get_null_bitmap(
+        const NullBitmapResolver* /*resolver*/) {
+    return _null_bitmap.isEmpty() ? nullptr : &_null_bitmap;
+}
+
+ScorerPtr make_exclude(ScorerPtr underlying, ScorerPtr excluding, roaring::Roaring exclude_null,
+                       const NullBitmapResolver* resolver) {
+    return std::make_shared<Exclude<ScorerPtr, ScorerPtr>>(
+            std::move(underlying), std::move(excluding), std::move(exclude_null), resolver);
 }
 
 template class Exclude<ScorerPtr, ScorerPtr>;
 
-} // namespace doris::segment_v2::inverted_index::query_v2
+} // namespace doris::segment_v2::inverted_index::query_v2

be/src/olap/rowset/segment_v2/inverted_index/query_v2/exclude_scorer.h

@@ -17,6 +17,8 @@
 
 #pragma once
 
+#include <roaring/roaring.hh>
+
 #include "olap/rowset/segment_v2/inverted_index/query_v2/scorer.h"
 
 namespace doris::segment_v2::inverted_index::query_v2 {
@@ -29,7 +31,8 @@ inline bool is_within(TDocSetExclude& docset, uint32_t doc) {
 template <typename TDocSet, typename TDocSetExclude>
 class Exclude final : public Scorer {
 public:
-    Exclude(TDocSet underlying_docset, TDocSetExclude excluding_docset);
+    Exclude(TDocSet underlying_docset, TDocSetExclude excluding_docset,
+            roaring::Roaring exclude_null = {}, const NullBitmapResolver* resolver = nullptr);
     ~Exclude() override = default;
 
     uint32_t advance() override;
@@ -38,13 +41,20 @@ class Exclude final : public Scorer {
     uint32_t size_hint() const override;
     float score() override;
 
+    bool has_null_bitmap(const NullBitmapResolver* resolver = nullptr) override;
+    const roaring::Roaring* get_null_bitmap(const NullBitmapResolver* resolver = nullptr) override;
+
 private:
     TDocSet _underlying_docset;
     TDocSetExclude _excluding_docset;
+    roaring::Roaring _exclude_null;
+    roaring::Roaring _null_bitmap;
 };
 
 using ExcludeScorerPtr = std::shared_ptr<Exclude<ScorerPtr, ScorerPtr>>;
 
-ScorerPtr make_exclude(ScorerPtr underlying, ScorerPtr excluding);
+ScorerPtr make_exclude(ScorerPtr underlying, ScorerPtr excluding,
+                       roaring::Roaring exclude_null = {},
+                       const NullBitmapResolver* resolver = nullptr);
 
 } // namespace doris::segment_v2::inverted_index::query_v2

be/src/vec/functions/function_search.cpp

@@ -641,7 +641,7 @@ Status FunctionSearch::build_query_recursive(const TSearchClause& clause,
                     occur = map_thrift_occur(child_clause.occur);
                 }
 
-                builder->add(child_query, occur);
+                builder->add(child_query, occur, std::move(child_binding_key));
             }
         }