Add flag for marking phrase always expensive.

10 files changed, 79 insertions, 9 deletions

diff --git a/searchcore/src/tests/proton/matching/unpacking_iterators_optimizer/unpacking_iterators_optimizer_test.cpp b/searchcore/src/tests/proton/matching/unpacking_iterators_optimizer/unpacking_iterators_optimizer_test.cpp
index eb222518710..2e1a28402a1 100644
--- a/searchcore/src/tests/proton/matching/unpacking_iterators_optimizer/unpacking_iterators_optimizer_test.cpp
+++ b/searchcore/src/tests/proton/matching/unpacking_iterators_optimizer/unpacking_iterators_optimizer_test.cpp
@@ -254,12 +254,34 @@ std::string split_query_tree_dump =
         "  Term a cheap\n"
         "  Term b cheap\n"
         "  Term c cheap\n";
+std::string split_query_tree_dump_always_expensive =
+        "And 7\n"
+        "  Or 3\n"
+        "    Term t2\n"
+        "    Phrase 3 expensive\n"
+        "      Term a\n"
+        "      Term b\n"
+        "      Term c\n"
+        "    Term x1\n"
+        "  Term x2\n"
+        "  Phrase 3 expensive\n"
+        "    Term a\n"
+        "    Term b\n"
+        "    Term c\n"
+        "  Term t1\n"
+        "  Term a cheap\n"
+        "  Term b cheap\n"
+        "  Term c cheap\n";
 #endif
 
 //-----------------------------------------------------------------------------
 
 Node::UP optimize(Node::UP root, bool white_list) {
-    return UnpackingIteratorsOptimizer::optimize(std::move(root), white_list);
+    return UnpackingIteratorsOptimizer::optimize(std::move(root), white_list, false);
+}
+
+Node::UP optimize(Node::UP root, bool white_list, bool always_mark_phrase_expensive) {
+    return UnpackingIteratorsOptimizer::optimize(std::move(root), white_list, always_mark_phrase_expensive);
 }
 
 TEST(UnpackingIteratorsOptimizerTest, require_that_root_phrase_node_can_be_left_alone) {
@@ -301,4 +323,14 @@ TEST(UnpackingIteratorsOptimizerTest, require_that_query_tree_can_be_split) {
     EXPECT_EQ(actual2, expect);
 }
 
+TEST(UnpackingIteratorsOptimizerTest, require_that_query_tree_can_be_split_always) {
+    std::string actual1 = dump_query(*optimize(make_query_tree(), false, false));
+    std::string actual2 = dump_query(*optimize(make_query_tree(), true, false));
+    std::string actual3 = dump_query(*optimize(make_query_tree(), true, true));
+    std::string expect = split_query_tree_dump;
+    EXPECT_EQ(actual1, expect);
+    EXPECT_EQ(actual2, expect);
+    EXPECT_EQ(actual3, split_query_tree_dump_always_expensive);
+}
+
 GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp b/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
index 7e7532a3182..7beecaca613 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
@@ -193,7 +193,8 @@ MatchToolsFactory(QueryLimiter               & queryLimiter,
     trace.addEvent(4, "Start query setup");
     _query.setWhiteListBlueprint(metaStore.createWhiteListBlueprint());
     trace.addEvent(5, "Deserialize and build query tree");
-    _valid = _query.buildTree(queryStack, location, viewResolver, indexEnv);
+    _valid = _query.buildTree(queryStack, location, viewResolver, indexEnv,
+                              AlwaysMarkPhraseExpensive::check(_queryEnv.getProperties(), rankSetup.always_mark_phrase_expensive()));
     if (_valid) {
         _query.extractTerms(_queryEnv.terms());
         _query.extractLocations(_queryEnv.locations());
diff --git a/searchcore/src/vespa/searchcore/proton/matching/query.cpp b/searchcore/src/vespa/searchcore/proton/matching/query.cpp
index d4f4ae8015d..071e914b405 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/query.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/query.cpp
@@ -165,7 +165,8 @@ Query::~Query() = default;
 
 bool
 Query::buildTree(vespalib::stringref stack, const string &location,
-                 const ViewResolver &resolver, const IIndexEnvironment &indexEnv)
+                 const ViewResolver &resolver, const IIndexEnvironment &indexEnv,
+                 bool always_mark_phrase_expensive)
 {
     SimpleQueryStackDumpIterator stack_dump_iterator(stack);
     _query_tree = QueryTreeCreator<ProtonNodeTypes>::create(stack_dump_iterator);
@@ -173,7 +174,7 @@ Query::buildTree(vespalib::stringref stack, const string &location,
         SameElementModifier prefixSameElementSubIndexes;
         _query_tree->accept(prefixSameElementSubIndexes);
         exchange_location_nodes(location, _query_tree, _locations);
-        _query_tree = UnpackingIteratorsOptimizer::optimize(std::move(_query_tree), bool(_whiteListBlueprint));
+        _query_tree = UnpackingIteratorsOptimizer::optimize(std::move(_query_tree), bool(_whiteListBlueprint), always_mark_phrase_expensive);
         ResolveViewVisitor resolve_visitor(resolver, indexEnv);
         _query_tree->accept(resolve_visitor);
         return true;
diff --git a/searchcore/src/vespa/searchcore/proton/matching/query.h b/searchcore/src/vespa/searchcore/proton/matching/query.h
index b67672ec3ef..6ea326834a5 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/query.h
+++ b/searchcore/src/vespa/searchcore/proton/matching/query.h
@@ -55,7 +55,15 @@ public:
     bool buildTree(vespalib::stringref stack,
                    const vespalib::string &location,
                    const ViewResolver &resolver,
-                   const search::fef::IIndexEnvironment &idxEnv);
+                   const search::fef::IIndexEnvironment &idxEnv)
+    {
+        return buildTree(stack, location, resolver, idxEnv, false);
+    }
+    bool buildTree(vespalib::stringref stack,
+                   const vespalib::string &location,
+                   const ViewResolver &resolver,
+                   const search::fef::IIndexEnvironment &idxEnv,
+                   bool always_mark_phrase_expensive);
 
     /**
      * Extract query terms from the query tree; to be used to build
diff --git a/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.cpp b/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.cpp
index e8dc8ab85ba..c9cfbbfd40e 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.cpp
@@ -73,6 +73,8 @@ struct TermExpander : QueryVisitor {
 
 struct NodeTraverser : TemplateTermVisitor<NodeTraverser, ProtonNodeTypes>
 {
+    bool _always_mark_phrase_expensive;
+    NodeTraverser(bool always_mark_phrase_expensive) : _always_mark_phrase_expensive(always_mark_phrase_expensive) {}
     template <class TermNode> void visitTerm(TermNode &) {}
     void visit(ProtonNodeTypes::And &n) override {
         for (Node *child: n.getChildren()) {
@@ -84,14 +86,19 @@ struct NodeTraverser : TemplateTermVisitor<NodeTraverser, ProtonNodeTypes>
         }
         expander.flush(n);
     }
+    void visit(Phrase &n) override {
+        if (_always_mark_phrase_expensive) {
+            n.set_expensive(true);
+        }
+    }
 };
 
 } // namespace proton::matching::<unnamed>
 
 search::query::Node::UP
-UnpackingIteratorsOptimizer::optimize(search::query::Node::UP root, bool has_white_list)
+UnpackingIteratorsOptimizer::optimize(search::query::Node::UP root, bool has_white_list, bool always_mark_phrase_expensive)
 {
-    NodeTraverser traverser;
+    NodeTraverser traverser(always_mark_phrase_expensive);
     root->accept(traverser);
     if (has_white_list) {
         TermExpander expander;
diff --git a/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.h b/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.h
index f698b79dd0c..fc08ae3cfdd 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.h
+++ b/searchcore/src/vespa/searchcore/proton/matching/unpacking_iterators_optimizer.h
@@ -12,7 +12,7 @@ namespace proton::matching {
  * expensive.
  **/
 struct UnpackingIteratorsOptimizer {
-    static search::query::Node::UP optimize(search::query::Node::UP root, bool has_white_list);
+    static search::query::Node::UP optimize(search::query::Node::UP root, bool has_white_list, bool always_mark_phrase_expensive);
 };
 
 }
diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
index 9b111c4bd5d..f063bad66e1 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
@@ -454,6 +454,12 @@ FuzzyAlgorithm::lookup(const Properties& props, vespalib::FuzzyMatchingAlgorithm
     return vespalib::fuzzy_matching_algorithm_from_string(value, default_value);
 }
 
+const vespalib::string AlwaysMarkPhraseExpensive::NAME("vespa.matching.always_mark_phrase_expensive");
+const bool AlwaysMarkPhraseExpensive::DEFAULT_VALUE(false);
+bool AlwaysMarkPhraseExpensive::check(const Properties &props, bool fallback) {
+    return lookupBool(props, NAME, fallback);
+}
+
 } // namespace matching
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.h b/searchlib/src/vespa/searchlib/fef/indexproperties.h
index c528c4366d6..348ce3ab5e2 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.h
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.h
@@ -339,6 +339,17 @@ namespace matching {
         static vespalib::FuzzyMatchingAlgorithm lookup(const Properties& props);
         static vespalib::FuzzyMatchingAlgorithm lookup(const Properties& props, vespalib::FuzzyMatchingAlgorithm default_value);
     };
+
+    /**
+     * When enabled, the unpacking part of the phrase iterator will be tagged as expensive
+     * under all intermediate iterators, not only AND.
+     **/
+    struct AlwaysMarkPhraseExpensive {
+        static const vespalib::string NAME;
+        static const bool DEFAULT_VALUE;
+        static bool check(const Properties &props) { return check(props, DEFAULT_VALUE); }
+        static bool check(const Properties &props, bool fallback);
+    };
 }
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
index 33e7dbda04b..806be9af47c 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
@@ -60,6 +60,7 @@ RankSetup::RankSetup(const BlueprintFactory &factory, const IIndexEnvironment &i
       _compiled(false),
       _compileError(false),
       _degradationAscendingOrder(false),
+      _always_mark_phrase_expensive(false),
       _diversityAttribute(),
       _diversityMinGroups(1),
       _diversityCutoffFactor(10.0),
@@ -135,6 +136,7 @@ RankSetup::configure()
     _mutateOnSummary._operation = mutate::on_summary::Operation::lookup(_indexEnv.getProperties());
     _mutateAllowQueryOverride = mutate::AllowQueryOverride::check(_indexEnv.getProperties());
     _enableNestedMultivalueGrouping = temporary::EnableNestedMultivalueGrouping::check(_indexEnv.getProperties());
+    _always_mark_phrase_expensive = matching::AlwaysMarkPhraseExpensive::check(_indexEnv.getProperties());
 }
 
 void
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.h b/searchlib/src/vespa/searchlib/fef/ranksetup.h
index 6f4651939ad..0e98c3f1c5d 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.h
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.h
@@ -32,7 +32,7 @@ public:
             : _attribute(attribute),
               _operation(operation)
         {}
-        bool enabled() const { return !_attribute.empty() && !_operation.empty(); }
+        bool enabled() const noexcept { return !_attribute.empty() && !_operation.empty(); }
         vespalib::string _attribute;
         vespalib::string _operation;
     };
@@ -69,6 +69,7 @@ private:
     bool                     _compiled;
     bool                     _compileError;
     bool                     _degradationAscendingOrder;
+    bool                     _always_mark_phrase_expensive;
     vespalib::string         _diversityAttribute;
     uint32_t                 _diversityMinGroups;
     double                   _diversityCutoffFactor;
@@ -221,6 +222,7 @@ public:
     bool isDegradationOrderAscending() const {
         return _degradationAscendingOrder;
     }
+    bool always_mark_phrase_expensive() const noexcept { return _always_mark_phrase_expensive; }
     /** get number of hits to collect during graceful degradation in match phase */
     uint32_t getDegradationMaxHits() const {
         return _degradationMaxHits;