Also make a faster DotproductExecutor when using enumerated values.

9 files changed, 133 insertions, 20 deletions

diff --git a/searchlib/src/tests/features/prod_features.cpp b/searchlib/src/tests/features/prod_features.cpp
index 1d9433c739f..ce2260e1681 100644
--- a/searchlib/src/tests/features/prod_features.cpp
+++ b/searchlib/src/tests/features/prod_features.cpp
@@ -1217,9 +1217,7 @@ Test::testDotProduct()
         vespalib::Stash stash;
         FeatureExecutor &exc = bp.createExecutor(ft.getQueryEnv(), stash);
         // check that we have the optimized enum version
-        dotproduct::wset::DotProductExecutorByCopy<dotproduct::wset::EnumVector, WeightedEnumContent> * myExc =
-            dynamic_cast<dotproduct::wset::DotProductExecutorByCopy<dotproduct::wset::EnumVector, WeightedEnumContent> *>(&exc);
-        EXPECT_TRUE(myExc != nullptr);
+        EXPECT_EQUAL("search::features::dotproduct::wset::(anonymous namespace)::DotProductExecutorByEnum", exc.getClassName());
         EXPECT_EQUAL(1u, deps.output.size());
     }
 }
diff --git a/searchlib/src/vespa/searchlib/attribute/attributevector.h b/searchlib/src/vespa/searchlib/attribute/attributevector.h
index af308044292..58779a17e00 100644
--- a/searchlib/src/vespa/searchlib/attribute/attributevector.h
+++ b/searchlib/src/vespa/searchlib/attribute/attributevector.h
@@ -552,6 +552,7 @@ public:
     virtual SearchContext::UP getSearch(QueryTermSimpleUP term, const attribute::SearchContextParams &params) const = 0;
     virtual const EnumStoreBase *getEnumStoreBase() const;
     virtual const attribute::MultiValueMappingBase *getMultiValueBase() const;
+
 private:
     /**
      * This is called before adding docs will commence.
diff --git a/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp b/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
index e7704cf19c1..05e83012421 100644
--- a/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
@@ -2,8 +2,14 @@
 
 #include "multienumattribute.h"
 #include "multienumattribute.hpp"
+#include <stdexcept>
 
 namespace search {
 
+uint32_t
+IWeightedIndexVector::getEnumHandles(uint32_t, const WeightedIndex * &) const {
+    throw std::runtime_error("IWeightedIndexVector::getEnumHandles() not implmented");
+}
+
 } // namespace search
 
diff --git a/searchlib/src/vespa/searchlib/attribute/multienumattribute.h b/searchlib/src/vespa/searchlib/attribute/multienumattribute.h
index 9300d93168b..ee77baf778f 100644
--- a/searchlib/src/vespa/searchlib/attribute/multienumattribute.h
+++ b/searchlib/src/vespa/searchlib/attribute/multienumattribute.h
@@ -5,9 +5,25 @@
 #include "multivalueattribute.h"
 #include "enumstorebase.h"
 #include "loadedenumvalue.h"
+#include "multivalue.h"
 
 namespace search {
 
+class IWeightedIndexVector {
+public:
+    virtual ~IWeightedIndexVector() = default;
+    using WeightedIndex = multivalue::WeightedValue<EnumStoreBase::Index>;
+    /**
+     * Provides a reference to the underlying enum/weight pairs.
+     * This method should only be invoked if @ref getCollectionType(docId) returns CollectionType::WEIGHTED_SET.
+     *
+     * @param doc document identifier
+     * @param values Reference to values and weights
+     * @return the number of values for this document
+     **/
+    virtual uint32_t getEnumHandles(uint32_t doc, const WeightedIndex * & values) const;
+};
+
 class ReaderBase;
 
 /*
@@ -18,7 +34,8 @@ class ReaderBase;
  * M: MultiValueType
  */
 template <typename B, typename M>
-class MultiValueEnumAttribute : public MultiValueAttribute<B, M>
+class MultiValueEnumAttribute : public MultiValueAttribute<B, M>,
+                                public IWeightedIndexVector
 {
 protected:
     typedef typename B::UniqueSet UniqueSet;
@@ -67,6 +84,8 @@ protected:
 public:
     MultiValueEnumAttribute(const vespalib::string & baseFileName, const AttributeVector::Config & cfg);
 
+    uint32_t getEnumHandles(DocId doc, const IWeightedIndexVector::WeightedIndex * & values) const override final;
+
     void onCommit() override;
     void onUpdateStat() override;
 
@@ -84,6 +103,7 @@ public:
             return indices[0].value().ref();
         }
     }
+
     uint32_t get(DocId doc, EnumHandle * e, uint32_t sz) const override {
         WeightedIndexArrayRef indices(this->_mvMapping.get(doc));
         uint32_t valueCount = indices.size();
diff --git a/searchlib/src/vespa/searchlib/attribute/multienumattribute.hpp b/searchlib/src/vespa/searchlib/attribute/multienumattribute.hpp
index f8f8e84b41e..cb31dbb4b14 100644
--- a/searchlib/src/vespa/searchlib/attribute/multienumattribute.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/multienumattribute.hpp
@@ -132,6 +132,30 @@ MultiValueEnumAttribute(const vespalib::string &baseFileName,
 {
 }
 
+namespace {
+
+template<typename T>
+const IWeightedIndexVector::WeightedIndex *
+extract(const T *) {
+    throw std::runtime_error("IWeightedIndexVector::getEnumHandles not implemented");
+}
+
+template <>
+inline const IWeightedIndexVector::WeightedIndex *
+extract(const IWeightedIndexVector::WeightedIndex * values) {
+    return values;
+}
+
+}
+
+template <typename B, typename M>
+uint32_t
+MultiValueEnumAttribute<B, M>::getEnumHandles(DocId doc, const IWeightedIndexVector::WeightedIndex * & values) const  {
+    WeightedIndexArrayRef indices(this->_mvMapping.get(doc));
+    values = extract(&indices[0]);
+    return indices.size();
+}
+
 template <typename B, typename M>
 void
 MultiValueEnumAttribute<B, M>::onCommit()
diff --git a/searchlib/src/vespa/searchlib/attribute/multivalue.h b/searchlib/src/vespa/searchlib/attribute/multivalue.h
index 330e69a534e..c59f975e00a 100644
--- a/searchlib/src/vespa/searchlib/attribute/multivalue.h
+++ b/searchlib/src/vespa/searchlib/attribute/multivalue.h
@@ -2,11 +2,9 @@
 
 #pragma once
 
-#include <stdint.h>
+#include <cstdint>
 
-namespace search {
-
-namespace multivalue {
+namespace search::multivalue {
 
 template <typename T>
 class Value {
@@ -29,7 +27,7 @@ public:
     bool operator >(const Value<T> & rhs) const { return _v > rhs._v; }
     static bool hasWeight() { return false; }
 
-    static const bool _hasWeight = false;
+    static constexpr bool _hasWeight = false;
 private:
     T _v;
 };
@@ -52,12 +50,10 @@ public:
     bool operator >(const WeightedValue<T> & rhs) const { return _v > rhs._v; }
     static bool hasWeight() { return true; }
 
-    static const bool _hasWeight = true;
+    static constexpr bool _hasWeight = true;
 private:
     T       _v;
     int32_t _w;
 };
 
 }
-}
-
diff --git a/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp b/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
index 55a550837e1..60a8fb372d7 100644
--- a/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
+++ b/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
@@ -9,6 +9,7 @@
 #include <vespa/searchlib/attribute/imported_attribute_vector_read_guard.h>
 #include <vespa/searchlib/attribute/floatbase.h>
 #include <vespa/searchlib/attribute/multinumericattribute.h>
+#include <vespa/searchlib/attribute/multienumattribute.h>
 #include <type_traits>
 
 #include <vespa/log/log.h>
@@ -77,7 +78,6 @@ DotProductExecutorByCopy<Vector, Buffer>::execute(uint32_t docId)
 }
 
 StringVector::StringVector() = default;
-
 StringVector::~StringVector() = default;
 
 template <typename BaseType>
@@ -124,6 +124,48 @@ DotProductExecutor<A>::getAttributeValues(uint32_t docId, const AT * & values)
     return _attribute->getRawValues(docId, values);
 }
 
+namespace {
+
+class DotProductExecutorByEnum : public fef::FeatureExecutor {
+public:
+    using V  = VectorBase<EnumHandle, EnumHandle, feature_t>;
+private:
+    const IWeightedIndexVector * _attribute;
+    V _queryVector;
+    const typename V::HashMap::const_iterator _end;
+public:
+    DotProductExecutorByEnum(const IWeightedIndexVector * attribute, V queryVector);
+    ~DotProductExecutorByEnum() override;
+    void execute(uint32_t docId) override;
+};
+
+DotProductExecutorByEnum::DotProductExecutorByEnum(const IWeightedIndexVector * attribute, V queryVector)
+    : FeatureExecutor(),
+      _attribute(attribute),
+      _queryVector(std::move(queryVector)),
+      _end(_queryVector.syncMap().getDimMap().end())
+{
+}
+
+DotProductExecutorByEnum::~DotProductExecutorByEnum() = default;
+
+void DotProductExecutorByEnum::execute(uint32_t docId) {
+    feature_t val = 0;
+    if (!_queryVector.getDimMap().empty()) {
+        const IWeightedIndexVector::WeightedIndex *values(nullptr);
+        uint32_t sz = _attribute->getEnumHandles(docId, values);
+        for (size_t i = 0; i < sz; ++i) {
+            typename V::HashMap::const_iterator itr = _queryVector.getDimMap().find(values[i].value().ref());
+            if (itr != _end) {
+                val += values[i].weight() * itr->second;
+            }
+        }
+    }
+    outputs().set_number(0, val);
+}
+
+}
+
 }
 
 namespace dotproduct::array {
@@ -431,6 +473,19 @@ bool supportsGetRawValues(const A & attr) noexcept {
     }
 }
 
+bool supportsGetEnumHandles(const IWeightedIndexVector * attr) noexcept {
+    if (attr == nullptr) return false;
+    try {
+        const IWeightedIndexVector::WeightedIndex * tmp = nullptr;
+        attr->getEnumHandles(0, tmp); // Throws if unsupported
+        return true;
+    } catch (const std::runtime_error & e) {
+        (void) e;
+        return false;
+    }
+}
+
+
 // Precondition: attribute->isImported() == false
 template <typename A>
 FeatureExecutor &
@@ -633,6 +688,10 @@ createTypedWsetExecutor(const IAttributeVector * attribute, const Property & pro
     if (attribute->hasEnum()) {
         EnumVector vector(attribute);
         WeightedSetParser::parse(prop.get(), vector);
+        const IWeightedIndexVector * getEnumHandles = dynamic_cast<const IWeightedIndexVector *>(attribute);
+        if (supportsGetEnumHandles(getEnumHandles)) {
+            return &stash.create<DotProductExecutorByEnum>(getEnumHandles, std::move(vector));
+        }
         return &stash.create<DotProductExecutorByCopy<EnumVector, WeightedEnumContent>>(attribute, std::move(vector));
     } else {
         if (attribute->isStringType()) {
diff --git a/searchlib/src/vespa/searchlib/fef/featureexecutor.cpp b/searchlib/src/vespa/searchlib/fef/featureexecutor.cpp
index 02d69fbc5ca..1ea3da3939c 100644
--- a/searchlib/src/vespa/searchlib/fef/featureexecutor.cpp
+++ b/searchlib/src/vespa/searchlib/fef/featureexecutor.cpp
@@ -1,14 +1,17 @@
 // Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 
 #include "featureexecutor.h"
+#include <vespa/vespalib/util/classname.h>
 
-namespace search {
-namespace fef {
+namespace search::fef {
 
-FeatureExecutor::FeatureExecutor()
-    : _inputs(),
-      _outputs()
+FeatureExecutor::FeatureExecutor() = default;
+
+
+vespalib::string
+FeatureExecutor::getClassName() const
 {
+    return vespalib::getClassName(*this);
 }
 
 bool
@@ -52,5 +55,4 @@ FeatureExecutor::bind_match_data(const MatchData &md)
     handle_bind_match_data(md);
 }
 
-} // namespace fef
-} // namespace search
+}
diff --git a/searchlib/src/vespa/searchlib/fef/featureexecutor.h b/searchlib/src/vespa/searchlib/fef/featureexecutor.h
index dfc46230e18..dc8a4ba6075 100644
--- a/searchlib/src/vespa/searchlib/fef/featureexecutor.h
+++ b/searchlib/src/vespa/searchlib/fef/featureexecutor.h
@@ -112,6 +112,13 @@ public:
      **/
     FeatureExecutor();
 
+    /**
+    * Obtain the fully qualified name of the concrete class for this object.
+    *
+    * @return fully qualified class name
+    **/
+    vespalib::string getClassName() const;
+
     // bind order per executor: inputs, outputs, match_data
     void bind_inputs(vespalib::ConstArrayRef<LazyValue> inputs);
     void bind_outputs(vespalib::ArrayRef<NumberOrObject> outputs);