Factor out ReferenceMappings from ReferenceAttribute.

8 files changed, 222 insertions, 96 deletions

diff --git a/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp b/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
index a638309576a..7e2e8904170 100644
--- a/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
+++ b/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
@@ -15,6 +15,7 @@ LOG_SETUP("reference_attribute_test");
 using search::MemoryUsage;
 using vespalib::ArrayRef;
 using generation_t = vespalib::GenerationHandler::generation_t;
+using search::attribute::Reference;
 using search::attribute::ReferenceAttribute;
 using search::attribute::Config;
 using search::attribute::BasicType;
@@ -99,7 +100,7 @@ struct Fixture
     }
 
     const GlobalId *get(uint32_t doc) {
-        const ReferenceAttribute::Reference *ref = _attr->getReference(doc);
+        const Reference *ref = _attr->getReference(doc);
         if (ref != nullptr) {
             return &ref->gid();
         } else {
@@ -107,7 +108,7 @@ struct Fixture
         }
     }
 
-    const ReferenceAttribute::Reference *getRef(uint32_t doc) {
+    const Reference *getRef(uint32_t doc) {
         return _attr->getReference(doc);
     }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt b/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
index 9308ae015b5..6838119ccd9 100644
--- a/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
+++ b/searchlib/src/vespa/searchlib/attribute/CMakeLists.txt
@@ -79,6 +79,7 @@ vespa_add_library(searchlib_attribute OBJECT
     readerbase.cpp
     reference_attribute.cpp
     reference_attribute_saver.cpp
+    reference_mappings.cpp
     singleenumattribute.cpp
     singleenumattributesaver.cpp
     singlenumericattribute.cpp
diff --git a/searchlib/src/vespa/searchlib/attribute/reference.h b/searchlib/src/vespa/searchlib/attribute/reference.h
new file mode 100644
index 00000000000..db8c94b4a11
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/attribute/reference.h
@@ -0,0 +1,42 @@
+// Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include <vespa/document/base/globalid.h>
+#include <vespa/searchlib/datastore/entryref.h>
+
+namespace search::attribute {
+
+/*
+ * Class representing a single reference in a reference attribute.
+ */
+class Reference {
+    using EntryRef = search::datastore::EntryRef;
+    using GlobalId = document::GlobalId;
+    GlobalId _gid;
+    mutable uint32_t _lid;  // referenced lid
+    mutable EntryRef _revMapIdx; // map from gid to lids referencing gid
+public:
+    Reference()
+        : _gid(),
+          _lid(0u),
+          _revMapIdx()
+    {
+    }
+    Reference(const GlobalId &gid_)
+        : _gid(gid_),
+          _lid(0u),
+          _revMapIdx()
+    {
+    }
+    bool operator<(const Reference &rhs) const {
+        return _gid < rhs._gid;
+    }
+    const GlobalId &gid() const { return _gid; }
+    uint32_t lid() const { return _lid; }
+    EntryRef revMapIdx() const { return _revMapIdx; }
+    void setLid(uint32_t referencedLid) const { _lid = referencedLid; }
+    void setRevMapIdx(EntryRef newRevMapIdx) const { _revMapIdx = newRevMapIdx; }
+};
+
+}
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp b/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
index 2798127e981..ee911c9a8cc 100644
--- a/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
@@ -7,7 +7,6 @@
 #include <vespa/searchlib/datastore/unique_store_builder.h>
 #include <vespa/searchlib/datastore/datastore.hpp>
 #include <vespa/searchlib/datastore/unique_store.hpp>
-#include <vespa/searchlib/btree/btreestore.hpp>
 #include <vespa/searchlib/common/i_gid_to_lid_mapper_factory.h>
 #include <vespa/searchlib/common/i_gid_to_lid_mapper.h>
 #include <vespa/vespalib/data/fileheader.h>
@@ -37,8 +36,7 @@ ReferenceAttribute::ReferenceAttribute(const vespalib::stringref baseFileName,
       _indices(getGenerationHolder()),
       _cachedUniqueStoreMemoryUsage(),
       _gidToLidMapperFactory(),
-      _reverseMappingIndices(getGenerationHolder()),
-      _reverseMapping()
+      _referenceMappings(getGenerationHolder())
 {
     setEnum(true);
     enableEnumeratedSave(true);
@@ -46,16 +44,13 @@ ReferenceAttribute::ReferenceAttribute(const vespalib::stringref baseFileName,
 
 ReferenceAttribute::~ReferenceAttribute()
 {
-    _reverseMapping.clearBuilder();
+    _referenceMappings.clearBuilder();
     incGeneration(); // Force freeze
     const auto &store = _store;
     const auto saver = _store.getSaver();
     saver.foreach_key([&store,this](EntryRef ref)
                       {   const Reference &entry = store.get(ref);
-                          EntryRef revMapIdx = entry.revMapIdx();
-                          if (revMapIdx.valid()) {
-                              _reverseMapping.clear(revMapIdx);
-                          }
+                          _referenceMappings.clearMapping(entry);
                       });
     incGeneration(); // Force freeze
 }
@@ -83,46 +78,24 @@ ReferenceAttribute::addDoc(DocId &doc)
 }
 
 void
-ReferenceAttribute::syncReverseMappingIndices(const Reference &entry)
-{
-    uint32_t referencedLid = entry.lid();
-    if (referencedLid != 0u) {
-        _reverseMappingIndices.ensure_size(referencedLid + 1);
-        _reverseMappingIndices[referencedLid] = entry.revMapIdx();
-    }
-}
-
-void
 ReferenceAttribute::removeReverseMapping(EntryRef oldRef, uint32_t lid)
 {
     const auto &entry = _store.get(oldRef);
-    EntryRef revMapIdx = entry.revMapIdx();
-    _reverseMapping.apply(revMapIdx, nullptr, nullptr, &lid, &lid + 1);
-    std::atomic_thread_fence(std::memory_order_release);
-    entry.setRevMapIdx(revMapIdx);
-    syncReverseMappingIndices(entry);
+    _referenceMappings.removeReverseMapping(entry, lid);
 }
 
 void
 ReferenceAttribute::addReverseMapping(EntryRef newRef, uint32_t lid)
 {
     const auto &entry = _store.get(newRef);
-    EntryRef revMapIdx = entry.revMapIdx();
-    ReverseMapping::KeyDataType add(lid, btree::BTreeNoLeafData());
-    _reverseMapping.apply(revMapIdx, &add, &add + 1, nullptr, nullptr);
-    std::atomic_thread_fence(std::memory_order_release);
-    entry.setRevMapIdx(revMapIdx);
-    syncReverseMappingIndices(entry);
+    _referenceMappings.addReverseMapping(entry, lid);
 }
 
 void
 ReferenceAttribute::buildReverseMapping(EntryRef newRef, const std::vector<ReverseMapping::KeyDataType> &adds)
 {
     const auto &entry = _store.get(newRef);
-    EntryRef revMapIdx = entry.revMapIdx();
-    assert(!revMapIdx.valid());
-    _reverseMapping.apply(revMapIdx, &adds[0], &adds[adds.size()], nullptr, nullptr);
-    entry.setRevMapIdx(revMapIdx);
+    _referenceMappings.buildReverseMapping(entry, adds);
 }
 
 void
@@ -174,7 +147,7 @@ ReferenceAttribute::clearDoc(DocId doc)
 void
 ReferenceAttribute::removeOldGenerations(generation_t firstUsed)
 {
-    _reverseMapping.trimHoldLists(firstUsed);
+    _referenceMappings.trimHoldLists(firstUsed);
     _store.trimHoldLists(firstUsed);
     getGenerationHolder().trimHoldLists(firstUsed);
 }
@@ -182,9 +155,9 @@ ReferenceAttribute::removeOldGenerations(generation_t firstUsed)
 void
 ReferenceAttribute::onGenerationChange(generation_t generation)
 {
-    _reverseMapping.freeze();
+    _referenceMappings.freeze();
     _store.freeze();
-    _reverseMapping.transferHoldLists(generation - 1);
+    _referenceMappings.transferHoldLists(generation - 1);
     _store.transferHoldLists(generation - 1);
     getGenerationHolder().transferHoldLists(generation - 1);
 }
@@ -287,7 +260,7 @@ ReferenceAttribute::update(DocId doc, const GlobalId &gid)
     }
 }
 
-const ReferenceAttribute::Reference *
+const Reference *
 ReferenceAttribute::getReference(DocId doc)
 {
     assert(doc < _indices.size());
@@ -360,14 +333,7 @@ ReferenceAttribute::notifyGidToLidChange(const GlobalId &gid, DocId referencedLi
     EntryRef ref = _store.find(gid);
     if (ref.valid()) {
         const auto &entry = _store.get(ref);
-        uint32_t oldReferencedLid = entry.lid();
-        if (oldReferencedLid != referencedLid) {
-            if (oldReferencedLid != 0u && oldReferencedLid < _reverseMappingIndices.size()) {
-                _reverseMappingIndices[oldReferencedLid] = EntryRef();
-            }
-            entry.setLid(referencedLid);
-        }
-        syncReverseMappingIndices(entry);
+        _referenceMappings.notifyGidToLidChange(entry, referencedLid);
         commit();
     }
 }
@@ -383,7 +349,7 @@ ReferenceAttribute::populateReferencedLids()
         saver.foreach_key([&store,&mapper,this](EntryRef ref)
                           {   const Reference &entry = store.get(ref);
                               entry.setLid(mapper.mapGidToLid(entry.gid()));
-                              syncReverseMappingIndices(entry); });
+                              _referenceMappings.syncReverseMappingIndices(entry); });
     }
     commit();
 }
@@ -419,7 +385,7 @@ IMPLEMENT_IDENTIFIABLE_ABSTRACT(ReferenceAttribute, AttributeVector);
 
 namespace datastore {
 
-using Reference = attribute::ReferenceAttribute::Reference;
+using Reference = attribute::Reference;
 
 template class UniqueStore<Reference, EntryRefT<22>>;
 template class UniqueStoreBuilder<Reference, EntryRefT<22>>;
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_attribute.h b/searchlib/src/vespa/searchlib/attribute/reference_attribute.h
index 5647acfd163..a751213dc9a 100644
--- a/searchlib/src/vespa/searchlib/attribute/reference_attribute.h
+++ b/searchlib/src/vespa/searchlib/attribute/reference_attribute.h
@@ -3,10 +3,10 @@
 #pragma once
 
 #include "not_implemented_attribute.h"
-#include <vespa/document/base/globalid.h>
+#include "reference_mappings.h"
+#include "reference.h"
 #include <vespa/searchlib/datastore/unique_store.h>
 #include <vespa/searchlib/common/rcuvector.h>
-#include "postingstore.h"
 
 namespace search {
 
@@ -28,32 +28,6 @@ class ReferenceAttribute : public NotImplementedAttribute
 public:
     using EntryRef = search::datastore::EntryRef;
     using GlobalId = document::GlobalId;
-    class Reference {
-        GlobalId _gid;
-        mutable uint32_t _lid;  // referenced lid
-        mutable EntryRef _revMapIdx; // map from gid to lids referencing gid
-    public:
-        Reference()
-            : _gid(),
-              _lid(0u),
-              _revMapIdx()
-        {
-        }
-        Reference(const GlobalId &gid_)
-            : _gid(gid_),
-              _lid(0u),
-              _revMapIdx()
-        {
-        }
-        bool operator<(const Reference &rhs) const {
-            return _gid < rhs._gid;
-        }
-        const GlobalId &gid() const { return _gid; }
-        uint32_t lid() const { return _lid; }
-        EntryRef revMapIdx() const { return _revMapIdx; }
-        void setLid(uint32_t referencedLid) const { _lid = referencedLid; }
-        void setRevMapIdx(EntryRef newRevMapIdx) const { _revMapIdx = newRevMapIdx; }
-    };
     using ReferenceStore = datastore::UniqueStore<Reference>;
     using ReferenceStoreIndices = RcuVectorBase<EntryRef>;
     using IndicesCopyVector = vespalib::Array<EntryRef>;
@@ -68,12 +42,7 @@ private:
     ReferenceStoreIndices _indices;
     MemoryUsage _cachedUniqueStoreMemoryUsage;
     std::shared_ptr<IGidToLidMapperFactory> _gidToLidMapperFactory;
-    // Vector containing references to trees of lids referencing given
-    // referenced lid.
-    ReverseMappingIndices _reverseMappingIndices;
-    // Store of B-Trees, used to map from gid or referenced lid to
-    // referencing lids.
-    ReverseMapping _reverseMapping;
+    ReferenceMappings _referenceMappings;
 
     virtual void onAddDocs(DocId docIdLimit) override;
     virtual void removeOldGenerations(generation_t firstUsed) override;
@@ -87,7 +56,6 @@ private:
     bool considerCompact(const CompactionStrategy &compactionStrategy);
     void compactWorst();
     IndicesCopyVector getIndicesCopy(uint32_t size) const;
-    void syncReverseMappingIndices(const Reference &entry);
     void removeReverseMapping(EntryRef oldRef, uint32_t lid);
     void addReverseMapping(EntryRef newRef, uint32_t lid);
     void buildReverseMapping(EntryRef newRef, const std::vector<ReverseMapping::KeyDataType> &adds);
@@ -113,18 +81,10 @@ public:
 
     template <typename FunctionType>
     void
-    foreach_lid(uint32_t referencedLid, FunctionType &&func) const;
-};
-
-template <typename FunctionType>
-void
-ReferenceAttribute::foreach_lid(uint32_t referencedLid, FunctionType &&func) const
-{
-    if (referencedLid < _reverseMappingIndices.size()) {
-        EntryRef revMapIdx = _reverseMappingIndices[referencedLid];
-        _reverseMapping.foreach_frozen_key(revMapIdx, func);
+    foreach_lid(uint32_t referencedLid, FunctionType &&func) const {
+        _referenceMappings.foreach_lid(referencedLid, std::forward<FunctionType>(func));
     }
-}
+};
 
 }
 }
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_attribute_saver.h b/searchlib/src/vespa/searchlib/attribute/reference_attribute_saver.h
index d1ee5e0339a..7d85c60c3e6 100644
--- a/searchlib/src/vespa/searchlib/attribute/reference_attribute_saver.h
+++ b/searchlib/src/vespa/searchlib/attribute/reference_attribute_saver.h
@@ -8,6 +8,7 @@
 #include <vespa/searchlib/datastore/unique_store_saver.h>
 #include <vespa/searchlib/common/rcuvector.h>
 #include "reference_attribute.h"
+#include "reference.h"
 
 namespace search {
 namespace attribute {
@@ -31,7 +32,6 @@ private:
     using GlobalId = document::GlobalId;
     using IndicesCopyVector = ReferenceAttribute::IndicesCopyVector;
     using Store = ReferenceAttribute::ReferenceStore;
-    using Reference = ReferenceAttribute::Reference;
     using Saver = Store::Saver;
     IndicesCopyVector _indices;
     const Store &_store;
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_mappings.cpp b/searchlib/src/vespa/searchlib/attribute/reference_mappings.cpp
new file mode 100644
index 00000000000..67f02c50d88
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/attribute/reference_mappings.cpp
@@ -0,0 +1,82 @@
+// Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include "reference_mappings.h"
+#include "reference.h"
+#include <vespa/searchlib/datastore/datastore.hpp>
+#include <vespa/searchlib/btree/btreestore.hpp>
+
+namespace search::attribute {
+
+ReferenceMappings::ReferenceMappings(GenerationHolder &genHolder)
+    : _reverseMappingIndices(genHolder),
+      _reverseMapping()
+{
+}
+
+ReferenceMappings::~ReferenceMappings()
+{
+}
+
+void
+ReferenceMappings::clearMapping(const Reference &entry)
+{
+    EntryRef revMapIdx = entry.revMapIdx();
+    if (revMapIdx.valid()) {
+        _reverseMapping.clear(revMapIdx);
+    }
+}
+
+void
+ReferenceMappings::syncReverseMappingIndices(const Reference &entry)
+{
+    uint32_t referencedLid = entry.lid();
+    if (referencedLid != 0u) {
+        _reverseMappingIndices.ensure_size(referencedLid + 1);
+        _reverseMappingIndices[referencedLid] = entry.revMapIdx();
+    }
+}
+
+void
+ReferenceMappings::removeReverseMapping(const Reference &entry, uint32_t lid)
+{
+    EntryRef revMapIdx = entry.revMapIdx();
+    _reverseMapping.apply(revMapIdx, nullptr, nullptr, &lid, &lid + 1);
+    std::atomic_thread_fence(std::memory_order_release);
+    entry.setRevMapIdx(revMapIdx);
+    syncReverseMappingIndices(entry);
+}
+
+void
+ReferenceMappings::addReverseMapping(const Reference &entry, uint32_t lid)
+{
+    EntryRef revMapIdx = entry.revMapIdx();
+    ReverseMapping::KeyDataType add(lid, btree::BTreeNoLeafData());
+    _reverseMapping.apply(revMapIdx, &add, &add + 1, nullptr, nullptr);
+    std::atomic_thread_fence(std::memory_order_release);
+    entry.setRevMapIdx(revMapIdx);
+    syncReverseMappingIndices(entry);
+}
+
+void
+ReferenceMappings::buildReverseMapping(const Reference &entry, const std::vector<ReverseMapping::KeyDataType> &adds)
+{
+    EntryRef revMapIdx = entry.revMapIdx();
+    assert(!revMapIdx.valid());
+    _reverseMapping.apply(revMapIdx, &adds[0], &adds[adds.size()], nullptr, nullptr);
+    entry.setRevMapIdx(revMapIdx);
+}
+
+void
+ReferenceMappings::notifyGidToLidChange(const Reference &entry, uint32_t referencedLid)
+{
+    uint32_t oldReferencedLid = entry.lid();
+    if (oldReferencedLid != referencedLid) {
+        if (oldReferencedLid != 0u && oldReferencedLid < _reverseMappingIndices.size()) {
+            _reverseMappingIndices[oldReferencedLid] = EntryRef();
+        }
+        entry.setLid(referencedLid);
+    }
+    syncReverseMappingIndices(entry);
+}
+
+}
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_mappings.h b/searchlib/src/vespa/searchlib/attribute/reference_mappings.h
new file mode 100644
index 00000000000..129ccbbea25
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/attribute/reference_mappings.h
@@ -0,0 +1,74 @@
+// Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include <vespa/searchlib/btree/btreestore.h>
+#include <vespa/searchlib/common/rcuvector.h>
+
+namespace search::attribute {
+
+class Reference;
+
+/*
+ * Class representing mappings in a reference attribute.
+ */
+class ReferenceMappings
+{
+    using GenerationHolder = vespalib::GenerationHolder;
+    using EntryRef = search::datastore::EntryRef;
+    using ReverseMappingIndices = RcuVectorBase<EntryRef>;
+    using ReverseMapping = btree::BTreeStore<uint32_t, btree::BTreeNoLeafData,
+                                             btree::NoAggregated,
+                                             std::less<uint32_t>,
+                                             btree::BTreeDefaultTraits,
+                                             btree::NoAggrCalc>;
+    using generation_t = vespalib::GenerationHandler::generation_t;
+
+    // Vector containing references to trees of lids referencing given
+    // referenced lid.
+    ReverseMappingIndices _reverseMappingIndices;
+    // Store of B-Trees, used to map from gid or referenced lid to
+    // referencing lids.
+    ReverseMapping _reverseMapping;
+
+
+public:
+    ReferenceMappings(GenerationHolder &genHolder);
+
+    ~ReferenceMappings();
+
+    // Cleanup helpers, to free resources
+    void clearBuilder() { _reverseMapping.clearBuilder(); }
+    void clearMapping(const Reference &entry);
+
+    // Hold list management & freezing
+    void trimHoldLists(generation_t usedGen) { _reverseMapping.trimHoldLists(usedGen); }
+    void freeze() { _reverseMapping.freeze(); }
+    void transferHoldLists(generation_t generation) { _reverseMapping.transferHoldLists(generation); }
+
+    // Handle mapping changes
+    void notifyGidToLidChange(const Reference &entry, uint32_t referencedLid);
+    void syncReverseMappingIndices(const Reference &entry);
+    void removeReverseMapping(const Reference &entry, uint32_t lid);
+    void addReverseMapping(const Reference &entry, uint32_t lid);
+
+    // Setup mapping after load
+    void buildReverseMapping(const Reference &entry, const std::vector<ReverseMapping::KeyDataType> &adds);
+
+    // Reader API, reader must hold generation guard
+    template <typename FunctionType>
+    void
+    foreach_lid(uint32_t referencedLid, FunctionType &&func) const;
+};
+
+template <typename FunctionType>
+void
+ReferenceMappings::foreach_lid(uint32_t referencedLid, FunctionType &&func) const
+{
+    if (referencedLid < _reverseMappingIndices.size()) {
+        EntryRef revMapIdx = _reverseMappingIndices[referencedLid];
+        _reverseMapping.foreach_frozen_key(revMapIdx, std::forward<FunctionType>(func));
+    }
+}
+
+}