Change posting list for memory index to have PostingListEntry as value.

7 files changed, 106 insertions, 18 deletions

diff --git a/searchlib/src/tests/memoryindex/field_index/field_index_test.cpp b/searchlib/src/tests/memoryindex/field_index/field_index_test.cpp
index 05c905cdc32..0f1c966ad5d 100644
--- a/searchlib/src/tests/memoryindex/field_index/field_index_test.cpp
+++ b/searchlib/src/tests/memoryindex/field_index/field_index_test.cpp
@@ -171,7 +171,7 @@ assertPostingList(const std::string &exp,
         uint32_t docId = itr.getKey();
         ss << docId;
         if (store != nullptr) { // consider features as well
-            EntryRef ref(itr.getData());
+            EntryRef ref(itr.getData().get_features());
             store->setupForField(0, decoder);
             store->setupForUnpackFeatures(ref, decoder);
             decoder.unpackFeatures(matchData, docId);
diff --git a/searchlib/src/vespa/searchlib/memoryindex/field_index.cpp b/searchlib/src/vespa/searchlib/memoryindex/field_index.cpp
index e2e1c99a9b9..8daf87e1899 100644
--- a/searchlib/src/vespa/searchlib/memoryindex/field_index.cpp
+++ b/searchlib/src/vespa/searchlib/memoryindex/field_index.cpp
@@ -127,36 +127,35 @@ FieldIndex::compactFeatures()
             const PostingList *tree = _postingListStore.getTreeEntry(pidx);
             auto pitr = tree->begin(_postingListStore.getAllocator());
             for (; pitr.valid(); ++pitr) {
-                EntryRef oldFeatures(pitr.getData());
+                const PostingListEntry &posting_entry(pitr.getData());
 
                 // Filter on which buffers to move features from when
                 // performing incremental compaction.
 
-                EntryRef newFeatures = _featureStore.moveFeatures(packedIndex, oldFeatures);
+                EntryRef newFeatures = _featureStore.moveFeatures(packedIndex, posting_entry.get_features());
 
                 // Features must be written before reference is updated.
                 std::atomic_thread_fence(std::memory_order_release);
 
-                // Ugly, ugly due to const_cast in iterator
-                pitr.writeData(newFeatures.ref());
+                // Reference the moved data
+                posting_entry.update_features(newFeatures);
             }
         } else {
             const PostingListKeyDataType *shortArray = _postingListStore.getKeyDataEntry(pidx, clusterSize);
             const PostingListKeyDataType *ite = shortArray + clusterSize;
             for (const PostingListKeyDataType *it = shortArray; it < ite; ++it) {
-                EntryRef oldFeatures(it->getData());
+                const PostingListEntry &posting_entry(it->getData());
 
                 // Filter on which buffers to move features from when
                 // performing incremental compaction.
 
-                EntryRef newFeatures = _featureStore.moveFeatures(packedIndex, oldFeatures);
+                EntryRef newFeatures = _featureStore.moveFeatures(packedIndex, posting_entry.get_features());
 
                 // Features must be written before reference is updated.
                 std::atomic_thread_fence(std::memory_order_release);
 
-                // Ugly, ugly due to const_cast, but new data is
-                // semantically equal to old data
-                const_cast<PostingListKeyDataType *>(it)->setData(newFeatures.ref());
+                // Reference the moved data
+                posting_entry.update_features(newFeatures);
             }
         }
     }
@@ -189,7 +188,7 @@ FieldIndex::dump(search::index::IndexBuilder & indexBuilder)
             assert(pitr.valid());
             for (; pitr.valid(); ++pitr) {
                 uint32_t docId = pitr.getKey();
-                EntryRef featureRef(pitr.getData());
+                EntryRef featureRef(pitr.getData().get_features());
                 _featureStore.setupForReadFeatures(featureRef, decoder);
                 decoder.readFeatures(features);
                 features.set_doc_id(docId);
@@ -202,7 +201,7 @@ FieldIndex::dump(search::index::IndexBuilder & indexBuilder)
             const PostingListKeyDataType *kde = kd + clusterSize;
             for (; kd != kde; ++kd) {
                 uint32_t docId = kd->_key;
-                EntryRef featureRef(kd->getData());
+                EntryRef featureRef(kd->getData().get_features());
                 _featureStore.setupForReadFeatures(featureRef, decoder);
                 decoder.readFeatures(features);
                 features.set_doc_id(docId);
diff --git a/searchlib/src/vespa/searchlib/memoryindex/field_index.h b/searchlib/src/vespa/searchlib/memoryindex/field_index.h
index dba57f553b5..d5df2fa49c8 100644
--- a/searchlib/src/vespa/searchlib/memoryindex/field_index.h
+++ b/searchlib/src/vespa/searchlib/memoryindex/field_index.h
@@ -5,6 +5,7 @@
 #include "feature_store.h"
 #include "field_index_remover.h"
 #include "word_store.h"
+#include "posting_list_entry.h"
 #include <vespa/searchlib/index/docidandfeatures.h>
 #include <vespa/searchlib/index/field_length_calculator.h>
 #include <vespa/searchlib/index/indexbuilder.h>
@@ -35,8 +36,8 @@ class OrderedFieldIndexInserter;
 class FieldIndex {
 public:
     // Mapping from docid -> feature ref
-    using PostingList = btree::BTreeRoot<uint32_t, uint32_t, search::btree::NoAggregated>;
-    using PostingListStore = btree::BTreeStore<uint32_t, uint32_t,
+    using PostingList = btree::BTreeRoot<uint32_t, PostingListEntry, search::btree::NoAggregated>;
+    using PostingListStore = btree::BTreeStore<uint32_t, PostingListEntry,
                                                search::btree::NoAggregated,
                                                std::less<uint32_t>,
                                                btree::BTreeDefaultTraits>;
diff --git a/searchlib/src/vespa/searchlib/memoryindex/ordered_field_index_inserter.cpp b/searchlib/src/vespa/searchlib/memoryindex/ordered_field_index_inserter.cpp
index 637a13d67be..1d38e88b747 100644
--- a/searchlib/src/vespa/searchlib/memoryindex/ordered_field_index_inserter.cpp
+++ b/searchlib/src/vespa/searchlib/memoryindex/ordered_field_index_inserter.cpp
@@ -111,7 +111,7 @@ OrderedFieldIndexInserter::add(uint32_t docId,
     assert(_prevDocId == noDocId || _prevDocId < docId ||
            (_prevDocId == docId && !_prevAdd));
     datastore::EntryRef featureRef = _fieldIndex.addFeatures(features);
-    _adds.push_back(PostingListKeyDataType(docId, featureRef.ref()));
+    _adds.push_back(PostingListKeyDataType(docId, featureRef));
     _listener.insert(_dItr.getKey()._wordRef, docId);
     _prevDocId = docId;
     _prevAdd = true;
diff --git a/searchlib/src/vespa/searchlib/memoryindex/posting_iterator.cpp b/searchlib/src/vespa/searchlib/memoryindex/posting_iterator.cpp
index 63040aab66f..290aa16dfe4 100644
--- a/searchlib/src/vespa/searchlib/memoryindex/posting_iterator.cpp
+++ b/searchlib/src/vespa/searchlib/memoryindex/posting_iterator.cpp
@@ -6,6 +6,7 @@
 #include <vespa/vespalib/btree/btreenodeallocator.hpp>
 #include <vespa/vespalib/btree/btreenodestore.hpp>
 #include <vespa/vespalib/btree/btreeroot.hpp>
+#include <vespa/vespalib/btree/btreestore.hpp>
 
 #include <vespa/log/log.h>
 LOG_SETUP(".searchlib.memoryindex.posting_iterator");
@@ -62,7 +63,7 @@ PostingIterator::doUnpack(uint32_t docId)
     assert(docId == getDocId());
     assert(_itr.valid());
     assert(docId == _itr.getKey());
-    datastore::EntryRef featureRef(_itr.getData());
+    datastore::EntryRef featureRef(_itr.getData().get_features());
     _featureStore.setupForUnpackFeatures(featureRef, _featureDecoder);
     _featureDecoder.unpackFeatures(_matchData, docId);
     setUnpacked();
@@ -70,3 +71,59 @@ PostingIterator::doUnpack(uint32_t docId)
 
 }
 
+namespace search::btree {
+
+template class BTreeNodeTT<uint32_t,
+                           search::memoryindex::PostingListEntry,
+                           search::btree::NoAggregated,
+                           BTreeDefaultTraits::INTERNAL_SLOTS>;
+
+template class BTreeLeafNode<uint32_t,
+                             search::memoryindex::PostingListEntry,
+                             search::btree::NoAggregated,
+                             BTreeDefaultTraits::LEAF_SLOTS>;
+
+template class BTreeNodeStore<uint32_t,
+                              search::memoryindex::PostingListEntry,
+                              search::btree::NoAggregated,
+                              BTreeDefaultTraits::INTERNAL_SLOTS,
+                              BTreeDefaultTraits::LEAF_SLOTS>;
+
+template class BTreeIteratorBase<uint32_t,
+                                 search::memoryindex::PostingListEntry,
+                                 search::btree::NoAggregated,
+                                 BTreeDefaultTraits::INTERNAL_SLOTS,
+                                 BTreeDefaultTraits::LEAF_SLOTS,
+                                 BTreeDefaultTraits::PATH_SIZE>;
+
+template class BTreeIterator<uint32_t,
+                             search::memoryindex::PostingListEntry,
+                             search::btree::NoAggregated,
+                             std::less<uint32_t>,
+                             BTreeDefaultTraits>;
+
+template class BTree<uint32_t,
+                     search::memoryindex::PostingListEntry,
+                     search::btree::NoAggregated,
+                     std::less<uint32_t>,
+                     BTreeDefaultTraits>;
+
+template class BTreeRoot<uint32_t,
+                         search::memoryindex::PostingListEntry,
+                         search::btree::NoAggregated,
+                         std::less<uint32_t>,
+                         BTreeDefaultTraits>;
+
+template class BTreeRootBase<uint32_t,
+                             search::memoryindex::PostingListEntry,
+                             search::btree::NoAggregated,
+                             BTreeDefaultTraits::INTERNAL_SLOTS,
+                             BTreeDefaultTraits::LEAF_SLOTS>;
+
+template class BTreeNodeAllocator<uint32_t,
+                                  search::memoryindex::PostingListEntry,
+                                  search::btree::NoAggregated,
+                                  BTreeDefaultTraits::INTERNAL_SLOTS,
+                                  BTreeDefaultTraits::LEAF_SLOTS>;
+
+}
diff --git a/searchlib/src/vespa/searchlib/memoryindex/posting_list_entry.h b/searchlib/src/vespa/searchlib/memoryindex/posting_list_entry.h
new file mode 100644
index 00000000000..af204c84bbb
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/memoryindex/posting_list_entry.h
@@ -0,0 +1,31 @@
+// Copyright 2019 Oath Inc. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+# pragma once
+
+#include <vespa/vespalib/datastore/entryref.h>
+
+namespace search::memoryindex {
+
+/**
+ * Entry per document in memory index posting list.
+ */
+class PostingListEntry {
+    mutable datastore::EntryRef _features; // reference to compressed features
+
+public:
+    PostingListEntry(datastore::EntryRef features)
+        : _features(features)
+    {
+    }
+
+    PostingListEntry()
+        : _features()
+    {
+    }
+       datastore::EntryRef get_features() const { return _features; }
+
+    // Reference moved data (used when compacting FeatureStore)
+    void update_features(datastore::EntryRef features) const { _features = features; }
+};
+
+}
diff --git a/searchlib/src/vespa/searchlib/test/fakedata/fakememtreeocc.cpp b/searchlib/src/vespa/searchlib/test/fakedata/fakememtreeocc.cpp
index 54c0aa866b4..f5300430bea 100644
--- a/searchlib/src/vespa/searchlib/test/fakedata/fakememtreeocc.cpp
+++ b/searchlib/src/vespa/searchlib/test/fakedata/fakememtreeocc.cpp
@@ -261,13 +261,13 @@ FakeMemTreeOccMgr::flush()
         lastWord = wordIdx;
         if (i->getRemove()) {
             if (itr.valid() && itr.getKey() == docId) {
-                uint64_t bits = _featureStore.bitSize(fw->getPackedIndex(), EntryRef(itr.getData()));
+                uint64_t bits = _featureStore.bitSize(fw->getPackedIndex(), EntryRef(itr.getData().get_features()));
                 _featureSizes[wordIdx] -= RefType::align((bits + 7) / 8) * 8;
                 tree.remove(itr);
             }
         } else {
             if (!itr.valid() || docId < itr.getKey()) {
-                tree.insert(itr, docId, i->getFeatureRef().ref());
+                tree.insert(itr, docId, i->getFeatureRef());
             }
         }
     }