Merge pull request #20391 from vespa-engine/toregge/use-entry-ref-filter-for-remapping-after-compacting-dictionary-keys

Use EntryRefFilter to filter calls to UniqueStoreRemapper::remap() for

2 files changed, 14 insertions, 7 deletions

diff --git a/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp b/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
index b114a355bb4..8790bdd9885 100644
--- a/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/multienumattribute.cpp
@@ -30,13 +30,17 @@ remap_enum_store_refs(const EnumIndexRemapper& remapper, AttributeVector& v, att
     v.logEnumStoreEvent("compactfixup", "drain");
     {
         AttributeVector::EnumModifier enum_guard(v.getEnumModifier());
+        auto& filter = remapper.get_entry_ref_filter();
         v.logEnumStoreEvent("compactfixup", "start");
         for (uint32_t doc = 0; doc < v.getNumDocs(); ++doc) {
             vespalib::ConstArrayRef<WeightedIndex> indicesRef(multi_value_mapping.get(doc));
             WeightedIndexVector indices(indicesRef.cbegin(), indicesRef.cend());
             for (uint32_t i = 0; i < indices.size(); ++i) {
-                EnumIndex oldIndex = indices[i].value();
-                indices[i] = WeightedIndex(remapper.remap(oldIndex), indices[i].weight());
+                EnumIndex ref = indices[i].value();
+                if (ref.valid() && filter.has(ref)) {
+                    ref = remapper.remap(ref);
+                }
+                indices[i] = WeightedIndex(ref, indices[i].weight());
             }
             std::atomic_thread_fence(std::memory_order_release);
             multi_value_mapping.replace(doc, indices);
diff --git a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
index 4323e57f6b1..18805a7b20f 100644
--- a/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/singleenumattribute.cpp
@@ -49,13 +49,16 @@ SingleValueEnumAttributeBase::remap_enum_store_refs(const EnumIndexRemapper& rem
 {
     // update _enumIndices with new EnumIndex values after enum store has been compacted.
     v.logEnumStoreEvent("reenumerate", "reserved");
-    auto new_indexes = std::make_unique<vespalib::Array<EnumIndex>>();
-    new_indexes->reserve(_enumIndices.capacity());
+    vespalib::Array<EnumIndex> new_indexes;
+    new_indexes.reserve(_enumIndices.capacity());
     v.logEnumStoreEvent("reenumerate", "start");
+    auto& filter = remapper.get_entry_ref_filter();
     for (uint32_t i = 0; i < _enumIndices.size(); ++i) {
-        EnumIndex old_index = _enumIndices[i];
-        EnumIndex new_index = remapper.remap(old_index);
-        new_indexes->push_back_fast(new_index);
+        EnumIndex ref = _enumIndices[i];
+        if (ref.valid() && filter.has(ref)) {
+            ref = remapper.remap(ref);
+        }
+        new_indexes.push_back_fast(ref);
     }
     v.logEnumStoreEvent("compactfixup", "drain");
     {