From ceab4f213058cefa2458a76e82244cf49956bacc Mon Sep 17 00:00:00 2001
From: Henning Baldersheim <balder@yahoo-inc.com>
Date: Mon, 11 Dec 2023 13:40:03 +0100
Subject: Revert "Revert "relative estimate""

---
 .../lid_allocator/lid_allocator_test.cpp           | 23 ++++++++++++++++++----
 .../proton/documentmetastore/lid_allocator.cpp     |  4 ++++
 2 files changed, 23 insertions(+), 4 deletions(-)

(limited to 'searchcore/src')

diff --git a/searchcore/src/tests/proton/documentmetastore/lid_allocator/lid_allocator_test.cpp b/searchcore/src/tests/proton/documentmetastore/lid_allocator/lid_allocator_test.cpp
index b2fa8675835..e136e491f05 100644
--- a/searchcore/src/tests/proton/documentmetastore/lid_allocator/lid_allocator_test.cpp
+++ b/searchcore/src/tests/proton/documentmetastore/lid_allocator/lid_allocator_test.cpp
@@ -102,10 +102,15 @@ protected:
         }
         return result;
     }
-
-    SimpleResult get_active_lids_in_search_iterator(uint32_t docid_limit, bool filter) {
+    
+    Blueprint::UP make_whitelist_blueprint(uint32_t docid_limit) {
         auto blueprint = _allocator.createWhiteListBlueprint();
         blueprint->setDocIdLimit(docid_limit);
+        return blueprint;
+    }
+    
+    SimpleResult get_active_lids_in_search_iterator(uint32_t docid_limit, bool filter) {
+        auto blueprint = make_whitelist_blueprint(docid_limit);
         std::unique_ptr<SearchIterator> iterator;
         MatchData md(MatchData::params());
         if (filter) {
@@ -119,8 +124,7 @@ protected:
     }
 
     Trinary filter_search_iterator_matches_any(uint32_t docid_limit) {
-        auto blueprint = _allocator.createWhiteListBlueprint();
-        blueprint->setDocIdLimit(docid_limit);
+        auto blueprint = make_whitelist_blueprint(docid_limit);
         auto iterator = blueprint->createFilterSearch(true, Blueprint::FilterConstraint::UPPER_BOUND);
         return iterator->matches_any();
     }
@@ -170,6 +174,17 @@ TEST_F(LidAllocatorTest, filter_search_iterator_matches_all_when_all_lids_are_ac
     EXPECT_EQ(SimpleResult({1, 2, 3, 4}), get_active_lids_in_search_iterator(6, false));
 }
 
+TEST_F(LidAllocatorTest, whitelist_blueprint_can_maximize_relative_estimate)
+{
+    register_lids({ 1, 2, 3, 4 });
+    activate_lids({ 1, 2, 3, 4 }, true);
+    // the number of hits are overestimated based on the number of
+    // documents that could be active (100 in this test fixture)
+    EXPECT_EQ(make_whitelist_blueprint(1000)->estimate(), 0.1);
+    EXPECT_EQ(make_whitelist_blueprint(200)->estimate(), 0.5);
+    EXPECT_EQ(make_whitelist_blueprint(5)->estimate(), 1.0);
+}
+
 class LidAllocatorPerformanceTest : public LidAllocatorTest,
                                     public testing::WithParamInterface<bool>
 {
diff --git a/searchcore/src/vespa/searchcore/proton/documentmetastore/lid_allocator.cpp b/searchcore/src/vespa/searchcore/proton/documentmetastore/lid_allocator.cpp
index 95e4eac437c..9e9199bc8ba 100644
--- a/searchcore/src/vespa/searchcore/proton/documentmetastore/lid_allocator.cpp
+++ b/searchcore/src/vespa/searchcore/proton/documentmetastore/lid_allocator.cpp
@@ -222,6 +222,10 @@ public:
         setEstimate(HitEstimate(_activeLids.size(), false));
     }
 
+    double calculate_relative_estimate() const final {
+        return abs_to_rel_est(getState().estimate().estHits, get_docid_limit());
+    }
+
     bool isWhiteList() const noexcept final { return true; }
 
     SearchIterator::UP createFilterSearch(bool strict, FilterConstraint) const override {
-- 
cgit v1.2.3