Add unit tests and update a lot of things.

Signed-off-by: Johannes Kalmbach <[email protected]>

src/index/LocatedTriples.cpp

@@ -251,22 +251,21 @@ void LocatedTriplesPerBlock::setOriginalMetadata(
   updateAugmentedMetadata();
 }
 
-// Update the `blockMetadata`, such that its metadata pertaining to the
-// contained graphs is consistent with the `locatedTriples` which are added to
-// that block. In particular, all graphs to which at least one triple is added
-// is added to the graph metadata, and if the number of total graphs is larger
-// than the configured threshold, then the graph metadata is set to `nullopt`,
-// which means that there is no metadata.
+// Update the `blockMetadata`, such that its graph info is consistent with the
+// `locatedTriples` which are added to that block. In particular, all graphs to
+// which at least one triple is inserted become part of the graph info, and if
+// the number of total graphs becomes larger than the configured threshold, then
+// the graph info is set to `nullopt`, which means that there is no info.
 static auto updateGraphMetadata(CompressedBlockMetadata& blockMetadata,
                                 const LocatedTriples& locatedTriples) {
   // We do not know anything about the triples contained in the block, so we
-  // also Cannot know if the `locatedTriples` introduces duplicates. We thus
+  // also cannot know if the `locatedTriples` introduces duplicates. We thus
   // have to be conservative and assume that there are duplicates.
   blockMetadata.containsDuplicatesWithDifferentGraphs_ = true;
   auto& graphs = blockMetadata.graphInfo_;
   if (!graphs.has_value()) {
     // The original block already contains too many graphs, don't store any
-    // metadata.
+    // graph info.
     return;
   }
 
@@ -281,14 +280,16 @@ static auto updateGraphMetadata(CompressedBlockMetadata& blockMetadata,
     }
     newGraphs.insert(lt.triple_.ids_.at(ADDITIONAL_COLUMN_GRAPH_ID));
     // Handle the case that with the newly added triples we have too many
-    // distinct graphs to store them in the metadata.
+    // distinct graphs to store them in the graph info.
     if (newGraphs.size() > MAX_NUM_GRAPHS_STORED_IN_BLOCK_METADATA) {
       graphs.reset();
       return;
     }
   }
   graphs.emplace(newGraphs.begin(), newGraphs.end());
-  // Sort the stored graphs, to make the testing easier.
+
+  // Sort the stored graphs. Note: this is currently not expected by the code
+  // that uses the graph info, but makes testing much easier.
   std::ranges::sort(graphs.value());
 }
 

test/LocatedTriplesTest.cpp

@@ -886,9 +886,25 @@ TEST_F(LocatedTriplesTest, augmentedMetadataGraphInfo) {
 
     // Note: the GraphInfo hasn't changed, because the new triples all were
     // deleted.
-    auto m = locatedTriplesPerBlock.getAugmentedMetadata();
-    EXPECT_THAT(locatedTriplesPerBlock.getAugmentedMetadata(),
+    auto actualMetadata = locatedTriplesPerBlock.getAugmentedMetadata();
+    EXPECT_THAT(actualMetadata,
                 testing::ElementsAreArray(expectedAugmentedMetadata));
+
+    // Test the case that a block loses its graph info if the added located
+    // triples have too many distinct graphs.
+    ASSERT_TRUE(actualMetadata[1].graphInfo_.has_value());
+    std::vector<IdTriple<0>> triples;
+    for (size_t i = 30; i < 30 + 2 * MAX_NUM_GRAPHS_STORED_IN_BLOCK_METADATA;
+         ++i) {
+      auto tr = T3;
+      tr.ids_.at(ADDITIONAL_COLUMN_GRAPH_ID) = V(i);
+      triples.push_back(tr);
+    }
+    // Add the located triples.
+    locatedTriplesPerBlock.add(LocatedTriple::locateTriplesInPermutation(
+        triples, metadata, {0, 1, 2}, true, handle));
+    actualMetadata = locatedTriplesPerBlock.getAugmentedMetadata();
+    ASSERT_FALSE(actualMetadata[1].graphInfo_.has_value());
   }
 }
 

test/PrefilterExpressionIndexTest.cpp

@@ -157,21 +157,32 @@ class PrefilterExpressionOnMetadataTest : public ::testing::Test {
 
 //______________________________________________________________________________
 TEST_F(PrefilterExpressionOnMetadataTest, testBlockFormatForDebugging) {
+  auto toString = [](const CompressedBlockMetadata& b) {
+    return (std::stringstream{} << b).str();
+  };
+
+  auto matcher = [&toString](const std::string& substring) {
+    return ::testing::ResultOf(toString, ::testing::HasSubstr(substring));
+  };
   EXPECT_THAT(
-      (std::stringstream() << b5).str(),
-      ::testing::HasSubstr(
+      b5,
+      matcher(
           "#BlockMetadata\n(first) Triple: I:0 V:10 D:33.000000 V:0\n(last) "
           "Triple: I:0 V:10 D:33.000000 V:0\nnum. rows: 0.\n"));
   EXPECT_THAT(
-      (std::stringstream() << b11).str(),
-      ::testing::HasSubstr(
+      b11,
+      matcher(
           "#BlockMetadata\n(first) Triple: I:-4 V:10 D:33.000000 V:0\n(last) "
           "Triple: D:2.000000 V:10 D:33.000000 V:0\nnum. rows: 0.\n"));
   EXPECT_THAT(
-      (std::stringstream() << b21).str(),
-      ::testing::HasSubstr(
+      b21,
+      matcher(
           "#BlockMetadata\n(first) Triple: V:14 V:10 D:33.000000 V:0\n(last) "
           "Triple: V:17 V:10 D:33.000000 V:0\nnum. rows: 0.\n"));
+
+  auto blockWithGraphInfo = b21;
+  blockWithGraphInfo.graphInfo_.emplace({IntId(12), IntId(13)});
+  EXPECT_THAT(blockWithGraphInfo, matcher("Graphs: I:12, I:13\n"));
 }
 
 // Test Relational Expressions