fix list reader bug

kuzudb · Sep 26, 2023 · e34483c · e34483c
1 parent 5cbb422
commit e34483c
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 10 deletions.
diff --git a/src/include/storage/store/var_list_node_column.h b/src/include/storage/store/var_list_node_column.h
@@ -27,15 +27,17 @@ namespace storage {
 
 struct ListOffsetInfoInStorage {
     common::offset_t prevNodeListOffset;
-    std::unique_ptr<common::ValueVector> offsetVector;
+    std::vector<std::unique_ptr<common::ValueVector>> offsetVectors;
 
-    ListOffsetInfoInStorage(
-        common::offset_t prevNodeListOffset, std::unique_ptr<common::ValueVector> offsetVector)
-        : prevNodeListOffset{prevNodeListOffset}, offsetVector{std::move(offsetVector)} {}
+    ListOffsetInfoInStorage(common::offset_t prevNodeListOffset,
+        std::vector<std::unique_ptr<common::ValueVector>> offsetVectors)
+        : prevNodeListOffset{prevNodeListOffset}, offsetVectors{std::move(offsetVectors)} {}
 
     inline common::offset_t getListOffset(uint64_t nodePos) const {
+        auto offsetVector = offsetVectors[(nodePos - 1) / common::DEFAULT_VECTOR_CAPACITY].get();
         return nodePos == 0 ? prevNodeListOffset :
-                              offsetVector->getValue<common::offset_t>(nodePos - 1);
+                              offsetVector->getValue<common::offset_t>(
+                                  (nodePos - 1) % common::DEFAULT_VECTOR_CAPACITY);
     }
 
     inline uint64_t getListLength(uint64_t nodePos) const {

diff --git a/src/storage/store/var_list_node_column.cpp b/src/storage/store/var_list_node_column.cpp
@@ -151,13 +151,26 @@ offset_t VarListNodeColumn::readOffset(
 ListOffsetInfoInStorage VarListNodeColumn::getListOffsetInfoInStorage(Transaction* transaction,
     node_group_idx_t nodeGroupIdx, offset_t startOffsetInNodeGroup, offset_t endOffsetInNodeGroup,
     std::shared_ptr<DataChunkState> state) {
-    auto offsetVector = std::make_unique<ValueVector>(LogicalTypeID::INT64);
-    offsetVector->setState(std::move(state));
-    NodeColumn::scan(transaction, nodeGroupIdx, startOffsetInNodeGroup, endOffsetInNodeGroup,
-        offsetVector.get());
+    auto numOffsetsToRead = endOffsetInNodeGroup - startOffsetInNodeGroup;
+    auto numOffsetVectors = numOffsetsToRead / DEFAULT_VECTOR_CAPACITY +
+                            (numOffsetsToRead % DEFAULT_VECTOR_CAPACITY ? 1 : 0);
+    std::vector<std::unique_ptr<ValueVector>> offsetVectors;
+    offsetVectors.reserve(numOffsetVectors);
+    uint64_t numOffsetsRead = 0;
+    for (auto i = 0u; i < numOffsetVectors; i++) {
+        auto offsetVector = std::make_unique<ValueVector>(LogicalTypeID::INT64);
+        auto numOffsetsToReadInCurBatch =
+            std::min(numOffsetsToRead - numOffsetsRead, DEFAULT_VECTOR_CAPACITY);
+        offsetVector->setState(state);
+        NodeColumn::scan(transaction, nodeGroupIdx, startOffsetInNodeGroup + numOffsetsRead,
+            startOffsetInNodeGroup + numOffsetsRead + numOffsetsToReadInCurBatch,
+            offsetVector.get());
+        offsetVectors.push_back(std::move(offsetVector));
+        numOffsetsRead += numOffsetsToReadInCurBatch;
+    }
     auto prevNodeListOffsetInStorage =
         readListOffsetInStorage(transaction, nodeGroupIdx, startOffsetInNodeGroup);
-    return {prevNodeListOffsetInStorage, std::move(offsetVector)};
+    return {prevNodeListOffsetInStorage, std::move(offsetVectors)};
 }
 
 } // namespace storage