Method getNextTuplesInternal

src/processor/operator/scan/count_rel_table.cpp:33–140

Count rels by using CSR metadata, accounting for deletions and uncommitted data. This is more efficient than scanning through all edges.

// Count rels by using CSR metadata, accounting for deletions and uncommitted data.
// This is more efficient than scanning through all edges.
bool CountRelTable::getNextTuplesInternal(ExecutionContext* context) {
    if (hasExecuted) {
        return false;
    }

    auto transaction = Transaction::Get(*context->clientContext);
    auto* memoryManager = context->clientContext->getDatabase()->getMemoryManager();

    for (auto* relTable : relTables) {
        if (dynamic_cast<ColumnarRelTableBase*>(relTable) != nullptr) {
            totalCount += relTable->getNumTotalRows(transaction);
            continue;
        }

        // Get the RelTableData for the specified direction
        auto* relTableData = relTable->getDirectedTableData(direction);
        auto numNodeGroups = relTableData->getNumNodeGroups();
        auto* csrLengthColumn = relTableData->getCSRLengthColumn();

        // For each node group in the rel table
        for (node_group_idx_t nodeGroupIdx = 0; nodeGroupIdx < numNodeGroups; nodeGroupIdx++) {
            auto* nodeGroup = relTableData->getNodeGroup(nodeGroupIdx);
            if (!nodeGroup) {
                continue;
            }

            auto& csrNodeGroup = nodeGroup->cast<CSRNodeGroup>();

            // Count from persistent (checkpointed) data
            if (auto* persistentGroup = csrNodeGroup.getPersistentChunkedGroup()) {
                // Sum the actual relationship lengths from the CSR header instead of using
                // getNumRows() which includes dummy rows added for CSR offset array gaps
                auto& csrPersistentGroup = persistentGroup->cast<ChunkedCSRNodeGroup>();
                auto& csrHeader = csrPersistentGroup.getCSRHeader();

                // Get the number of nodes in this CSR header
                auto numNodes = csrHeader.length->getNumValues();
                if (numNodes == 0) {
                    continue;
                }

                // Create an in-memory chunk to scan the CSR length column into
                auto lengthChunk =
                    ColumnChunkFactory::createColumnChunkData(*memoryManager, LogicalType::UINT64(),
                        false /*enableCompression*/, StorageConfig::NODE_GROUP_SIZE,
                        ResidencyState::IN_MEMORY, false /*initializeToZero*/);

                // Initialize scan state and scan the length column from disk
                ChunkState chunkState;
                csrHeader.length->initializeScanState(chunkState, csrLengthColumn);
                csrLengthColumn->scan(chunkState, lengthChunk.get(), 0 /*offsetInChunk*/, numNodes);

                // Sum all the lengths
                auto* lengthData = reinterpret_cast<const uint64_t*>(lengthChunk->getData());
                row_idx_t groupRelCount = 0;
                for (offset_t i = 0; i < numNodes; ++i) {
                    groupRelCount += lengthData[i];
                }