Method scanInternal

src/storage/table/arrow_node_table.cpp:82–136

First run always fails due to arrowScanState.scanCompleted == true because either scanState.source = NONE or scanState.currentBatchIdx = INVALID_NODE_GROUP_IDX on the first run(look at initScanState function) tableScanSharedState.nextMorsel will drive scanInternal completely

// run(look at initScanState function) tableScanSharedState.nextMorsel will drive scanInternal
// completely
bool ArrowNodeTable::scanInternal([[maybe_unused]] transaction::Transaction* transaction,
    TableScanState& scanState) {
    auto& arrowScanState = scanState.cast<ArrowNodeTableScanState>();
    if (arrowScanState.scanCompleted) {
        return false;
    }

    if (arrowScanState.currentBatchIdx >= arrays.size() ||
        arrowScanState.currentMorselStartOffset >= arrowScanState.currentMorselEndOffset) {
        arrowScanState.scanCompleted = true;
        return false;
    }

    const auto& batch = arrays[arrowScanState.currentBatchIdx];
    auto batchLength = getArrowBatchLength(batch);

    if (batchLength == 0 || !batch.children || !schema.children || batch.n_children <= 0) {
        arrowScanState.scanCompleted = true;
        return false;
    }

    scanState.resetOutVectors();

    // Calculate the size of the current morsel
    auto morselStart = arrowScanState.currentMorselStartOffset;
    auto morselEnd = std::min((uint64_t)arrowScanState.currentMorselEndOffset, batchLength);
    auto outputSize = static_cast<uint64_t>(morselEnd - morselStart);

    auto nextGlobalRowOffset = batchStartOffsets[arrowScanState.currentBatchIdx] + morselStart;

    scanState.outState->getSelVectorUnsafe().setSelSize(outputSize);

    NodeTable::applySemiMaskFilter(scanState, nextGlobalRowOffset, outputSize,
        scanState.outState->getSelVectorUnsafe());

    if (scanState.outState->getSelVector().getSelSize() == 0) {
        return false;
    }

    const auto outputToArrowColumnIdx = getOutputToArrowColumnIdx(scanState.columnIDs);
    DASSERT(scanState.outputVectors.size() == outputToArrowColumnIdx.size());
    copyArrowMorselToOutputVectors(batch, arrowScanState.currentMorselStartOffset, outputSize,
        scanState.outputVectors, outputToArrowColumnIdx);

    auto tableID = this->getTableID();
    for (uint64_t i = 0; i < outputSize; ++i) {
        auto& nodeID = scanState.nodeIDVector->getValue<common::nodeID_t>(i);
        nodeID.tableID = tableID;
        nodeID.offset = nextGlobalRowOffset + i;
    }

    arrowScanState.currentMorselStartOffset += outputSize;

    return true;
}

common::node_group_idx_t ArrowNodeTable::getNumBatches(
    [[maybe_unused]] const transaction::Transaction* transaction) const {