Function buildDenseIndptr

src/main/query_result/arrow_query_result.cpp:83–101

Build a dense global indptr of size numSourceRows+1 from sparse (srcRows, counts) runs. indptr[src+1] is set to counts[i] for the touched src = srcRows[i], then prefix-summed so indptr[src] gives the offset of src's edges in the (source-sorted) indices vector. Source rows absent from srcRows keep their slot at 0, i.e. no edges. Returns an empty vector on validation failure (bad src/count, or the d

// disjointness invariant is violated — a source row appearing more than
// once would silently corrupt the merged CSR).
static std::vector<int64_t> buildDenseIndptr(int64_t numSourceRows,
    const std::vector<int64_t>& srcRows, const std::vector<int64_t>& counts) {
    std::vector<int64_t> indptr(static_cast<size_t>(numSourceRows) + 1, 0);
    for (auto i = 0u; i < srcRows.size(); ++i) {
        const auto src = srcRows[i];
        const auto count = counts[i];
        if (src < 0 || src >= numSourceRows || count < 0) {
            return {};
        }
        if (indptr[static_cast<size_t>(src) + 1] != 0) {
            return {};
        }
        indptr[static_cast<size_t>(src) + 1] = count;
    }
    for (size_t i = 0; i + 1 < indptr.size(); ++i) {
        indptr[i + 1] += indptr[i];
    }
    return indptr;
}

// K-way merge of per-batch sparse CSR metadata chunks (in batch_index
// order) into a single flat CSRMetadata with a dense global indptr.