Method DataProviderSamplerReorderByIndices

catboost/libs/data/sampler.cpp:108–135

TDataProviderPtr NCB::DataProviderSamplerReorderByIndices(
    const TDataProviderSampleParams& params,
    TDataProviderPtr dataProvider,
    TConstArrayRef<ui32> indices
) {
    // resort dataset according to original indices
    TVector<ui32> currentOrder(indices.size());
    Iota(currentOrder.begin(), currentOrder.end(), ui32(0));
    Sort(currentOrder, [&](ui32 l, ui32 r) { return indices[l] < indices[r]; });
    TVector<ui32> finalOrder(indices.size());
    for (size_t i : xrange(indices.size())) {
        finalOrder[currentOrder[i]] = i;
    }

    if (params.OnlyFeaturesData) {
        DegroupDataset(dataProvider.Get());
    }

    return dataProvider->GetSubset(
        GetGroupingSubsetFromObjectsSubset(
            dataProvider->ObjectsGrouping,
            finalOrder,
            EObjectsOrder::RandomShuffled
        ),
        params.CpuUsedRamLimit,
        params.LocalExecutor
    );
}


TDataProviderPtr NCB::LinesFileSampleByIndices(const TDataProviderSampleParams& params, TConstArrayRef<ui32> indices) {