Function LoadPools

catboost/app/mode_select_features.cpp:72–113

static TDataProviders LoadPools(
    const TCatBoostOptions& catBoostOptions,
    const TPoolLoadParams& poolLoadParams,
    TOption<bool>* hasTimeFlag,
    NPar::ILocalExecutor* executor
) {
    TVector<NJson::TJsonValue> classLabels = catBoostOptions.DataProcessingOptions->ClassLabels;
    const auto objectsOrder = hasTimeFlag->Get() ? EObjectsOrder::Ordered : EObjectsOrder::Undefined;
    CB_ENSURE(poolLoadParams.TestSetPaths.size() <= 1, "Features selection mode doesn't support several eval sets.");
    const bool haveLearnFeaturesInMemory = HaveFeaturesInMemory(
        catBoostOptions,
        poolLoadParams.LearnSetPath);
    TVector<TDatasetSubset> testDatasetSubsets;
    for (const auto& testSetPath : poolLoadParams.TestSetPaths) {
        testDatasetSubsets.push_back(
            TDatasetSubset::MakeColumns(HaveFeaturesInMemory(catBoostOptions, testSetPath)));
    }
    auto pools = NCB::ReadTrainDatasets(
        catBoostOptions.GetTaskType(),
        poolLoadParams,
        objectsOrder,
        /*readTestData*/true,
        TDatasetSubset::MakeColumns(haveLearnFeaturesInMemory),
        testDatasetSubsets,
        catBoostOptions.DataProcessingOptions->ForceUnitAutoPairWeights,
        &classLabels,
        executor,
        /*profile*/nullptr
    );
    CB_ENSURE(pools.Learn != nullptr, "Train data must be provided");

    const auto learnDataOrder = pools.Learn->ObjectsData->GetOrder();
    if (learnDataOrder == EObjectsOrder::Ordered) {
        *hasTimeFlag = true;
    }
    pools.Learn = ReorderByTimestampLearnDataIfNeeded(catBoostOptions, pools.Learn, executor);

    TRestorableFastRng64 rand(catBoostOptions.RandomSeed);
    pools.Learn = ShuffleLearnDataIfNeeded(catBoostOptions, pools.Learn, executor, &rand);

    return pools;
}


void SaveSummaryToFile(const TFeaturesSelectionSummary& summary, const TString& resultPath) {