Method scan_sample_next_tuple

cpp/deeplake_pg/table_am.cpp:909–950

}

bool deeplake_table_am_routine::scan_sample_next_tuple(TableScanDesc scan,
                                                       struct SampleScanState* scanstate,
                                                       TupleTableSlot* slot)
{
    CHECK_FOR_INTERRUPTS();

    DeeplakeScanData* scan_data = get_scan_data(scan);
    if (!scan_data || !scan_data->sample_scan_active) {
        return false;
    }

    // Reset memory context to prevent unbounded growth
    if (scan_data->scan_state.get_current_position() % num_tuples_to_reset_memory_context == 0) {
        MemoryContextReset(scan_data->memory_context);
    }

    // Switch to the dedicated memory context for this scan
    pg::utils::memory_context_switcher context_switcher(scan_data->memory_context);

    constexpr double DEFAULT_SAMPLE_FRACTION = 0.1;   // Default sample fraction
    double sample_fraction = DEFAULT_SAMPLE_FRACTION; // default fallback
    if (scan_data->current_sample_scanstate && scan_data->current_sample_scanstate->tsm_state) {
        // If your TAM stores the fraction in tsm_state, cast and read it here
        struct SampleState
        {
            double fraction;
        };
        SampleState* sampler = static_cast<SampleState*>(scan_data->current_sample_scanstate->tsm_state);
        sample_fraction = sampler->fraction;
    }

    while (scan_data->scan_state.get_next_tuple(slot)) {
        // random fraction, should come from scanstate->tsm_state
        const double random_value = (double)random() / RAND_MAX;

        if (random_value <= sample_fraction) {
            return true;
        }
    }

    return false;
}

IndexFetchTableData* deeplake_table_am_routine::begin_index_fetch(Relation rel)
{