Function BuildBuckets

library/cpp/text_processing/dictionary/mmap_hash_table.h:53–79

    template <typename TElementRange, typename TExctractor>
    void BuildBuckets(
        const TElementRange& range,
        const TExctractor& getTokenInfo,
        TVector<TBucket>* buckets,
        ui64* seed
    ) {
        const ui32 bucketCount = ComputeCorrectBucketCount(range.size());
        *seed = 0;
        buckets->yresize(bucketCount);
        for (; *seed < MAX_SEED_CHOICE_COUNT; ++(*seed)) {
            bool isBucketSearchStepLimitExceeded = false;
            std::fill(buckets->begin(), buckets->end(), TBucket());
            for (const auto& element : range) {
                const auto hashAndTokenId = getTokenInfo(element, *seed);
                ui32 bucketSearchStepCount;
                auto index = GetBucketIndex(hashAndTokenId.first, *buckets, &bucketSearchStepCount);
                (*buckets)[index] = {hashAndTokenId.first, hashAndTokenId.second};
                isBucketSearchStepLimitExceeded |= bucketSearchStepCount > MAX_BUCKET_SEARCH_STEPS;
            }

            if (!isBucketSearchStepLimitExceeded) {
                return;
            }
        }

        Y_ENSURE(false, "Couldn't find a mapping without collisions.");
    }
}