VectorSearch/lib_vector_search/src/bucket_finder.cpp

#include "bucket_finder.h"

#include <mutex>
#include <strings.h>
#include <thread>

void Bucket::insert(const WordList &word_list, size_t first_index,
                    size_t last_index) {
  for (size_t index = first_index; index < last_index; ++index) {
    const auto &current_word = word_list[index];
    groups_[current_word.front()].push_back(&current_word);
  }
}

std::forward_list<const std::string *>
Bucket::find_prefix(std::string_view search_term) const {
  const auto group = groups_.find(search_term.front());
  if (group == groups_.cend()) {
    return {};
  }

  std::forward_list<const std::string *> result;
  for (const auto *word : group->second) {
    if (word->starts_with(search_term)) {
      result.push_front(word);
    }
  }
  return result;
}

BucketFinder::BucketFinder(const WordList &word_list) {
  if (word_list.empty()) {
    return;
  }

  const size_t word_list_size = word_list.size();
  const size_t bucket_count =
      std::min<size_t>(std::thread::hardware_concurrency(), word_list_size);
  const size_t bucket_size = word_list_size / bucket_count;

  buckets_.resize(bucket_count);

  std::vector<std::thread> insert_threads;
  for (auto bucket_index = 0; bucket_index < bucket_count; ++bucket_index) {
    const size_t first_index = bucket_index * bucket_size;
    const size_t last_index = (bucket_index == bucket_count - 1)
                                  ? word_list_size
                                  : (bucket_index + 1) * bucket_size;

    auto &bucket = buckets_[bucket_index];

    insert_threads.emplace_back([&bucket, &word_list, first_index, last_index] {
      bucket.insert(word_list, first_index, last_index);
    });
  }

  for (auto &thread : insert_threads) {
    thread.join();
  }
}

std::forward_list<const std::string *>
BucketFinder::find_prefix(std::string_view search_term) const {
  std::forward_list<const std::string *> result;
  std::mutex result_mutex;

  std::vector<std::thread> search_threads;
  for (const auto &bucket : buckets_) {

    search_threads.emplace_back([&] {
      auto thread_result = bucket.find_prefix(search_term);
      if (!thread_result.empty()) {
        std::lock_guard<std::mutex> result_lock(result_mutex);
        result.merge(thread_result);
      }
    });
  }

  for (auto &thread : search_threads) {
    thread.join();
  }

  return result;
};