85 lines
2.2 KiB
C++
85 lines
2.2 KiB
C++
#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 ¤t_word = word_list[index];
|
|
groups_[current_word.front()].push_back(¤t_word);
|
|
}
|
|
}
|
|
|
|
WordRefList Bucket::find_prefix(std::string_view search_term) const {
|
|
auto group_it = groups_.find(search_term.front());
|
|
if (group_it == groups_.cend()) {
|
|
return {};
|
|
}
|
|
|
|
WordRefList result;
|
|
for (const auto *word : group_it->second) {
|
|
if (word->starts_with(search_term)) {
|
|
result.push_back(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> threads;
|
|
for (auto bucket_index = 0; bucket_index < bucket_count; ++bucket_index) {
|
|
auto &bucket = buckets_[bucket_index];
|
|
|
|
bool is_last_bucket = bucket_index == bucket_count - 1;
|
|
|
|
const size_t first_index = bucket_index * bucket_size;
|
|
const size_t last_index =
|
|
is_last_bucket ? word_list_size : first_index + bucket_size;
|
|
|
|
threads.emplace_back([&, first_index, last_index] {
|
|
bucket.insert(word_list, first_index, last_index);
|
|
});
|
|
}
|
|
|
|
for (auto &thread : threads) {
|
|
thread.join();
|
|
}
|
|
}
|
|
|
|
WordRefList BucketFinder::find_prefix(std::string_view search_term) const {
|
|
WordRefList search_results;
|
|
std::mutex search_results_mutex;
|
|
|
|
std::vector<std::thread> threads;
|
|
for (const auto &bucket : buckets_) {
|
|
threads.emplace_back([&] {
|
|
auto thread_search_results = bucket.find_prefix(search_term);
|
|
if (!thread_search_results.empty()) {
|
|
std::lock_guard result_lock(search_results_mutex);
|
|
std::move(thread_search_results.begin(), thread_search_results.end(),
|
|
std::back_inserter(search_results));
|
|
}
|
|
});
|
|
}
|
|
|
|
for (auto &thread : threads) {
|
|
thread.join();
|
|
}
|
|
|
|
return search_results;
|
|
};
|