refactor: extract WordRefList and thread finder

main
mandlm 2024-03-21 09:57:13 +01:00
parent 48f05ddb4f
commit 95cc7223e8
Signed by: mandlm
GPG Key ID: 4AA25D647AA54CC7
8 changed files with 84 additions and 63 deletions

View File

@ -4,15 +4,14 @@
#include "word_list.h" #include "word_list.h"
#include <map> #include <map>
#include <vector>
class GroupedFinder : public Finder { class GroupedFinder : public Finder {
private: private:
std::map<char, std::vector<const std::string *>> groups_; std::map<char, WordRefList> groups_;
public: public:
GroupedFinder(const WordList &word_list); GroupedFinder(const WordList &word_list);
virtual std::forward_list<const std::string *> virtual std::forward_list<const std::string *>
find_prefix(std::string_view search_term) const override; find_prefix(std::string_view search_prefix) const override;
}; };

View File

@ -11,5 +11,5 @@ public:
ParallelFinder(const WordList &word_list); ParallelFinder(const WordList &word_list);
std::forward_list<const std::string *> std::forward_list<const std::string *>
find_prefix(std::string_view search_term) const override; find_prefix(std::string_view search_prefix) const override;
}; };

View File

@ -3,11 +3,9 @@
#include "finder.h" #include "finder.h"
#include "word_list.h" #include "word_list.h"
#include <vector>
class SortedLinearFinder : public Finder { class SortedLinearFinder : public Finder {
private: private:
std::vector<const std::string *> word_list_; WordRefList word_list_;
public: public:
SortedLinearFinder(const WordList &word_list); SortedLinearFinder(const WordList &word_list);

View File

@ -1,6 +1,8 @@
#pragma once #pragma once
#include <filesystem> #include <filesystem>
#include <forward_list>
#include <mutex>
#include <string> #include <string>
#include <vector> #include <vector>
@ -11,4 +13,17 @@ public:
static WordList fourCaps(); static WordList fourCaps();
static WordList fromFile(const std::filesystem::path &path); static WordList fromFile(const std::filesystem::path &path);
static void find_prefix_in_range(
const WordList &word_list, const std::string_view &search_prefix,
size_t start_index, size_t end_index,
std::forward_list<const std::string *> &result, std::mutex &result_mutex);
};
class WordRefList : public std::vector<const std::string *> {
public:
static void find_prefix_in_range(
const WordRefList &word_list, const std::string_view &search_prefix,
size_t start_index, size_t end_index,
std::forward_list<const std::string *> &result, std::mutex &result_mutex);
}; };

View File

@ -5,8 +5,8 @@
#include <thread> #include <thread>
#include <vector> #include <vector>
using std::mutex, std::vector, std::thread, std::lock_guard, std::string, using std::mutex, std::vector, std::thread, std::string, std::forward_list,
std::forward_list, std::string_view; std::string_view;
GroupedFinder::GroupedFinder(const WordList &word_list) { GroupedFinder::GroupedFinder(const WordList &word_list) {
for (const auto &word : word_list) { for (const auto &word : word_list) {
@ -15,8 +15,8 @@ GroupedFinder::GroupedFinder(const WordList &word_list) {
} }
std::forward_list<const std::string *> std::forward_list<const std::string *>
GroupedFinder::find_prefix(std::string_view search_term) const { GroupedFinder::find_prefix(std::string_view search_prefix) const {
const auto group = groups_.find(search_term.front()); const auto group = groups_.find(search_prefix.front());
if (group == groups_.cend()) { if (group == groups_.cend()) {
return {}; return {};
} }
@ -27,41 +27,26 @@ GroupedFinder::find_prefix(std::string_view search_term) const {
const auto thread_count = const auto thread_count =
std::min<size_t>(std::thread::hardware_concurrency(), word_list_size); std::min<size_t>(std::thread::hardware_concurrency(), word_list_size);
forward_list<const string *> matching_words; forward_list<const string *> result;
mutex matching_words_mutex; mutex result_mutex;
vector<thread> search_threads; vector<thread> search_threads;
for (size_t thread_index = 0; thread_index < thread_count; ++thread_index) { for (size_t thread_index = 0; thread_index < thread_count; ++thread_index) {
const size_t first_word_index = const size_t first_index = thread_index * (word_list_size / thread_count);
thread_index * (word_list_size / thread_count);
const size_t last_word_index = const size_t last_index =
(thread_index == thread_count - 1) (thread_index == thread_count - 1)
? word_list_size ? word_list_size
: (thread_index + 1) * (word_list_size / thread_count); : (thread_index + 1) * (word_list_size / thread_count);
search_threads.emplace_back( search_threads.emplace_back(
[](const vector<const string *> &word_list, WordRefList::find_prefix_in_range, cref(word_list), cref(search_prefix),
const string_view &search_term, forward_list<const string *> &result, first_index, last_index, ref(result), ref(result_mutex));
size_t start_index, size_t end_index, mutex &result_mutex) {
forward_list<const string *> thread_results;
for (size_t index = start_index; index < end_index; ++index) {
const auto &current_word = word_list[index];
if (current_word->starts_with(search_term)) {
thread_results.push_front(current_word);
}
}
if (!thread_results.empty()) {
const lock_guard<mutex> lock(result_mutex);
result.merge(thread_results);
}
},
cref(word_list), cref(search_term), ref(matching_words),
first_word_index, last_word_index, ref(matching_words_mutex));
} }
for (auto &thread : search_threads) { for (auto &thread : search_threads) {
thread.join(); thread.join();
} }
return matching_words; return result;
} }

View File

@ -3,51 +3,36 @@
#include <mutex> #include <mutex>
#include <thread> #include <thread>
using std::mutex, std::thread, std::lock_guard, std::vector, std::forward_list, using std::mutex, std::thread, std::vector, std::forward_list, std::string,
std::string, std::string_view; std::string_view;
ParallelFinder::ParallelFinder(const WordList &word_list) ParallelFinder::ParallelFinder(const WordList &word_list)
: word_list_(word_list) {} : word_list_(word_list) {}
forward_list<const string *> forward_list<const string *>
ParallelFinder::find_prefix(string_view search_term) const { ParallelFinder::find_prefix(string_view search_prefix) const {
forward_list<const string *> result; forward_list<const string *> result;
mutex result_mutex; mutex result_mutex;
const auto word_list_size = word_list_.size(); const size_t word_list_size = word_list_.size();
const size_t thread_count =
std::min<size_t>(thread::hardware_concurrency(), word_list_size);
const size_t thread_count = thread::hardware_concurrency(); vector<thread> search_threads;
vector<thread> threads;
for (size_t thread_index = 0; thread_index < thread_count; ++thread_index) { for (size_t thread_index = 0; thread_index < thread_count; ++thread_index) {
const size_t first_word_index = const size_t first_index = thread_index * (word_list_size / thread_count);
thread_index * (word_list_size / thread_count);
const size_t last_word_index = const size_t last_index =
(thread_index == thread_count - 1) (thread_index == thread_count - 1)
? word_list_size ? word_list_size
: (thread_index + 1) * (word_list_size / thread_count); : (thread_index + 1) * (word_list_size / thread_count);
threads.emplace_back( search_threads.emplace_back(
[](const WordList &word_list, const string_view &search_term, WordList::find_prefix_in_range, cref(word_list_), cref(search_prefix),
forward_list<const string *> &result, size_t start_index, first_index, last_index, ref(result), ref(result_mutex));
size_t end_index, mutex &result_mutex) {
forward_list<const string *> thread_results;
for (size_t index = start_index; index < end_index; ++index) {
const auto &current_word = word_list[index];
if (current_word.starts_with(search_term)) {
thread_results.push_front(&current_word);
}
}
if (!thread_results.empty()) {
const lock_guard<mutex> lock(result_mutex);
result.merge(thread_results);
}
},
cref(word_list_), cref(search_term), ref(result), first_word_index,
last_word_index, ref(result_mutex));
} }
for (auto &thread : threads) { for (auto &thread : search_threads) {
thread.join(); thread.join();
} }

View File

@ -14,6 +14,7 @@ SortedLinearFinder::SortedLinearFinder(const WordList &word_list) {
word_list_.begin(), word_list_.end(), word_list_.begin(), word_list_.end(),
[](const string *left, const string *right) { return *left < *right; }); [](const string *left, const string *right) { return *left < *right; });
} }
forward_list<const string *> forward_list<const string *>
SortedLinearFinder::find_prefix(string_view search_term) const { SortedLinearFinder::find_prefix(string_view search_term) const {
forward_list<const string *> matching_words; forward_list<const string *> matching_words;

View File

@ -57,3 +57,41 @@ WordList WordList::fromFile(const std::filesystem::path &path) {
return word_list; return word_list;
} }
void WordList::find_prefix_in_range(
const WordList &word_list, const std::string_view &search_prefix,
size_t start_index, size_t end_index,
std::forward_list<const std::string *> &result, std::mutex &result_mutex) {
std::forward_list<const std::string *> local_results;
for (size_t index = start_index; index < end_index; ++index) {
const auto &current_word = word_list[index];
if (current_word.starts_with(search_prefix)) {
local_results.push_front(&current_word);
}
}
if (!local_results.empty()) {
const std::lock_guard<std::mutex> lock(result_mutex);
result.merge(local_results);
}
};
void WordRefList::find_prefix_in_range(
const WordRefList &word_list, const std::string_view &search_prefix,
size_t start_index, size_t end_index,
std::forward_list<const std::string *> &result, std::mutex &result_mutex) {
std::forward_list<const std::string *> local_results;
for (size_t index = start_index; index < end_index; ++index) {
const auto *current_word = word_list[index];
if (current_word->starts_with(search_prefix)) {
local_results.push_front(current_word);
}
}
if (!local_results.empty()) {
const std::lock_guard<std::mutex> lock(result_mutex);
result.merge(local_results);
}
};