doc: update README

README.md

@@ -1,9 +1,31 @@
-## 参考资料
-1. [Lucene原理](https://www.infoq.cn/article/ejeg02vroegvalw4j_ll)
+<div align="center">
+  <h1>OurSearchEngine</code></h1>
 
+  <p>
+    <strong>A simple search engine written in <code>Cpp</code></strong>
+  </p>
+</div>
 
-## Requires
+## Features
+- 
+
+
+## Architecture
+![Architecture](doc/pic/Arch.png)
+
+
+## How to use
+### Build
 ```bash
-yay -S jsoncpp-cmake
+cd OurSearchEngine
 git submodule update --init --recursive
-```
+cmake -B build
+cmake --build build
+```
+
+### Use
+
+
+
+## Reference
+1. [Lucene](https://www.infoq.cn/article/ejeg02vroegvalw4j_ll)

src/backends/src/ServerCtl.cpp

@@ -21,8 +21,8 @@ void ServerCtl::search(const drogon::HttpRequestPtr &req, ResponseCallback &&cal
 
     Json::StreamWriterBuilder builder;
     builder.settings_["emitUTF8"] = true;
-    spdlog::info("\n{}", Json::writeString(builder, ret));
-    auto resp = drogon::HttpResponse::newHttpJsonResponse(ret);
+    // spdlog::info("\n{}", Json::writeString(builder, ret));
+    auto resp                     = drogon::HttpResponse::newHttpJsonResponse(ret);
 
 
     // std::ifstream           inputFile("../assets/test.json");

src/query/include/SearchResultBuilder.h

@@ -12,6 +12,7 @@ class SearchResultBuilder {
     void        addItem(SearchResultItem &&item);
     void        addItems(std::vector<std::unique_ptr<SG::SearchResultItem>> &&items);
     void        addPartsInfo(const std::map<std::string, uint64_t> &parts);
+    void        addPartsInfo(std::map<std::string, uint64_t> &&parts);
     void        addItemTotalNumber(uint64_t tot);
     Json::Value build();
 

src/query/src/Queryer.cpp

@@ -17,9 +17,9 @@ using SG::Core::SkipList;
 Json::Value Queryer::get(const std::string &content, uint64_t rkBegin, uint64_t rkEnd) {
     SG::SearchResultBuilder ret;
 
-    SG::PartsInfo partsInfo = createPartsInfo(content);
-    ret.addPartsInfo(partsInfo);
+    SG::PartsInfo partsInfo     = createPartsInfo(content);
     auto [resultList, totalCnt] = createResultList(partsInfo, rkBegin, rkEnd);
+    ret.addPartsInfo(std::move(partsInfo));
     ret.addItems(std::move(resultList));
     ret.addItemTotalNumber(totalCnt);
     return ret.build();
@@ -58,7 +58,7 @@ std::pair<std::vector<std::unique_ptr<SearchResultItem>>, uint64_t> Queryer::cre
         uint64_t      outputID = output / 400;
         std::ifstream inputFile("../assets/library/skl/" + std::to_string(outputID) + ".lib");
         if (!inputFile.is_open()) {
-            spdlog::error("[Queryer::creatResultList] Failed to open json file");
+            spdlog::error("[Queryer::creatResultList] Failed to open skl library file {}", std::to_string(outputID) + ".lib");
             inputFile.close();
             break;
         }
@@ -78,11 +78,11 @@ std::pair<std::vector<std::unique_ptr<SearchResultItem>>, uint64_t> Queryer::cre
 
     std::vector<std::vector<Doc>> combineResult = SkipList<Doc>::combine(sls);
 
-    std::cout << combineResult.size() << std::endl;
-    for (auto &i : combineResult) {
-        std::cout << i.size() << std::endl;
-    }
-    std::cout << idfs.size() << std::endl;
+    // std::cout << combineResult.size() << std::endl;
+    // for (auto &i : combineResult) {
+    //     std::cout << i.size() << std::endl;
+    // }
+    // std::cout << idfs.size() << std::endl;
 
     for (int i = 0; i < combineResult.size(); ++i) {
         double score = 0;

src/query/src/SearchResultBuilder.cpp

@@ -14,6 +14,10 @@ void SearchResultBuilder::addPartsInfo(const std::map<std::string, uint64_t> &pa
     partsInfo = parts;
 }
 
+void SearchResultBuilder::addPartsInfo(std::map<std::string, uint64_t> &&parts) {
+    partsInfo = std::move(parts);
+}
+
 void SearchResultBuilder::addItem(SearchResultItem &&item) {
     results.push_back(std::make_unique<SearchResultItem>(item));
 }

test/CMakeLists.txt

@@ -32,6 +32,17 @@ target_link_libraries(test_queryer PUBLIC
     query
 )
 
+find_package(benchmark REQUIRED)
+
+add_executable(test_benchmark test_benchmark.cpp)
+target_link_libraries(test_benchmark PUBLIC
+    query
+)
+target_link_libraries(test_benchmark PUBLIC
+    benchmark::benchmark
+)
+
+
 if(UNIX AND CMAKE_BUILD_TYPE MATCHES "Debug")
     target_compile_options(test_skiplist PRIVATE -fno-omit-frame-pointer)
     target_compile_options(test_skiplist PRIVATE -fsanitize=address)

test/test_benchmark.cpp

@@ -0,0 +1,16 @@
+#include "Queryer.h"
+#include <spdlog/spdlog.h>
+#include <benchmark/benchmark.h>
+using namespace SG;
+
+
+static void BM_Query(benchmark::State &state) {
+    Queryer &queryer = Queryer::getInstance();
+    for (auto _ : state) {
+        auto ans = queryer.get("寿命", 0, 10);
+        benchmark::DoNotOptimize(ans);
+    }
+}
+BENCHMARK(BM_Query);
+
+BENCHMARK_MAIN();

test/test_queryer.cpp

@@ -2,9 +2,12 @@
 #include <spdlog/spdlog.h>
 using namespace SG;
 int main() {
-    Queryer                  &queryer = Queryer::getInstance();
-    auto                      ans     = queryer.get("使用寿命", 0, 10);
+    Queryer &queryer = Queryer::getInstance();
+    auto     ans     = queryer.get("使用寿命", 0, 10);
+    auto     ans2    = queryer.get("使用寿命", 0, 10);
+
     Json::StreamWriterBuilder builder;
     builder.settings_["emitUTF8"] = true;
     spdlog::info("\n{}", Json::writeString(builder, ans));
+    spdlog::info("\n{}", Json::writeString(builder, ans2));
 }

test/test_skp_eff.cpp

@@ -0,0 +1,109 @@
+#include "SkipList.h"
+#include <chrono>
+#include <fstream>
+#include <iostream>
+#include <vector>
+#include <json/json.h>
+
+struct Item {
+    int id;
+    Item() = default;
+    Item(int id)
+        : id{id} {}
+    bool operator==(const Item &t) const {
+        return id == t.id;
+    }
+    bool operator<(const Item &t) const & {
+        return id < t.id;
+    }
+    bool operator<(const Item &t) const && {
+        return id < t.id;
+    }
+
+    friend std::ostream &operator<<(std::ostream &os, const Item &t) {
+        os << '(' << t.id << ')';
+        return os;
+    }
+
+    static Json::Value toJson(const Item &t) {
+        Json::Value ret;
+        ret["id"] = t.id;
+        return ret;
+    }
+    static Item fromJson(const Json::Value &src) {
+        return {src["id"].asInt()};
+    }
+};
+using SG::Core::SkipList;
+
+void test(const int &size, std::ofstream &outputFile) {
+    srand(time(nullptr));
+
+    SkipList<Item> skipList;
+
+    const int largeSize = size;
+
+    // test insertion
+    std::vector<int> insertData;
+    srand(time(nullptr));
+    for (int i = 0; i < largeSize; i++) {
+        insertData.push_back(rand() % largeSize);
+    }
+
+    auto start = std::chrono::steady_clock::now();
+    for (int i = 0; i < insertData.size(); i++) {
+        skipList.insert(Item(insertData[i]));
+    }
+    auto end      = std::chrono::steady_clock::now();
+    auto duration = std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+    outputFile << size << "," << duration.count();
+
+    // outputFile << "\n========\n"
+    //            << skipList.dump().toStyledString() << "\n========\n";
+    // std::cout << "=====\n";
+    // test deletion
+    // std::vector<int> removeData;
+    // srand(time(nullptr));
+    // for (int i = 0; i < largeSize; i++) {
+    //     removeData.push_back(rand() % largeSize);
+    // }
+
+    // start = std::chrono::steady_clock::now();
+    // for (int i = 0; i < removeData.size(); i++) {
+    //     skipList.remove(removeData[i]);
+    // }
+    // end      = std::chrono::steady_clock::now();
+    // duration = std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+    // outputFile << "," << duration.count();
+
+    // test search
+    std::vector<int> searchData;
+    for (int i = 0; i < largeSize; i++) {
+        searchData.push_back(rand() % largeSize);
+    }
+
+    start = std::chrono::steady_clock::now();
+    for (int i = 0; i < searchData.size(); i++) {
+        auto l = skipList.search(Item(searchData[i]));
+    }
+    end      = std::chrono::steady_clock::now();
+    duration = std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+    outputFile << "," << duration.count() << std::endl;
+
+    // std::cout << "=====\n";
+
+    // skipList.print();
+}
+
+int main() {
+    std::ofstream outputFile("result.csv");
+    outputFile << "DataSize,InsertionTime,SearchTime" << std::endl;
+
+    for (int i = 100; i <= 10000000; i *= 10) {
+        test(i, outputFile);
+    }
+
+    outputFile.close();
+
+    return 0;
+}