Added a basic benchmark.

CMakeLists.txt

@@ -37,8 +37,10 @@ link_directories(${CMAKE_SOURCE_DIR}/external/${ROCKSDB_NAME})
 add_executable(typesense_test test/forarray_test.cpp test/art_test.cpp test/collection_test.cpp test/topster_test.cpp
                ${SRC_FILES})
 add_executable(search ${HEADER_FILES} ${SRC_FILES} src/main/main.cpp)
+add_executable(benchmark ${HEADER_FILES} ${SRC_FILES} src/main/benchmark.cpp)
 add_executable(typesense-server ${HEADER_FILES} ${SRC_FILES} src/main/server.cpp)
 
 target_link_libraries(typesense_test for rocksdb gtest gtest_main)
 target_link_libraries(search for pthread rocksdb)
+target_link_libraries(benchmark for pthread rocksdb)
 target_link_libraries(typesense-server for curl h2o-evloop pthread rocksdb ssl crypto)

src/main/benchmark.cpp

@@ -0,0 +1,60 @@
+#include <stdlib.h>
+#include <iostream>
+#include <fstream>
+#include <vector>
+#include <numeric>
+#include <chrono>
+#include <art.h>
+#include <unordered_map>
+#include <queue>
+#include "string_utils.h"
+#include "collection.h"
+#include "collection_manager.h"
+
+using namespace std;
+
+int main() {
+    system("rm -rf /tmp/typesense-data && mkdir -p /tmp/typesense-data");
+
+    std::vector<field> fields_to_index = {field("title", field_types::STRING)};
+    std::vector<std::string> rank_fields = {"points"};
+    Store *store = new Store("/tmp/typesense-data");
+    CollectionManager & collectionManager = CollectionManager::get_instance();
+    collectionManager.init(store);
+
+    Collection *collection = collectionManager.get_collection("collection");
+    if(collection == nullptr) {
+        collection = collectionManager.create_collection("collection", fields_to_index, rank_fields);
+    }
+
+    std::ifstream infile("/Users/kishore/Downloads/hnstories.jsonl");
+
+    std::string json_line;
+
+    while (std::getline(infile, json_line)) {
+        collection->add(json_line);
+    }
+
+    infile.close();
+    cout << "FINISHED INDEXING!" << endl << flush;
+
+    std::vector<std::string> search_fields = {"title"};
+
+    std::vector<string> queries = {"the", "and", "to", "of", "in"};
+    auto counter = 0;
+    uint64_t results_total = 0; // to prevent optimizations!
+
+    auto begin = std::chrono::high_resolution_clock::now();
+
+    while(counter < 3000) {
+        auto i = counter % 5;
+        auto results = collection->search(queries[i], search_fields, 1, 100);
+        results_total += results.size();
+        counter++;
+    }
+
+    long long int timeMillis = std::chrono::duration_cast<std::chrono::milliseconds>(std::chrono::high_resolution_clock::now() - begin).count();
+    cout << "Time taken: " << timeMillis << "ms" << endl;
+    cout << "Total: " << results_total << endl;
+    return 0;
+}