using XxHashMap = std::unordered_map<std::string, std::vector<std::string>>;

XxHashMap read_file_to_map(const std::string& file) {
    XxHashMap map;
    map.reserve(5000000);
    std::ifstream infile(file);
    std::string line;
    size_t start = __rdtsc();
    while (std::getline(infile, line)) {
        size_t pos = line.find(',');
        map[line.substr(0, pos)].emplace_back(line.substr(pos + 1));
    }
    size_t end = __rdtsc();
    std::cout << "Read cycles: " << end - start << '\n';
    return map;
}

void print_map(const XxHashMap& map) {
    for (const auto& [key, vec] : map) {
        std::cout << key << ':';
        for (const auto& val : vec) {
            std::cout << val << ',';
        }
        std::cout << '\n';
    }
}

void join(const XxHashMap& f1, const XxHashMap& f2, const XxHashMap& f3, const XxHashMap& f4) {
    std::ostream& buffer = std::cout;
    std::string line;
    for (const auto& [key, vec1] : f1) {
        auto it2 = f2.find(key);
        auto it3 = f3.find(key);
        if (it2 != f2.end() && it3 != f3.end()) {
            for (const auto& x1 : vec1) {
                for (const auto& x2 : it2->second) {
                    for (const auto& x3 : it3->second) {
                        auto it4 = f4.find(x3);
                        if (it4 != f4.end()) {
                            for (const auto& x4 : it4->second) {
                                line.clear();
                                line.append(x3).append(",").append(key).append(",").append(x1).append(",").append(x2).append(",").append(x4).append("\n");
                                buffer.write(line.c_str(), line.size());
                            }
                        }
                    }
                }
            }
        }
    }
}

int main(int argc, char* argv[]) {
    size_t startAll = __rdtsc();

    if (argc < 5) {
        std::cerr << "Usage: " << argv[0] << " <file1> <file2> <file3> <file4>\n";
        return 1;
    }

    // measure the time
    size_t startRead = __rdtsc();
    const XxHashMap f1 = read_file_to_map(argv[1]);
    const XxHashMap f2 = read_file_to_map(argv[2]);
    const XxHashMap f3 = read_file_to_map(argv[3]);
    const XxHashMap f4 = read_file_to_map(argv[4]);
    size_t endRead = __rdtsc();
    std::cout << "Read cycles : " << endRead - startRead << '\n';

    // print_map(f1);
    size_t startJoin = __rdtsc();
    // join(f1, f2, f3, f4);
    size_t endJoin = __rdtsc();
    std::cout << "Join cycles: " << endJoin - startJoin << '\n';
    size_t endAll = __rdtsc();
    std::cout << "All cycles: " << endAll - startAll << '\n';
    std::cout << "Join time: " << (endJoin - startJoin) * 100.0 / (endAll - startAll) << "%\n";
    std::cout << "Read time: " << (endRead - startRead) * 100.0 / (endAll - startAll) << "%\n";
    return 0;
}