@Component
@Slf4j
public class BatchDao {

    @Autowired
    private SqlSessionTemplate sqlSessionTemplate;

    public <T,M> boolean batchSave(Collection<T> entityList, Class<M> mapper, BiConsumer<M,Collection<T>> fuc) {
        SqlSessionFactory sqlSessionFactory = sqlSessionTemplate.getSqlSessionFactory();
        SqlSession sqlSession = sqlSessionFactory.openSession(ExecutorType.BATCH, false);
        M entityMapper = sqlSession.getMapper(mapper);
        try {
            fuc.accept(entityMapper, entityList);
            sqlSession.flushStatements();
            sqlSession.clearCache();
            return true;
        } catch (Exception e) {
            log.error("{}",e.getMessage());
            sqlSession.rollback();
        } finally {
            sqlSession.close();
        }
        return false;
    }
}

/**
 * @author z201.coding@gmail.com
 **/
@Slf4j
@State(Scope.Benchmark)
public class AppApplicationJMHTest {

    private final static Integer MEASUREMENT_ITERATIONS = 1;
    private final static Integer WARMUP_ITERATIONS = 1;

    static ConfigurableApplicationContext context;

    protected TableDataDao tableDataDao;

    protected BatchDao batchDao;

    protected SqlSessionFactory sqlSessionFactory;

    @Setup(Level.Trial)
    public synchronized void initialize() {
        try {
            String args = "";
            if (context == null) {
                context = SpringApplication.run(AppApplication.class, args);
            }
            tableDataDao = context.getBean(TableDataDao.class);
            batchDao = context.getBean(BatchDao.class);
            sqlSessionFactory = context.getBean(SqlSessionFactory.class);
            tableDataDao.truncate();
        } catch (Exception e) {
            e.printStackTrace();
        }
    }
    // Throughput 整体吞吐量，例如”1秒内可以执行多少次调用”。
    // AverageTime: 调用的平均时间，例如”每次调用平均耗时xxx毫秒”。
    // SampleTime: 随机取样，最后输出取样结果的分布，例如”99%的调用在xxx毫秒以内，99.99%的调用在xxx毫秒以内”
    // SingleShotTime: 以上模式都是默认一次 iteration 是 1s，唯有 SingleShotTime 是只运行一次。往往同时把 warmup 次数设为0，用于测试冷启动时的性能。
    // All(“all”, “All benchmark modes”);
    @Test
    public void executeJmhRunner() throws RunnerException {
        Options opt = new OptionsBuilder()
                // 设置类名正则表达式基准为搜索当前类
                .include("\\." + AppApplicationJMHTest.class.getSimpleName() + "\\.")
                // 都是一些基本的参数，可以根据具体情况调整。一般比较重的东西可以进行大量的测试，放到服务器上运行。
                .warmupIterations(WARMUP_ITERATIONS) // 多少次预热
                .measurementIterations(MEASUREMENT_ITERATIONS) // 要做多少次测量m
                .timeUnit(TimeUnit.MILLISECONDS) // 毫秒
                // 不使用多线程
                .forks(0) // 进行 fork 的次数。如果 fork 数是2的话，则 JMH 会 fork 出两个进程来进行测试。
                .threads(1) // 每个进程中的测试线程，这个非常好理解，根据具体情况选择，一般为cpu乘以2。
                .mode(Mode.AverageTime)
                .shouldDoGC(true)
                .shouldFailOnError(true)
                .resultFormat(ResultFormatType.JSON) // 输出格式化
                .result("/dev/null") // set this to a valid filename if you want reports
//                .result("benchmark.json")
                .shouldFailOnError(true)
                .jvmArgs("-server")
                .build();
        new Runner(opt).run();
    }

    private List<TableData> mockData(int size) {
        List<TableData> list = new ArrayList<>();
        Long time = Clock.systemDefaultZone().millis();
        for (int i = 0; i < size; i++) {
            list.add(TableData.builder().createTime(time + i).build());
        }
        return list;
    }

    @Benchmark
    public void insert() {
        List<TableData> tableDataList = mockData(10);
        tableDataList.stream().forEach(i -> tableDataDao.insert(i));
    }

    @Benchmark
    public void batchInsert() {
        List<TableData> tableDataList = mockData(10);
        tableDataDao.batchInsert(tableDataList);
    }

    @Benchmark
    public void batchSave() {
        List<TableData> tableDataList = mockData(10);
        boolean result = batchDao.batchSave(tableDataList, TableDataDao.class, (mapper, data) -> {
            mapper.batchInsert(tableDataList);
        });
    }


    @Benchmark
    public void insert100() {
        List<TableData> tableDataList = mockData(100);
        tableDataList.stream().forEach(i -> tableDataDao.insert(i));
    }

    @Benchmark
    public void batchInsert100() {
        List<TableData> tableDataList = mockData(100);
        tableDataDao.batchInsert(tableDataList);
    }

    @Benchmark
    public void batchSave100() {
        List<TableData> tableDataList = mockData(100);
        boolean result = batchDao.batchSave(tableDataList, TableDataDao.class, (mapper, data) -> {
            mapper.batchInsert(tableDataList);
        });
    }

    @Benchmark
    public void insert1000() {
        List<TableData> tableDataList = mockData(1000);
        tableDataList.stream().forEach(i -> tableDataDao.insert(i));
    }

    @Benchmark
    public void batchInsert1000() {
        List<TableData> tableDataList = mockData(1000);
        tableDataDao.batchInsert(tableDataList);
    }

    @Benchmark
    public void batchSave1000() {
        List<TableData> tableDataList = mockData(1000);
        boolean result = batchDao.batchSave(tableDataList, TableDataDao.class, (mapper, data) -> {
            mapper.batchInsert(tableDataList);
        });
    }

    @Benchmark
    public void batchInsert5000() {
        List<TableData> tableDataList = mockData(5000);
        tableDataDao.batchInsert(tableDataList);
    }

    @Benchmark
    public void batchSave5000() {
        List<TableData> tableDataList = mockData(5000);
        boolean result = batchDao.batchSave(tableDataList, TableDataDao.class, (mapper, data) -> {
            mapper.batchInsert(tableDataList);
        });
    }

}