基于 mysql 异步驱动的非阻塞 Mybatis【待验证】

转自:https://segmentfault.com/a/1190000016196693

虽然 spring5 也推出了 WebFlux 这一套异步技术栈,这种极大提升吞吐的玩法在 node 里玩的风生水起,但 java 世界里异步依旧不是主流,Vertx 倒是做了不少对异步的支持,但是其对于数据访问层的封装依旧还是挺精简的,传统的 javaer 还是受不了这种没有对象映射的工具库,于是我尝试将 Mybatis 移植到了异步驱动上,让数据访问层的工作变得更简单一些。给个例子:

@Sql(User.class)
public interface CommonMapper {
   @Select(columns = "id,age,username")
   @OrderBy("id desc")
   @Page
   @ModelConditions({
           @ModelCondition(field = "username", criterion = Criterions.EQUAL),
           @ModelCondition(field = "maxAge", column = "age", criterion = Criterions.LESS),
           @ModelCondition(field = "minAge", column = "age", criterion = Criterions.GREATER)
   })
   void query(UserSearch userSearch, DataHandler<List<User>> handler);
}

上面是 mapper 接口定义,方法的最后一个参数因为异步的原因所以变成了一个回调,不同的是有很多注解来表达 sql,看到这些注解应该不难猜出 sql 语句吧。如果不喜欢你当然可以继续使用 mapper.xml 的方式来写 sql。

更多内容移步代码库吧~


AsyncDao

asyncDao是一款异步非阻塞模型下的数据访问层工具。

  • MySQL only. 基于MySQL的异步驱动
  • 借鉴了Mybatis的mapping 和 dynamicSQL的内容,Mybatiser可以无缝切换
  • 注解表达SQL的能力
  • 事务支持
  • SpringBoot支持

Mybatis like

使用上与Mybatis几乎一致,由于异步非阻塞的关系,数据的返回都会通过回调DataHandler来完成,所以方法定义参数的最后一个一定是DataHandler类型。由于需要提取方法的参数名,于是需要加上编译参数-parameters,请将它在IDE和maven里配置上。

public interface CommonDao {

    void query(User user, DataHandler<List<User>> handler);

    void querySingle(User user, DataHandler<User> handler);

    void querySingleMap(User user, DataHandler<Map> handler);

    void insert(User user,DataHandler<Long> handler);

    void update(User user,DataHandler<Long> handler);

    void delete(User user,DataHandler<Long> handler);
}

mapper.xml与Mybatis几乎一致的写法(覆盖常见标签,一些不常用标签可能不支持,动态SQL建议使用注解SQL功能)

<?xml version="1.0" encoding="UTF-8"?>
<mapper namespace="com.tg.async.mapper.CommonDao">
    <resultMap id="BaseResultMap" type="com.tg.async.mapper.User">
        <id column="id" property="id"/>
        <result column="old_address" property="oldAddress"/>
        <result column="created_at" property="createdAt"/>
        <result column="password" property="password"/>
        <result column="now_address" property="nowAddress"/>
        <result column="state" property="state"/>
        <result column="age" property="age"/>
        <result column="username" property="username"/>
        <result column="updated_at" property="updatedAt"/>
    </resultMap>

    <select id="query" resultMap="BaseResultMap">select * from T_User
        <where>
            <if test="user.username!=null and user.username!=‘‘">AND username = #{user.username}</if>
            <if test="user.age != null">OR age > #{user.age}</if>
        </where>
        order by id desc
    </select>


    <insert id="insert" useGeneratedKeys="true" keyProperty="id">insert into T_User
        <trim prefix="(" suffix=")" suffixOverrides=",">
            <if test="user.oldAddress != null">old_address,</if>
            <if test="user.createdAt != null">created_at,</if>
            <if test="user.password != null">password,</if>
            <if test="user.nowAddress != null">now_address,</if>
            <if test="user.state != null">state,</if>
            <if test="user.age != null">age,</if>
            <if test="user.username != null">username,</if>
            <if test="user.updatedAt != null">updated_at,</if>
        </trim>
        <trim prefix="values (" suffix=")" suffixOverrides=",">
            <if test="user.oldAddress != null">#{user.oldAddress},</if>
            <if test="user.createdAt != null">#{user.createdAt},</if>
            <if test="user.password != null">#{user.password},</if>
            <if test="user.nowAddress != null">#{user.nowAddress},</if>
            <if test="user.state != null">#{user.state},</if>
            <if test="user.age != null">#{user.age},</if>
            <if test="user.username != null">#{user.username},</if>
            <if test="user.updatedAt != null">#{user.updatedAt},</if>
        </trim>
    </insert>

    <update id="update">
        update T_User
        <set>
            <if test="user.password != null">password=#{user.password},</if>
            <if test="user.age != null">age=#{user.age},</if>
        </set>
        where id = #{user.id}
    </update>
</mapper>

注解SQL

在XML里写SQL对于一些常见SQL实在是重复劳动,so这里允许你利用注解来表达SQL,该怎么做呢?

Table与Model关联

@Table(name = "T_User")
public class User {
    @Id("id")
    private Long id;
    //建议全部用包装类型,并注意mysql中字段类型与java类型的对应关系,mysql的int不会自动装换到这里的long

    private String username;
    private Integer age;

    @Column("now_address")
    private String nowAddress;

    @Column("created_at")
    private LocalDateTime createdAt;
    //asyncDao 里sql的时间类型都用joda,注意不是JDK8提供的那个,而是第三方包org.joda.time

    @Ignore
    private String remrk;

@Table记录数据表的名字 @Id记录主键信息 @Column映射了表字段和属性的关系,如果表字段和类属性同名,那么可以省略这个注解 @Ingore忽略这个类属性,没有哪个表字段与它关联。

定义接口

@Sql(User.class)
public interface CommonDao {
    @Select(columns = "id,age,username")
    @OrderBy("id desc")
    @Page
    @ModelConditions({
            @ModelCondition(field = "username", criterion = Criterions.EQUAL),
            @ModelCondition(field = "maxAge", column = "age", criterion = Criterions.LESS),
            @ModelCondition(field = "minAge", column = "age", criterion = Criterions.GREATER)
    })
    void query(UserSearch userSearch, DataHandler<List<User>> handler);


    @Select(columns = "age,username")
    @OrderBy("id desc")
    void queryParam(@Condition String username,
                    @Condition(criterion = Criterions.GREATER) Integer age,
                    @OffSet int offset,
                    @Limit int limit,
                    DataHandler<List<User>> handler);


    @Select(columns = "username,age", sqlMode = SqlMode.COMMON)
    void queryList(@Condition(criterion = Criterions.IN, column = "id") int[] ids, DataHandler<List<User>> handler);

    @Insert(useGeneratedKeys = true, keyProperty = "id")
    void insert(User user, DataHandler<Long> handler);

    @Update
    @ModelConditions(@ModelCondition(field = "id"))
    void update(User user, DataHandler<Long> handler);

    @Delete
    @ModelConditions(@ModelCondition(field = "id"))
    void delete(User user, DataHandler<Long> handler);
}

看到这些注解你应该能猜出来SQL长什么样,接下来解释一下这些注解

查询

@Select(columns = "id,age,username")
@OrderBy("id desc")
@Page
@ModelConditions({
       @ModelCondition(field = "username", criterion = Criterions.EQUAL),
       @ModelCondition(field = "maxAge", column = "age", criterion = Criterions.LESS),
       @ModelCondition(field = "minAge", column = "age", criterion = Criterions.GREATER)
})
void query(UserSearch userSearch, DataHandler<List<User>> handler);
@Select
  • columns:默认 select *可以配置columns("username,age")选择部分字段;
  • SqlMode:有两个选择,SqlMode.SELECTIVE 和 SqlMode.COMMON,区别是selective会检查查询条件的字段是否为null来实现动态的查询,即值为null时不会成为查询条件。并且@Select@Count@Update@Delete都有selective这个属性。
@Condition
  • criterion:查询条件,=,<,>,in等,具体见Criterions
  • column:与表字段的对应,若与字段名相同可不配置
  • attach:连接 and,or, 默认是and
  • test:SqlMode为selective下的判断表达式,类似Mybatis<if test="username != null">里的test属性,动态化查询条件

@Limit@OffSet为分页字段。
方法的参数不加任何注解一样会被当做查询条件,如下面两个函数效果是一样的:

@Select()
void queryUser(Integer age,DataHandler<List<User>> handler);

@Select()
void queryUser(@Condition(criterion = Criterions.EQUAL, column = "age") Integer age,DataHandler<List<User>> handler);

查询Model

上面的例子在查询条件比较多时方法参数会比较多,我们可以把查询条件封装到一个类里,使用@ModelConditions来注解查询条件,注意被@ModelConditions注解的方法只能有两个参数,一个是查询model,一个是DataHandler。

@Select
@Page
@ModelConditions({
       @ModelCondition(field = "username", criterion = Criterions.EQUAL),
       @ModelCondition(field = "minAge", column = "age", criterion = Criterions.GREATER),
       @ModelCondition(field = "maxAge", column = "age", criterion = Criterions.LESS),
       @ModelCondition(field = "ids", column = "id", criterion = Criterions.IN)
})
void queryUser5(UserSearch userSearch,DataHandler<List<User>> handler);
@ModelCondition
  • field:必填,查询条件中类对应的属性
  • column:对应的表字段
  • test:动态SQL的判断表达式

@Page只能用在ModelConditions下的查询,并且方法参数的那个类应该有offsetlimit这两个属性,或者 使用@Page(offsetField = "offset",limitField = "limit")指定具体字段

统计

@Count
void count(DataHandler<Integer> handler);//返回Long类型

插入

@Insert(useGeneratedKeys = true, keyProperty = "id")//返回自增id
void insert(User user, DataHandler<Long> handler);

更新

@Update(columns = "username,age")//选择更新某几个列
void update(User user, DataHandler<Long> handler);//返回affectedRows

删除

@Delete
int delete(@Condition(criterion = Criterions.GREATER, column = "age") int min,
          @Condition(criterion = Criterions.LESS, column = "age") int max,
          DataHandler<Long> handler);

@Delete
@ModelConditions(@ModelCondition(field = "id"))
void delete(User user, DataHandler<Long> handler);

使用

简单的编程使用

AsyncConfig asyncConfig = new AsyncConfig();
PoolConfiguration configuration = new PoolConfiguration("username", "localhost", 3306, "password", "database-name");
asyncConfig.setPoolConfiguration(configuration);
asyncConfig.setMapperPackages("com.tg.async.mapper");//mapper接口
asyncConfig.setXmlLocations("mapper/");//xml目录,classpath的相对路径,不支持绝对路径
AsyncDaoFactory asyncDaoFactory = AsyncDaoFactory.build(asyncConfig);
CommonDao commonDao = asyncDaoFactory.getMapper(CommonDao.class);
   
UserSearch userSearch = new UserSearch();
userSearch.setUsername("ha");
userSearch.setMaxAge(28);
userSearch.setMinAge(8);
userSearch.setLimit(5);
CountDownLatch latch = new CountDownLatch(1);
commonDao.query(user, users -> {
  System.out.println(users);
  latch.countDown();
});
latch.await();

事务

Mybatis和Spring体系里有一个非常好用的@Translactional注解,我们知道事务本质就是依赖connection的rollback等操作,那么一个事务下多个SQL就要共用这一个connection,如何共享呢?传统的阻塞体系下ThreadLocal就成了实现这一点的完美解决方案。那么在异步世界里,要实现mybatis-spring一样的上层Api来完成事务操作是一件非常困难的事,难点就在于Api太上层,以至于无法实现connection共享。于是这里自能退而求其次,使用编程式的方式来使用事务,抽象出一个Translaction,具体的mapper通过translaction.getMapper()来获取,这样通过同一个Translaction得到的Mapper都将共用一个connection。

CountDownLatch latch = new CountDownLatch(1);
AsyncConfig asyncConfig = new AsyncConfig();
PoolConfiguration configuration = new PoolConfiguration("username", "localhost", 3306, "password", "database-name");
asyncConfig.setPoolConfiguration(configuration);
asyncConfig.setMapperPackages("com.tg.async.mapper");
asyncConfig.setXmlLocations("mapper/");
asyncDaoFactory = AsyncDaoFactory.build(asyncConfig);
asyncDaoFactory.startTranslation(res -> {
    Translaction translaction = res.result();
    System.out.println(translaction);
    CommonDao commonDao = translaction.getMapper(CommonDao.class);
    User user = new User();
    user.setUsername("insert");
    user.setPassword("1234");
    user.setAge(28);
    commonDao.insert(user, id -> {
        System.out.println(id);
        translaction.rollback(Void -> {
            latch.countDown();
        });
    });
});
latch.await();

SpringBoot

虽然Spring5推出了WebFlux,但异步体系在Spring里依旧不是主流。在异步化改造的过程中,大部分人也往往会保留Spring的IOC,而将其他交给Vertx,所以asyncDao对于Spring的支持就是将Mapper注入IOC容器。

quick start

YAML配置文件:

async:
    dao:
     mapperLocations: /mapper  #xml目录,classpath的相对路径,不支持绝对路径
     basePackages: com.tg.mapper #mapper所在包
     username: username
     host: localhost
     port: 3306
     password: pass
     database: database-name
     maxTotal: 12
     maxIdle: 12
     minIdle: 1
     maxWaitMillis: 10000

添加@Mapper来实现注入

@Mapper
@Sql(User.class)
public interface CommonDao {
    @Select(columns = "id,age,username")
    @OrderBy("id desc")
    @Page(offsetField = "offset", limitField = "limit")
    @ModelConditions({
            @ModelCondition(field = "username", criterion = Criterions.EQUAL),
            @ModelCondition(field = "maxAge", column = "age", criterion = Criterions.LESS),
            @ModelCondition(field = "minAge", column = "age", criterion = Criterions.GREATER)
    })
    void query(UserSearch userSearch, DataHandler<List<User>> handler);
}

通过@EnableAsyncDao来开启支持,简单示例:

@SpringBootApplication
@EnableAsyncDao
public class DemoApplication {

    public static void main(String[] args){
        ApplicationContext applicationContext = SpringApplication.run(DemoApplication.class);
        CommonDao commonDao = applicationContext.getBean(CommonDao.class);

        UserSearch userSearch = new UserSearch();
        userSearch.setUsername("ha");
        userSearch.setMaxAge(28);
        userSearch.setMinAge(8);
        userSearch.setLimit(5);

        commonDao.query(userSearch, users -> {
            System.out.println("result: " + users);
        });
    }
}
 

 

基于 mysql 异步驱动的非阻塞 Mybatis【待验证】

上一篇:mysql高级


下一篇:Oracle静默安装搭建