业务代码中先处理业务最后存储数据

背景说明:

在处理复杂业务的时候,特别是研发自测期间,经常会产生很多不必要的垃圾数据。

技术原理:

先将要存入数据库的数据放在缓存中,等所有业务代码执行完后,再统一保存;

代码如下:

@Slf4j
public class BaseService<M extends BaseMapper<T>, T> extends ServiceImpl<M, T> implements IService<T> {

    /**
     * 构建查询
     * @param e
     * @return
     */
    @SneakyThrows
    public <E extends Object> EntityWrapper<E> wrapperTable(E e, String tableName, Boolean like) {
        EntityWrapper<E> wrapper = new EntityWrapper<>();
        for (String field: JSON.parseObject(JSON.toJSONString(e)).keySet()) {
            Field ld = findField(e.getClass(), field);
            if(ld == null || !checkField(ld)) continue;
            String _s = columnName(ld);
            String m =null;
            if(_s.contains("_")){
                 m = StrUtil.toCamelCase("get_"+_s);
            }else {m=StrUtil.genGetter(_s);}

            Object o = e.getClass().getMethod(m).invoke(e);
            if(ObjectUtils.isEmpty(o)) continue;
            if(!ObjectUtils.isEmpty(tableName)) {
                _s = tableName+"."+_s;
            }
            if(like) wrapper.like(_s,o.toString());
            else wrapper.eq(_s , o);
        }
        return wrapper;
    }
    /**
     * 构建查询
     * @param e
     * @return
     */
    @SneakyThrows
    public <E extends Object> EntityWrapper<E> wrapper(E e, Boolean like) {
        return wrapperTable(e,null,like);
    }
    /**
     * 构建查询
     * @param e
     * @return
     */
    @SneakyThrows
    public <E extends Object> EntityWrapper<E> wrapper(E e) {
        return wrapperTable(e,null,false);
    }
    private boolean checkField(Field field) {
        if(!field.isAnnotationPresent(TableId.class) && !field.isAnnotationPresent(TableField.class)) return false;
        if(field.isAnnotationPresent(TableField.class) && !field.getAnnotation(TableField.class).exist()) return false;
        return true;
    }
    private Field findField(Class c,String field) {
        if(c == Object.class) return null;
        try {
            return c.getDeclaredField(field);
        } catch (NoSuchFieldException e) {
            return findField(c.getSuperclass(),field);
        }
    }

    private String columnName(Field field) {
        if(field.isAnnotationPresent(TableId.class) && ObjectUtils.isNotEmpty(field.getAnnotation(TableId.class).value())) return field.getAnnotation(TableId.class).value();
        if(field.isAnnotationPresent(TableField.class) && field.getAnnotation(TableField.class).exist() && ObjectUtils.isNotEmpty(field.getAnnotation(TableField.class).value())) return field.getAnnotation(TableField.class).value();
        return StrUtil.toUnderlineCase(field.getName());
    }

    @Transactional(rollbackFor = Exception.class)
    @Override
    public boolean insertOrUpdate(T entity){
        if(entity instanceof TransacationalEntity) {
            //截流所有 TransacationalEntity 下的新增
            if(((TransacationalEntity) entity).getSqlStatus().equalsIgnoreCase("begin")) {
                return TransactionalUtils.add(this,((TransacationalEntity) entity));
            } else if(((TransacationalEntity) entity).getSqlStatus().equalsIgnoreCase("commit")) {
                return TransactionalUtils.commit(((TransacationalEntity) entity).getSqlVersion());
            } else {
                return super.insertOrUpdate(entity);
            }
        } else {
            return super.insertOrUpdate(entity);
        }
    }

    @SneakyThrows
    @Transactional(rollbackFor = Exception.class)
    @Override
    public boolean insertOrUpdateBatch(List<T> entityList) {
        try {
            this.baseMapper.getClass().getMethod("insertOrUpdateBatch", List.class).invoke(this.baseMapper,entityList);
        } catch (NoSuchMethodException e) {
            log.warn("mapper类{},不存在方法{},替换为service执行",this.baseMapper.getClass(),"insertOrUpdateBatch");
            super.insertOrUpdateBatch(entityList);
        }
        return true;
    }


}
业务父类

其他方法可以忽略,都是懒得写xml但是版本又不允许升级,偷懒写的sql条件辅助工具。

主要要关注以下两个方法:

第一个,用于截流原来的保存方法,将要保存的数据存入缓存中。

@Transactional(rollbackFor = Exception.class)
    @Override
    public boolean insertOrUpdate(T entity){
        if(entity instanceof TransacationalEntity) {
            //截流所有 TransacationalEntity 下的新增
            if(((TransacationalEntity) entity).getSqlStatus().equalsIgnoreCase("begin")) {
                return TransactionalUtils.add(this,((TransacationalEntity) entity));
            } else if(((TransacationalEntity) entity).getSqlStatus().equalsIgnoreCase("commit")) {
                return TransactionalUtils.commit(((TransacationalEntity) entity).getSqlVersion());
            } else {
                return super.insertOrUpdate(entity);
            }
        } else {
            return super.insertOrUpdate(entity);
        }
    }

第二个,用于最后统一处理保存,我这里偷懒了,使用了代码生成器,所以xml中已经有了 insertOrupdataBatch.

@SneakyThrows
    @Transactional(rollbackFor = Exception.class)
    @Override
    public boolean insertOrUpdateBatch(List<T> entityList) {
        try {
            this.baseMapper.getClass().getMethod("insertOrUpdateBatch", List.class).invoke(this.baseMapper,entityList);
        } catch (NoSuchMethodException e) {
            log.warn("mapper类{},不存在方法{},替换为service执行",this.baseMapper.getClass(),"insertOrUpdateBatch");
            super.insertOrUpdateBatch(entityList);
        }
        return true;
@Data
@AllArgsConstructor
@NoArgsConstructor
public class TransacationalEntity {

    /**
     * begin null commit
     */
    @TableField(exist = false)
    @ApiModelProperty(hidden = true)
    private String sqlStatus;

    /**
     * begin null commit
     */
    @TableField(exist = false)
    @ApiModelProperty(hidden = true)
    private String sqlVersion;
}
Pojo父类
public class TransactionalUtils {
    static volatile ConcurrentHashMap<String, ConcurrentHashMap<BaseService, CopyOnWriteArrayList<Object>>> cMap = new ConcurrentHashMap<>();
    static volatile ConcurrentHashMap<String,Long> tMap = new ConcurrentHashMap<>();
     public static <E extends TransacationalEntity> boolean add(BaseService service, E e) {
         long millis = System.currentTimeMillis();
         synchronized(service.getClass()) {
            ConcurrentHashMap<BaseService, CopyOnWriteArrayList<Object>> map = cMap
                    .getOrDefault(e.getSqlVersion(), new ConcurrentHashMap<>());
            CopyOnWriteArrayList<Object> list = map.getOrDefault(service, new CopyOnWriteArrayList<>());
            list.add(e);
            map.put(service, list);
            cMap.put(e.getSqlVersion(), map);
            tMap.put(e.getSqlVersion(),tMap.getOrDefault(e.getSqlVersion(), 0L)+System.currentTimeMillis()-millis) ;
        }
         return true;
    }

    @SneakyThrows
    public static boolean commit(String key) {
        long millis = System.currentTimeMillis();
        int size = 0;
        for (Map.Entry<BaseService, CopyOnWriteArrayList<Object>> entry : cMap.remove(key).entrySet()) {
            entry.getKey().insertOrUpdateBatch(entry.getValue());
            size+=entry.getValue().size();
        }
        System.out.println("批量业务入队总量" + size);
        System.out.println("批量业务入队耗时" + tMap.remove(key));
        System.out.println("批量业务提交耗时" + (System.currentTimeMillis()-millis));
        return true;
    }

}
核心工具类

以下是调用示例:

String tranId = IdUtil.simpleUUID();
        Map<String, Object> userMap = new HashMap();
        userMap.put("sqlStatus","begin");
        userMap.put("sqlVersion", tranId);
        propretyDye(beans,userMap);
        insertOrUpdate(Arrays.asList(beans));
        TransactionalUtils.commit(tranId);

先给要截流的数据加上版本号和标识,其实只要一个版本号就可以了,sqlStatus这个字段懒得删除了。

版本号主要用来在缓存中分组,隔离其他业务提交。

附带一个参数处理的小工具,有时候要重复的一层层赋值,太懒了,所以写个工具,让机器自己给我处理所有数据的值,用法如上面的示例,没仔细检查,偶尔会发生栈溢出,但是不影响,应该是项目的全局配置不够造成的:

/**
     * 属性染色
     * 如果对象 o 包含 propreties 的key,则设置key对应的值
     * 处理对象和对象中所有的子对象
     * @param o
     * @param propreties
     * @return
     */
    @SneakyThrows
    public static Object propretyDye(Object o, Map<String,Object> propreties) {
        if(o instanceof Collection) for (Object object : (Collection<?>) o) {
            propretyDye(object,propreties);
        }
        Class<?> aClass = o.getClass();
        while (aClass instanceof Object) {
            for (Field field : aClass.getDeclaredFields()) {
                field.setAccessible(true);
                if(propreties.containsKey(field.getName())) field.set(o,propreties.get(field.getName()));
                if(field.get(o) instanceof Collection) for (Object object : (Collection<?>) field.get(o)) {
                    propretyDye(object,propreties);
                }
                if(field.get(o) instanceof Object && BeanUtil.isBean(field.get(o).getClass())) propretyDye(field.get(o),propreties);
            }
            aClass = aClass.getSuperclass();
        }
        return o;
    }

 

posted on 2024-10-29 11:41  instr  阅读(19)  评论(0编辑  收藏  举报

导航