1. 基于某个字段

比如要处理一批数据，把 id 作为查询条件，500 条数据作为一个批次

工具类代码如下：

public class DataLoopUtil {

    /**
     * 基于某个字段（比如 id, createTime 等）, 循环处理数据
     * P > Param
     * R > Result
     *
     * @param dataQueryFunction 查询数据
     * @param dataProcessConsumer 处理数据
     * @param extractParamFunction 从 R 中获取 P
     * @param startParam 起始参数
     */
    public static <P, R> void processData(Function<P, List<R>> dataQueryFunction,
                                          Consumer<List<R>> dataProcessConsumer,
                                          Function<R, P> extractParamFunction,
                                          P startParam) {
        List<R> dataList;
        do {
            dataList = dataQueryFunction.apply(startParam);
            if (CollUtil.isNotEmpty(dataList)) {
                log.info("data process result {}", LambdaUtils.mapToList(dataList, extractParamFunction));
                dataProcessConsumer.accept(dataList);
                startParam = extractParamFunction.apply(dataList.get(dataList.size() - 1));
            }
        } while (CollUtil.isNotEmpty(dataList));

        log.info("processData exist start param is {}", startParam);
    }
}

使用代码如下

@Component
public class DataLoopService {

    @Resource
    private MemberUserMapper memberUserMapper;

    public void loopGetMemberInfo() {
        DataLoopUtil.processData(getMemberQueryFn(), getMemberProcessConsumer(), MemberUserDO::getId, 600L);
    }

    // 查数据 Function
    private Function<Long, List<MemberUserDO>> getMemberQueryFn() {
        return startId -> memberUserMapper.selectList(
                new LambdaQueryWrapper<MemberUserDO>().gt(MemberUserDO::getId, startId).last(" limit 3"));
    }

    // 处理数据 Consumer
    private Consumer<List<MemberUserDO>> getMemberProcessConsumer() {
        return dataList -> dataList.forEach(member -> {
            MemberUserDO update = new MemberUserDO();
            update.setId(member.getId());
            update.setRegisterIp(member.getId().toString());
            memberUserMapper.updateById(update);
        });
    }
}

2. 基于上一批字段

比如初始化部门表的 parent_ids，先从第一批开始，然后开始第二批，轮子如下

工具类代码是如下

/**
 * 基于上次结果，循环处理数据
 * 比如查询部门，先用顶级部门 id 查询，查到下级，再用下级的 ids 查下下级
 */
public static <P, R> void processDataBatch(Function<Collection<P>, Collection<R>> dataQueryFunction,
                                           Consumer<Collection<R>> dataProcessConsumer,
                                           Function<Collection<R>, Collection<P>> extractParamFunction,
                                           Collection<P> startParam) {
    Collection<R> dataList;
    do {
        log.info("loop utils param is {}", JSONUtil.toJsonStr(startParam));
        dataList = dataQueryFunction.apply(startParam);
        if (CollUtil.isNotEmpty(dataList)) {
            dataProcessConsumer.accept(dataList);
            startParam = extractParamFunction.apply(dataList);
        }
    } while (CollUtil.isNotEmpty(dataList));

    log.info("processData exist start param is {}", startParam);
}

使用代码如下

@Override
public void initAncestor() {
   // 先查出第一批根部门
   List<DeptDO> deptDOList = deptMapper.selectListByParentId(Lists.newArrayList(DeptDO.PARENT_ID_ROOT));
   List<Long> parentIds = LambdaUtils.mapToList(deptDOList, DeptDO::getId);

   // 先处理根部门
   deptMapper.update(new LambdaUpdateWrapper<DeptDO>()
           .set(DeptDO::getParentIds, DeptDO.PARENT_ID_ROOT_STR)
           .eq(DeptDO::getParentId, 0L));

   // 循环处理
   DataLoopUtil.processDataBatch(
           getListDeptFn(),
           getDeptConsumer(),
           getParamExtractFn(),
           parentIds);
}

////////////////////////// 下面是参数获取方法 //////////////////////////

private Function<Collection<DeptDO>, Collection<Long>> getParamExtractFn() {
   return deptList -> LambdaUtils.mapToSet(deptList, DeptDO::getId);
}

private Function<Collection<Long>, Collection<DeptDO>> getListDeptFn() {
   return parentIds -> deptMapper.selectListByParentId(parentIds);
}

private Consumer<Collection<DeptDO>> getDeptConsumer() {
   return deptList -> {
       if (CollUtil.isEmpty(deptList)) {
           return;
       }
       // 批量查询父级部门
       Set<Long> parentIds = LambdaUtils.mapToSet(deptList, DeptDO::getParentId);
       List<DeptDO> parentDeptList = deptMapper.selectBatchIds(parentIds);
       Map<Long, DeptDO> parentMap = LambdaUtils.toMap(parentDeptList, DeptDO::getId);

       // 开始处理
       Map<Long, List<DeptDO>> deptMap = LambdaUtils.groupList(deptList, DeptDO::getParentId);
       deptMap.forEach((parentId, list) -> {
           DeptDO parent = parentMap.get(parentId);
           if (parent == null) {
               log.info("parent dept not exists, id {}", parentId);
               return;
           }
           list.forEach(dept -> deptMapper.update(new LambdaUpdateWrapper<DeptDO>()
                   .set(DeptDO::getParentIds, parent.getParentIds() + "," + parent.getId())
                   .eq(DeptDO::getId, dept.getId())));
       });
   };
}