SSM分页查询

分页处理，这是做JavaWeb项目中常见的场景。

背景：

系统架构：SpringCloud分布式
持久层：MyBatis
前端：前后分离vue.js/bootstrap等.
后台提供restful api 接口，前端访问后端接口展示数据。

2种方式提供分页处理方案：

一、直接MyBatis数据库进行分页

controller接口

@ApiImplicitParams({
@ApiImplicitParam(name = “categoryId”, value = “支付渠道大类Id”, required = false, dataType = “Long”, paramType = “query”),
@ApiImplicitParam(name = “payChannelId”, value = “支付渠道ID”, required = false, dataType = “Long”, paramType = “query”),
@ApiImplicitParam(name = “bankCode”, value = “银行名称code”, required = false, dataType = “String”, paramType = “query”),
@ApiImplicitParam(name = “startTime”, value = “创建时间起始（示例：2018-09-05）”, required = false, dataType = “String”, paramType = “query”, length = 10),
@ApiImplicitParam(name = “endTime”, value = “创建时间截止（示例：2018-09-05）”, required = false, dataType = “String”, paramType = “query”, length = 10),
@ApiImplicitParam(name = “pageNum”, value = “查询开始页”, required = true, dataType = “int”, paramType = “query”),
@ApiImplicitParam(name = “pageSize”, value = “查询的页面大小不需要分页则把此值填大一点”, required = true, dataType = “int”, paramType = “query”)})
public PageResult queryPayChannel(
@RequestParam(value = “categoryId”, required = false) Long categoryId,
@RequestParam(value = “payChannelId”, required = false) Long payChannelId,
@RequestParam(value = “bankCode”, required = false) String bankCode,
@RequestParam(value = “startTime”, required = false) String startTime,
@RequestParam(value = “endTime”, required = false) String endTime,
@RequestParam(value = “pageNum”, required = true) Integer pageNum,
@RequestParam(value = “pageSize”, required = true) Integer pageSize) {
PageResult response = new PageResult();
try {
int endRowNo = pageNum * pageSize;
int beginRowNo = (pageNum - 1) * pageSize + 1;
return channelBaseDataService.queryBaseData(categoryId,payChannelId,bankCode,startTime,endTime,beginRowNo,endRowNo);
} catch (Exception e) {
response = new PageResult();
if (e instanceof AppException) {
response.setCode(((AppException) e).getErrorCode());
response.setResult(((AppException) e).getErrorMsg());
} else {
response.setCode(CumReturnCode.SYSTEM_EXCEPTION.code);
response.setResult(CumReturnCode.SYSTEM_EXCEPTION.message);
}
return response;
}
}
Swagger-ui接口界面如下：

pageNum 和 pageSize 用于前端分页的参数，pageNum：表示页码第几页，pageSize：表示每页展示数据数量。

所有查询条件参数，在mapper-xml里进行处理，再利用数据库oracle本身的rownum行值进行分页。

xml代码：

select ID, PAY_CHANNEL_MAIN_ID, PAY_CHANNEL_ID, BANK_CODE, CREATE_TIME, UPDATE_TIME,UPDATOR from ( select A.*, rownum RN from ( select ID, PAY_CHANNEL_MAIN_ID, PAY_CHANNEL_ID, BANK_CODE, CREATE_TIME, UPDATE_TIME,UPDATOR from CUM_PAY_CHANNEL_BASE CREATE_TIME <![CDATA[ >= ]]> #{beginCreateTime,jdbcType=TIMESTAMP} AND CREATE_TIME <![CDATA[ <= ]]> #{endCreateTime,jdbcType=TIMESTAMP} AND PAY_CHANNEL_MAIN_ID = #{categoryId,jdbcType=DECIMAL} AND PAY_CHANNEL_ID = #{payChannelId,jdbcType=DECIMAL} AND BANK_CODE = #{bankCode,jdbcType=VARCHAR} order by CREATE_TIME desc ) A where rownum <= #{endRowNo,jdbcType=DECIMAL} ) where RN >= #{beginRowNo,jdbcType=DECIMAL} 优点：直观、方便、易排查问题。 缺点：访问数据库过于频繁，未利用到mybatis本身的缓存优势。

二、Java+缓存分页

这种方法对于前端而言是没变化，无感的。

只是在后端处理稍作变动，思路：

先把所有数据记录查询出来
数据库实体再次封装为查询实体
纯java代码进行分页
controller接口不变

service业务处理——>数据库crud操作变动

把所有数据库记录查询处理

select ID, PAY_CHANNEL_MAIN_ID, PAY_CHANNEL_ID, BANK_CODE, CREATE_TIME, UPDATE_TIME,UPDATOR from CUM_PAY_CHANNEL_BASE CREATE_TIME <![CDATA[ >= ]]> #{beginCreateTime,jdbcType=TIMESTAMP} AND CREATE_TIME <![CDATA[ <= ]]> #{endCreateTime,jdbcType=TIMESTAMP} AND PAY_CHANNEL_MAIN_ID = #{categoryId,jdbcType=DECIMAL} AND PAY_CHANNEL_ID = #{payChannelId,jdbcType=DECIMAL} AND BANK_CODE = #{bankCode,jdbcType=VARCHAR} order by CREATE_TIME desc 再按照pageNum和pageSize进行分页处理

    //以分页形式输出给前端
    List<ChannelRouteGroupResp> resultList = new ArrayList<>();
    if(groupRespList.size() >= endRowNo){
        for(int i=(beginRowNo-1); i< endRowNo; i++){
            resultList.add(groupRespList.get(i));
        }
    }else{
        for(int i=(beginRowNo-1); i< groupRespList.size(); i++){
            resultList.add(groupRespList.get(i));
        }
    }

优点：利用了mybatis缓存机制，分页查询快速，减少数据库访问次数。

缺点：当数据量大的时候，100W+；在没有条件查询数据库的时候，耗时久