MYSQL中遇到的问题以及解决方法（二）排序后筛选

MYSQL中没有row_number()over(partition by order by )，如何排序后进行筛选？

方法一：

    select OperatorID,sum(cnt) as '总人数',sum(case when Quality=1 then cnt  else  0 end )  as '质量合格的人数',
    sum(case when Quality=-1 then cnt  else  0 end )  as '质量一般的人数',
    sum(case when Quality=-2 then cnt  else  0 end )  as '质量差的人数',
    sum(case when Quality=-3 then cnt  else  0 end )  as '血压可疑的人数'
    from (select OperatorID,Quality,count(0)cnt from (
    select * from (select * from
     (SELECT
    p.DataID,p.Quality,p.PatientID , p.PPGFilePath, p.SBP, p.DBP, pa.Age, pa.Sex, pa.Height, pa.Weight,round( pa.BMI,2)BMI,p.ECGFilePath, p.OperatorID
FROM
    ppg_data AS p
    LEFT JOIN 
    (select *,pa.Weight/power(pa.Height,2)*10000 as BMI from patient AS pa)pa ON p.PatientID=pa.PatientID
    LEFT JOIN operator AS o ON o.OperatorID = p.OperatorID
WHERE
    p.TestEndTime >= '2020-06-01 00:00:00' 
    AND p.TestEndTime < '2020-07-01 00:00:00'
    AND pa.RecordIsDeleted = 0
    AND o.isCensus = 1
    AND p.OperatorID IN ('107','112','113','114','116','117','132','134','135','145','146','147','149','154','153')
    AND p.PPGFilePath IS NOT NULL
    AND p.SBP IS NOT NULL
ORDER BY p.PatientID)as p order by p.Quality desc,DataID desc
limit 20000)p
where  (p.OperatorID,p.PatientID)  IN (select  p.OperatorID,p.PatientID  FROM  patient p   group by  p.OperatorID,p.PatientID)
group by p.PatientID ,p.OperatorID)as a
group by OperatorID,Quality)A
group by OperatorID;

可以先进行排序后limit，必须进行limit，而且limit必须要大于真实的条数，后再进行group by 就会选择第一条数据，那么如何选择前两条呢？貌似这个方法解决不了？

方法二：

select OperatorID,sum(cnt) as '总人数',sum(case when Quality=1 then cnt  else  0 end )  as '质量合格的人数',
    sum(case when Quality=-1 then cnt  else  0 end )  as '质量一般的人数',
    sum(case when Quality=-2 then cnt  else  0 end )  as '质量差的人数',
    sum(case when Quality=-3 then cnt  else  0 end )  as '血压可疑的人数'
    from (select OperatorID,Quality,count(0)cnt from (
    select * from 
(select a.*,IF (@p=PatientID,@r:=@r+1,@r:=1) as rank,@p:=PatientID from
    (SELECT    p.DataID,p.Quality,p.PatientID , p.PPGFilePath, p.SBP, p.DBP, pa.Age, pa.Sex, pa.Height, pa.Weight,round( pa.BMI,2)BMI,p.ECGFilePath, p.OperatorID
FROM
    ppg_data AS p
    LEFT JOIN 
    (select *,pa.Weight/power(pa.Height,2)*10000 as BMI from patient AS pa)pa ON p.PatientID=pa.PatientID
    LEFT JOIN operator AS o ON o.OperatorID = p.OperatorID
WHERE
    p.TestEndTime >= '2020-06-01 00:00:00' 
    AND p.TestEndTime < '2020-07-01 00:00:00'
    AND pa.RecordIsDeleted = 0
    AND o.isCensus = 1
    AND p.OperatorID IN ('107','112','113','114','116','117','132','134','135','145','146','147','149','154','153')
    AND  (p.OperatorID,p.PatientID)  IN (select  p.OperatorID,p.PatientID  FROM  patient p   group by  p.OperatorID,p.PatientID)
    AND p.KRKPPGFilePath IS NOT NULL
    AND p.SBP IS NOT NULL
ORDER BY p.PatientID)a,(select @p:=NUll,@r:=0)r
ORDER BY PatientID,Quality desc,DataID desc)A
where rank=1)A
group by  OperatorID,Quality)A
group by OperatorID;

rank=1就是选择第一条， <3就是前两条

对于按照多个字段进行分组的，需要将IF后面的条件用and连接即可。@p=PatientID and @s=TestEndTime

参考链接： https://www.cnblogs.com/niniya/p/9046449.html