如何在MySql中获取组中的特定部分？

Question

我有一个SQL查询的问题，我会尝试解释我想要做什么。这是我的查询返回一些结果行。

SELECT 
sck.sckid, 
sck.prid, 
sck.paid, 
sck.sckcen, 
sum(scd.scdkiek) as count_of_goods, 
min(scn.scndat) as `date` 
FROM sck 
INNER JOIN scd ON scd.sckid = sck.sckid 
INNER JOIN scn ON scn.scnid = scd.scnid 
INNER JOIN sandeliai ON sandeliai.paid = sck.paid 
WHERE sck.prid = 1376 GROUP BY sckid 

而且该查询返回“表1”的结果：

表1：

sckid | prid | sckcen | count_of_goods |  date  | 
123  | 1376 | 10009  |   0   | 2012-12-31 | 
124  | 1376 | 10007  |   15   | 2013-01-25 | 
125  | 1376 | 10005  |   0   | 2013-02-13 | 
126  | 1376 | 10000  |   18   | 2013-03-15 | 

但我只需要一行与prid分组的所有数据，我写此查询：

SELECT 
sck.sckid, 
sck.prid, 
sck.paid, 
sck.sckcen, 
sum(scd.scdkiek) as count_of_goods, 
min(scn.scndat) as `date` 
FROM sck 
INNER JOIN scd ON scd.sckid = sck.sckid 
INNER JOIN scn ON scn.scnid = scd.scnid 
INNER JOIN sandeliai ON sandeliai.paid = sck.paid 
WHERE sck.prid = 1376 GROUP BY prid 

然后我在表格中得到一行这个数据：

表2：

sckid | prid | sckcen | count_of_goods |  date  | 
123  | 1376 | 10009  |   23   | 2012-12-31 | 

这似乎是正确的，但在date领域，我需要从表中的一个，其中有count_of_goods> 0回到最早期的日期，所以对我来说需要这样的结果：

表3：

sckid | prid | sckcen | count_of_goods |  date  | 
123  | 1376 | 10009  |   23   | 2013-01-25 | 

所以，任何想法我怎么能得到像表3中的结果？

Answer 1

尝试使用having，像

SELECT 
sck.sckid, 
sck.prid, 
sck.paid, 
sck.sckcen, 
sum(scd.scdkiek) as count_of_goods, 
min(scn.scndat) as `date` 
FROM sck 
INNER JOIN scd ON scd.sckid = sck.sckid 
INNER JOIN scn ON scn.scnid = scd.scnid 
INNER JOIN sandeliai ON sandeliai.paid = sck.paid 
WHERE sck.prid = 1376 GROUP BY prid HAVING count_of_goods > 0; 

这应该为你工作。这可能会跳过count = 0的行，我认为这应该没问题（只是我个人的意见）。

请以此为出发点，而不是最终解决方案。

编辑：与别名

Answer 2

有更新的总和（...）我现在我的问题是不是很合乎逻辑的我尝试这两种查询在我的问题，然后我得到一个很好的答案相结合，这是我的解决方案：

SELECT 
sck.sckid, 
sck.prid, 
sck.paid, 
sck.sckcen, 
sum(scd.scdkiek) as count_of_goods, 
(SELECT 
scn.scndat 
FROM sck 
INNER JOIN scd ON scd.sckid = sck.sckid 
INNER JOIN scn ON scn.scnid = scd.scnid 
WHERE sck.prid = 1376 GROUP BY sck.sckid having sum(scd.scdkiek) > 0 ORDER BY scn.scndat LIMIT 1) as `date` 
FROM sck 
INNER JOIN scd ON scd.sckid = sck.sckid 
INNER JOIN scn ON scn.scnid = scd.scnid 
WHERE sck.prid = 1376 GROUP BY prid; 

Answer 3

我认为我最初回来的答案可能已经非常复杂。假设scd.scdkiek是数字，是不是ID，那么你可能会得到你想要通过简单地改变INNER JOIN的标准是什么，即

SELECT 
    sck.sckid, 
    sck.prid, 
    sck.paid, 
    sck.sckcen, 
    scn.scndat, 
    SUM(scd.scdkiek) AS count_of_goods 
FROM sck 
INNER JOIN scd 
    ON scd.sckid = sck.sckid 
    AND scd.scdkiek > 0 
INNER JOIN sandeliai 
    ON sandeliai.paid = sck.paid 
INNER JOIN scn 
    ON scn.scnid = scd.scnid 
WHERE sck.prid = 1376 
GROUP BY 
    sck.prid 

做不到这一点，虽然我会考虑使用子查询。子查询将计算每行count_of_goods。然后，外部查询将使用它来决定是否将该行的日期设置为NULL。 MIN（... somedate ...，NULL）将返回日期。这反过来将意味着你应该得到该行的count_of_goods不为NULL的最小日期。

当您正在查找count_of_goods> 0的日期时，子查询按日期分组。

这仅仅是一个当然的选择，可能还有其他更有效的方法，但我相信这会按照你的目标是要做到

SELECT 
    sck.sckid, 
    sck.prid, 
    sck.paid, 
    sck.sckcen, 
    SUM(d.count_of_goods) AS count_of_goods, 
    MIN(IF(d.count_of_goods>0,d.scndat,NULL)) AS `date` 
FROM (
    -- Get the data and group it by date 
    SELECT 
     sck.sckid, 
     sck.prid, 
     sck.paid, 
     sck.sckcen, 
     scn.scndat, 
     SUM(scd.scdkiek) AS count_of_goods 
    FROM sck 
    INNER JOIN scd ON scd.sckid = sck.sckid 
    INNER JOIN sandeliai ON sandeliai.paid = sck.paid 
    INNER JOIN scn ON scn.scnid = scd.scnid 
    WHERE sck.prid = 1376 
    GROUP BY 
     sck.prid, -- may not really be needed 
     scn.scndat 
) AS d 
GROUP BY d.prid