tbl_sms：

Post_ID | User_ID | SMS_Sent_DT
---------------------------------
123     |   007   |   2010-05-14 10:15:25
123     |   008   |   2010-05-14 10:18:32
123     |   009   |   2010-05-14 10:23:05
123     |   010   |   2010-05-14 10:23:06

tbl_bids：

Post_ID | User_ID | Bid_DT
--------------------------
123     |   010   |   2010-05-14 10:27:25
123     |   008   |   2010-05-14 10:28:32
123     |   009   |  2010-05-14 10:28:47
123     |   007   |   2010-05-14 10:35:06

对于这两个表，我试图实现以下目标：

Post_ID | First_BID_Time | First_BID_SMS_TIME | Second_BID_Time | Second_BID_SMS_Time | Third_BID_Time | Third_BID_SMS_Time
-----------------------------------------------------------------------------------------------------------------------------------
123     | 2010-05-14 10:27:25 | 2010-05-14 10:23:06 | 2010-05-14 10:28:32 | 2010-05-14 10:18:32 | 2010-05-14 10:28:47 | 2010-05-14 10:23:05

我写的查询是：

   SELECT b.post_id,
          sms.message_sent_at as notif_sent1,
          b.message_sent_at as notif_accepted1,
          DATEDIFF(b.message_sent_at, sms.message_sent_at) AS delay1
     FROM tbl_bids b
LEFT JOIN tbl_sms_status sms ON (sms.jobid = b.post_id AND b.user_id = sms.userid)
    WHERE b.post_id = sms.jobid
 ORDER BY b.post_id ASC

这给了我正确的结果，但它们没有像我想要的那样转动。

请有人帮我解决这个问题。我欢迎任何解决方案，无论是冗长的查询还是程序。

Answer 1

在SQL Server中，使用ROW_NUMBER()函数或CROSS APPLY构造很容易实现。在MySQL中，这是harder。

一种解决方案是emulate ROW_NUMBER() in MySQL使用变量。通过这种方式，可以返回按出价时间排名的每个帖子ID的出价，并获取user_ids。从那里开始，将SMS时间LEFT JOIN加入post_id / user_id组合是一件容易的事。按照链接中的示例，代码将类似于：

SELECT tmp.Post_ID, tmp.ranking, tmp.user_ID, tmp.Bid_DT, s.SMS_DT
FROM ( 
  SELECT 
    b.Post_ID, b.user_ID, b.Bid_DT, 
    IF( @prev <> ID, @rownum := 1, @rownum := @rownum+1 ) AS ranking, 
    @prev := ID 
  FROM tbl_bids b 
  JOIN (SELECT @rownum := NULL, @prev := 0) AS r 
  ORDER BY b.Post_ID, b.BID_DT 
) AS tmp
LEFT JOIN tbl_sms s
  ON tmp.Post_ID = s.Post_ID AND tmp.user_ID = s.user_ID 
WHERE tmp.rank <= 3 -- Top 3, adjust when more are necessary 
ORDER BY post_ID, ranking;

然后你会有这样的输出：

Post_ID | Ranking  | User_ID | Bid_DT                 | SMS_DT
---------------------------------------------------------------------------
123     |    1     |   010   |   2010-05-14 10:27:25  | 2010-05-14 10:23:06
123     |    2     |   008   |   2010-05-14 10:28:32  | ....
123     |    3     |   009   |   2010-05-14 10:28:47  | ....
123     |    4     |   007   |   2010-05-14 10:35:06  | ....
124     |    1     | .......

您可以将此结果存储在临时表中：

CREATE TEMPORARY TABLE RankedBids(Post_ID INTEGER, Ranking INTEGER, User_ID INTEGER, Bid_DT DATETIME, SMS_DT DATETIME)
INSERT INTO Rankedbids SELECT.... (use above query)

不幸的是，由于MySQL limitation您无法在查询中对同一个临时表使用多个引用，因此您必须按排名拆分此表：

CREATE TEMPORARY TABLE RankedBids1(Post_ID INTEGER, User_ID INTEGER, Bid_DT DATETIME, SMS_DT DATETIME)
CREATE TEMPORARY TABLE RankedBids2....
INSERT INTO Rankedbids1 SELECT Post_ID, User_ID, Bid_DT, SMS_DT FROM RankedBids WHERE Ranking = 1
INSERT INTO RankedBids2...

如果记录集非常大，那么在Post_ID上分配一个（主键）索引会加快转动查询的速度。

现在您可以转移此数据：

SELECT R1.Post_ID, R1.Bid_DT AS Bid_DT1, R1.SMS_DT AS SMS_DT1 .... 
FROM RankedBids1 R1
LEFT JOIN RankedBids2 R2 ON R1.Post_ID = R2.Post_ID
LEFT JOIN RankedBids3 R3 ON ........

但是，OMG Ponies有一点，它可以在一个不透明的桌子周围构建你的系统。因此，如果您不需要转动，请不要。

MySQL将行旋转到列以获得类似的记录

tbl_sms：

tbl_bids：

1 个答案: