在 MySQL 中自联接表时在不同列中显示特定结果答案

【问题标题】：Showing specific results in different columns when self-joining a table in MySQL在 MySQL 中自联接表时在不同列中显示特定结果
【发布时间】：2016-09-26 16:01:46
【问题描述】：

我有一个表 email_activities_2，有一个标题为“action_type”的列，其中结果可以是“sent”、“open”、“click”等，每个 action、每个 email、每个 recipient_id 都有一条记录.例如，recipient_id 54 的 email_id 94607 有两行：第一行的 action_type 是“已发送”，第二行是“打开”。

我想要一个查询，它将在一行中返回，action_type 列列出了两次，但在左边，只显示 action_type = "sent"，而右边的列是 action_type = "open" 或 "null"

这是我最接近我想要的：

select email2.email_id, email2.recipient_id, email2.action_type as sent, 
opens.action_type as opened
  from email_activities_2 email2 
  left join email_activities_2 opens on email2.id = opens.id
  where email2.action_type = "sent"
union all
select email2.email_id, email2.recipient_id, email2.action_type as sent, 
opens.action_type as opened
  from email_activities_2 email2
  right join email_activities_2 opens on email2.id = opens.id
  where opens.action_type = "open"
  order by recipient_id, email_id asc;

这会返回：

email_id    |   recipient_id   |  sent  |   opened

94607              54             sent      sent
94607              54             open      open
94981              54             sent      sent
98479              54             sent      sent
98479              54             open      open

当我想要的是：

email_id    |   recipient_id   |  sent  |   opened

94607              54             sent      open
94981              54             sent      NULL
98479              54             sent      open

这可能吗？我是 SQL 的新手，我正在努力解决这个问题。

【问题讨论】：

您想查看“表别名”以创建表的虚拟“副本”，以便可以在查询中使用它两次。如果您有问题，请设置一个小提琴并发布它。

标签： mysql sql mysql-workbench self-join

【解决方案1】：

一种方法是条件聚合，不需要连接。

SELECT e.email_id
     , e.recipient_id
     , MAX(IF(e.action_type='sent',e.action_type,NULL)) AS sent
     , MAX(IF(e.action_type='open',e.action_type,NULL)) AS opened
  FROM email_activities_2 e
 WHERE e.action_type IN ('sent','open')
 GROUP
    BY e.email_id
     , e.recipient_id

GROUP BY 子句会将所有 email_id 和收件人 ID 值相等的行“折叠”成一行。 SELECT 子句中表达式中的条件测试将为每个组“挑选”出'open' 和'sent' 的值。 MAX 聚合将选择返回的最大非 NULL 值，并且仅当组中没有符合条件的行时才返回 NULL。

此查询有可能返回如下一行：

email_id    |   recipient_id   |  sent  |   opened
94999           54                NULL      open

在有open 的行但没有sent 的行的情况下。

如果需要删除这些行，可以添加 HAVING 子句：

HAVING MAX(IF(e.action_type='sent',e.action_type,NULL)) = 'sent'

如果您想从表中返回 email_id 和收件人的所有值，即使没有 action_type 为 'sent' 或 'open' 的行，您也可以完全省略 WHERE 子句。

【讨论】：

【解决方案2】：

我了解您希望获取所有已发送的电子邮件及其打开的活动（如果它们是打开的）查询应该是：

select email2.email_id, email2.recipient_id, email2.action_type as sent, 
opens.action_type as opened
  from email_activities_2 email2 
  left join (
select * from  email_activities_2
where email_activities_2.action_type = "open" 
) as opens on email2.id = opens.id
  where email2.action_type = "sent"

【讨论】：

【解决方案3】：

以下 SQL 应该可以为您完成这项工作：

Select email2.email_id,
       email2.recipient_id,
       MAX ( CASE WHEN email2.action_type = "sent" THEN email2.action_type  END ) As sent, 
       MAX (CASE WHEN opens.action_type = "open" THEN opens.action_type  END) As opened
  from email_activities_2 email2 
  left join email_activities_2 opens on email2.id = opens.id
  where email2.action_type = "sent" or opens.action_type = "open"
  group by email2.email_id
  order by recipient_id, email_id asc;

【讨论】：

【解决方案4】：

如果您只想显示已发送的电子邮件（或表格仅包含已发送的电子邮件），则 LEFT JOIN 解决方案适用于两种类型。

select
    sent.email_id,
    sent.recipient_id,
    sent.action_type as sent,
    open.action_type as opened
from email_activities_2 sent
left join email_activities_2 open
    on  open.email_id = sent.email_id
    and open.recipient_id = sent.recipient_id
    and open.action_type = 'open'
where sent.action_type = 'sent';

Demo

【讨论】：