Question

对此感到头疼：

我需要计算（*）现在和1天之前两个表中的行数，其中表1中的historyId！=表2中的historyId和userId =？。然后我需要返回两个表的所有行的总和。

如果表1中的historyId与表2中的historyId相等，则只需将其计为一行，而不是两行。

查询一个

    SELECT HOUR(date) as hr, historyId, COUNT(*) as num_rows 
    FROM webHistory WHERE userId=? AND date BETWEEN (SYSDATE() - INTERVAL 1 DAY) 
    AND SYSDATE() GROUP BY HOUR(date);

查询两个

    SELECT HOUR(date) as hr, historyId, COUNT(*) as num_rows 
    FROM locationHistory WHERE userId=? AND date BETWEEN (SYSDATE() - INTERVAL 1 DAY) 
    AND SYSDATE() GROUP BY HOUR(date);

我一直在看JOIN，但我完全陷入困境，不知道甚至要搜索下一个，更不用说冒险和测试的路线了。

Answer 1

你可以在这里使用A UNION：

  SELECT HOUR(date) as hr, historyId, COUNT(*) as num_rows 
    FROM webHistory WHERE userId=? AND date BETWEEN (SYSDATE() - INTERVAL 1 DAY) 
    AND SYSDATE() GROUP BY HOUR(date)
 UNION
  SELECT HOUR(date) as hr, historyId, COUNT(*) as num_rows 
   FROM locationHistory WHERE userId=? AND date BETWEEN (SYSDATE() - INTERVAL 1 DAY) 
   AND SYSDATE() GROUP BY HOUR(date);

Answer 2

我认为你想使用union：

select count(distinct historyid)
from (select hour(date) as hr, historyid
      from webhistory
      where userid = ? and date between sysdate() - interval 1 day and sysdate()
      union all
      select hour(date) as hr, historyid
      from locationhistory
      where userid = ? and date between sysdate() - interval 1 day and sysdate()
     ) t
group by hr;

在行ID不匹配的两个表上计数（*）行。

2 个答案: