SQL用于报告基于事件

Question

我有一个事件表是这样的：

----------------------------------------------------------- 
timestamp   | station | event_type | count 
----------------------------------------------------------- 
2013-02-22 01:00:00 | 1  | log_in  | -1 
2013-02-22 01:05:00 | 1  | alert  | 5 
2013-02-22 01:08:00 | 1  | alert  | 3 
2013-02-22 01:10:00 | 1  | log_out  | -1 
2013-02-22 01:30:00 | 2  | log_in  | -1 
2013-02-22 01:31:00 | 2  | alert  | 2 
2013-02-22 01:35:00 | 2  | log_out  | -1 
----------------------------------------------------------- 

我如何写SQL，产生这样的报告：

-------------------------------------------------------------------------- 
station | log_in    | log_out    | count 
-------------------------------------------------------------------------- 
1  | 2013-02-22 01:00:00 | 2013-02-22 01:10:00 | 8 
2  | 2013-02-22 01:30:00 | 2013-02-22 01:35:00 | 2 
-------------------------------------------------------------------------- 

基本上，我想总结注销登录和注销之间的警报。

有人能指点我正确的方向吗？

更新：下面的作品

JW的回答，但我做到了上面的表中的错误。对于警报事件，我不知道站点ID，所以表格应该如下所示。

----------------------------------------------------------- 
timestamp   | station | event_type | count 
----------------------------------------------------------- 
2013-02-22 01:00:00 | 1  | log_in  | -1 
2013-02-22 01:05:00 | -1  | alert  | 5 
2013-02-22 01:08:00 | -1  | alert  | 3 
2013-02-22 01:10:00 | 1  | log_out  | -1 
2013-02-22 01:30:00 | 2  | log_in  | -1 
2013-02-22 01:31:00 | -1  | alert  | 2 
2013-02-22 01:35:00 | 2  | log_out  | -1 
----------------------------------------------------------- 

Answer 1

SELECT station, 
     MIN(timestamp) login, 
     MAX(timestamp) logout, 
     SUM(CASE WHEN count > 0 THEN count ELSE 0 END) count 
FROM tbaleName 
GROUP BY station 

• SQLFiddle Demo

UPDATE 1

SELECT a.station, 
     a.login, 
     a.logout, 
     SUM(CASE WHEN count > 0 THEN count else 0 END) count 
FROM 
     (
      SELECT station, 
        MIN(timestamp) login, 
        MAX(timestamp) logout 
      FROM tableName2 
      GROUP BY station 
     ) a 
     INNER JOIN tableName2 b 
      ON b.timestamp BETWEEN a.logIN and a.logout AND 
      a.station <> -1 
GROUP BY a.station, 
      a.login, 
      a.logout 

• SQLFiddle Demo

为获得更快的性能，请在列timestamp上添加INDEX。