Question

我无法解决此查询问题，以获取两个处于OFF状态的两个不同IP地址的公共持续时间。

以下示例数据中的案例

1（简单情况） - 对于00:10:10到00：20：00，IP地址“10.0.1.2”保持为OFF，在此持续时间内“10.0.1.3”也为OFF，因此常用的持续时间为OFF两者都是00:10:10到00:20:00。

2（有问题） - IP地址“10.0.1.2”在13:00:00~13：25：00关闭，如果和其他IP地址核对的话，12：55：00~13为OFF ：20：00。因此，两者的共同持续时间为13:00:00至13:20:00。

示例数据：

ID   IP address  Status   Time
----------------------------------
1    10.0.1.2    OFF      00:10:00
1    10.0.1.2    ON       00:20:00
1    10.0.1.2    OFF      11:00:00
1    10.0.1.2    ON       11:20:00
1    10.0.1.2    OFF      13:00:00
1    10.0.1.2    ON       13:25:00
1    10.0.1.2    OFF      14:05:00
1    10.0.1.2    ON       14:10:00
1    10.0.1.2    OFF      15:35:00
1    10.0.1.2    ON       15:45:00
1    10.0.1.3    OFF      00:10:00
1    10.0.1.3    ON       00:20:00
1    10.0.1.3    OFF      11:05:00
1    10.0.1.3    ON       11:25:00
1    10.0.1.3    OFF      12:55:00
1    10.0.1.3    ON       13:20:00
1    10.0.1.3    OFF      17:10:00
1    10.0.1.3    ON       17:15:00
1    10.0.1.3    OFF      15:00:00
1    10.0.1.3    ON       16:45:00

输出：

ID   IP addresses       Status  Time
-----------------------------------------
1    10.0.1.3,10.0.1.2  OFF      00:10:00
1    10.0.1.3,10.0.1.2  ON       00:20:00
1    10.0.1.3,10.0.1.2  OFF      11:05:00
1    10.0.1.3,10.0.1.2  ON       11:20:00
1    10.0.1.3,10.0.1.2  OFF      13:00:00
1    10.0.1.3,10.0.1.2  ON       13:20:00
1    10.0.1.3,10.0.1.2  OFF      15:35:00
1    10.0.1.3,10.0.1.2  ON       15:45:00

Answer 1

这是你的首发。

为了便于阅读，我将IP地址缩减为int ip。
我将状态更改为文字。它应该是布尔值，如果MySQL没有，那么可能是char（1）或带有CHECK约束的int。
您需要考虑一些约束或唯一索引来保证状态切换并防止在它已经打开时将其打开（多次打开）？
声明适当的索引以加快查询速度。否则它是二次复杂度。

CREATE TABLE foo (ip int NOT NULL, status text NOT NULL,
    ts time NOT NULL, PRIMARY KEY (ip, status, ts));

INSERT INTO foo VALUES
(2, 'OFF', '00:10:00'),
(2, 'ON',  '00:20:00'),
(2, 'OFF', '11:00:00'),
(2, 'ON',  '11:20:00'),
(2, 'OFF', '13:00:00'),
(2, 'ON',  '13:25:00'),
(2, 'OFF', '14:05:00'),
(2, 'ON',  '14:10:00'),
(2, 'OFF', '15:35:00'),
(2, 'ON',  '15:45:00'),
(3, 'OFF', '00:10:00'),
(3, 'ON',  '00:20:00'),
(3, 'OFF', '11:05:00'),
(3, 'ON',  '11:25:00'),
(3, 'OFF', '12:55:00'),
(3, 'ON',  '13:20:00'),
(3, 'OFF', '17:10:00'),
(3, 'ON',  '17:15:00'),
(3, 'OFF', '15:00:00'),
(3, 'ON',  '16:45:00');

假设您在MySQL中使用了公用表表达式CTE（除此之外，您没有指定版本）。

如果您没有CTE，那么只需复制并替换CTE的所有引用（在本例中为off）并为其命名。最后一个示例不会使用WITH。

WITH off AS
(SELECT ip,
        ts "off_from",
        (SELECT ts FROM foo
         WHERE ip = a.ip AND a.ts <= ts AND status = 'ON'
         ORDER BY ts ASC LIMIT 1) "off_until"
 FROM foo a WHERE status = 'OFF'
)
SELECT * FROM off;

哪个给出了

 ip | off_from | off_until
----+----------+-----------
  2 | 00:10:00 | 00:20:00
  2 | 11:00:00 | 11:20:00
  2 | 13:00:00 | 13:25:00
  2 | 14:05:00 | 14:10:00
  2 | 15:35:00 | 15:45:00
  3 | 00:10:00 | 00:20:00
  3 | 11:05:00 | 11:25:00
  3 | 12:55:00 | 13:20:00
  3 | 17:10:00 | 17:15:00
  3 | 15:00:00 | 16:45:00

WITH off AS
(SELECT ip,
        ts "off_from",
        (SELECT ts FROM foo
         WHERE ip = a.ip AND a.ts <= ts AND status = 'ON'
         ORDER BY ts ASC LIMIT 1) "off_until"
 FROM foo a WHERE status = 'OFF'
)
SELECT *
FROM off x
INNER JOIN off y
ON  x.off_from <= y.off_from AND y.off_from < x.off_until
AND x.ip <> y.ip ;

 ip | off_from | off_until | ip | off_from | off_until
----+----------+-----------+----+----------+-----------
  2 | 00:10:00 | 00:20:00  |  3 | 00:10:00 | 00:20:00
  2 | 11:00:00 | 11:20:00  |  3 | 11:05:00 | 11:25:00
  3 | 00:10:00 | 00:20:00  |  2 | 00:10:00 | 00:20:00
  3 | 12:55:00 | 13:20:00  |  2 | 13:00:00 | 13:25:00
  3 | 15:00:00 | 16:45:00  |  2 | 15:35:00 | 15:45:00

要获得最小和最长时间使用

WITH off AS
(SELECT ip,
        ts "off_from",
        (SELECT ts FROM foo
         WHERE ip = a.ip AND a.ts <= ts AND status = 'ON'
         ORDER BY ts ASC LIMIT 1) "off_until"
 FROM foo a WHERE status = 'OFF'
)
SELECT x.ip "ip_a", y.ip "ip_b",
       greatest( x.off_from, y.off_from ) "off_from",
       least( x.off_until, y.off_until ) "off_until"
FROM off x
INNER JOIN off y
ON  x.off_from <= y.off_from AND y.off_from < x.off_until
AND x.ip <> y.ip ;

产生

 ip_a | ip_b | off_from | off_until
------+------+----------+-----------
    2 |    3 | 00:10:00 | 00:20:00
    2 |    3 | 11:05:00 | 11:20:00
    3 |    2 | 00:10:00 | 00:20:00
    3 |    2 | 13:00:00 | 13:20:00
    3 |    2 | 15:35:00 | 15:45:00

没有WITH（复制粘贴并命名CTE）。

SELECT x.ip "ip_a", y.ip "ip_b",
       greatest( x.off_from, y.off_from ) "off_from",
       least( x.off_until, y.off_until ) "off_until"
FROM
(SELECT ip,
        ts "off_from",
        (SELECT ts
         FROM foo
         WHERE ip = a.ip AND a.ts <= ts AND status = 'ON'
         ORDER BY ts ASC LIMIT 1) "off_until"
 FROM foo a WHERE status = 'OFF'
) x
INNER JOIN
(SELECT ip,
        ts "off_from",
        (SELECT ts
         FROM foo
         WHERE ip = a.ip AND a.ts <= ts AND status = 'ON'
         ORDER BY ts ASC LIMIT 1) "off_until"
 FROM foo a WHERE status = 'OFF'
) y
ON  x.off_from <= y.off_from
AND y.off_from < x.off_until
AND x.ip <> y.ip ;

对于LIMIT 1的内部选择，请考虑(ip, status, ts)上的索引。

对于连接，您的DBMS可能会使用ts上的索引。 CTE（WITH子句）仅实现一次虚拟表。这可能不适用于复制粘贴CTE几次（这里是两次）。

这对你来说应该是一个粗略的启动。到目前为止还不是完美或最好的解决方案。可能还有其他更好的。

Answer 2

一种方法是使用TIME_TO_SEC()将时间作为秒，并计算存储过程中的差异：

Create table common_duration (
ip varchar (10),
start_time time,
end_time time
)

CREATE PROCEDURE `comm_time`()
    BEGIN
    DECLARE curs1 CURSOR FOR SELECT `IP`, TIME_TO_SEC(`time`) as time, STATUS FROM TABLE;
    DECLARE ip varchar(20);
    DECLARE iptime time;
    DECLARE ipstime time;
    DECLARE ipstatus varchar(10);
    OPEN curs1;
    FETCH curs1 INTO ip,iptime,ipstatus;
    if (status='ON')
    insert into `common_duration`(ip, start_time, end_time) values(ip, ipstime, iptime);
    else
    ipstime=iptime;
    endif;
    CLOSE curs1;
    SELECT t1.ip SEC_TO_TIME(t1.end_time-t1.start_time) as time_duration FROM `common_duration t1, `common_duration t2
    WHERE t1.time_duration= t2.time_duration
          AND t1.ip != t2.ip;
    End

MySQL查询以获得常见的持续时间

2 个答案: