Question

我在MySQL中有五个表，我正在生成一些报告。为了显示目的，我将略微抽象表名。让我们假设我正在使用的表格如下所述：

History  (id, serial, date, time) : Stores historic data, this is the main
                                    column we query.
Serial   (id, serial)             : Stores serial numbers used in the next
                                    3 tables.
Employee (id, serial_id, name)
Event    (id, serial_id, name)
Location (id, serial_id, name)

编辑：重写整个问题，希望它更清晰，更容易理解......

我需要查询事件，如果在历史记录中直接找到员工序列 PRIOR 到具有与事件匹配的序列的行，该员工将与该事件相关联，同样，如果在历史记录中直接找到一个位置 AFTER 具有与事件匹配的序列的行，该位置将与该事件相关联。

示例数据：

History:
+------+------------+------------+----------+
| id   | serial     | date       | time     |
+------+------------+------------+----------+
| 1    | 00001234   | 2012-09-25 | 09:00:00 |
+------+------------+------------+----------+
| 2    | 00001235   | 2012-09-25 | 09:01:00 |
+------+------------+------------+----------+
| 3    | 00001236   | 2012-09-25 | 09:02:00 |
+------+------------+------------+----------+
| 4    | 00001235   | 2012-09-25 | 09:05:00 |
+------+------------+------------+----------+
| 5    | 00001236   | 2012-09-25 | 09:07:00 |
+------+------------+------------+----------+
| 6    | 00001235   | 2012-09-25 | 09:10:00 |
+------+------------+------------+----------+
| 7    | 00001235   | 2012-09-25 | 09:11:00 |
+------+------------+------------+----------+
| 8    | 00001235   | 2012-09-25 | 09:12:00 |
+------+------------+------------+----------+
| 9    | 00001236   | 2012-09-25 | 09:15:00 |
+------+------------+------------+----------+
| 10   | 00001234   | 2012-09-25 | 09:20:00 |
+------+------------+------------+----------+
| 11   | 00001236   | 2012-09-25 | 09:25:00 |
+------+------------+------------+----------+

Serial:
+----------+------------+
| id       | serial     |
+----------+------------+
| 1        | 00001234   |
+----------+------------+
| 2        | 00001235   |
+----------+------------+
| 3        | 00001236   |
+----------+------------+

Employee:
+----------+------------+-------------+
| id       | serial_id  | name        |
+----------+------------+-------------+
| 1        | 1          | John Smith  |
+----------+------------+-------------+

Event:
+----------+------------+-------------+
| id       | serial_id  | name        |
+----------+------------+-------------+
| 1        | 2          | Event 1     |
+----------+------------+-------------+

Location:
+----------+------------+-------------+
| id       | serial_id  | name        |
+----------+------------+-------------+
| 1        | 3          | Location 1  |
+----------+------------+-------------+

如果我要进行正常的LEFT JOIN查询（只是为了清晰起见结果的差异），使用以下查询，结果将是：

SELECT
    history.id,
    history.serial,
    history.date,
    history.time,
    employee.name as 'employee',
    event.name as 'event',
    location.name as 'location'
FROM history
LEFT JOIN serial ON history.serial = serial.serial
LEFT JOIN employee ON serial.id = employee.serial_id
LEFT JOIN event ON serial.id = event.serial_id
LEFT JOIN location ON location.id = location.serial_id
ORDER BY date, time

正常LEFT JOIN的结果是（仅为了清楚起见，我不需要这个结果集）：

+----+----------+------------+----------+------------+----------+------------+
| id | serial   | date       | time     | employee   | event    | location   |
+----+----------+------------+----------+------------+----------+------------+
| 1  | 00001234 | 2012-09-25 | 09:00:00 | John Smith | NULL     | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 2  | 00001235 | 2012-09-25 | 09:01:00 | NULL       | Event 1  | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 3  | 00001236 | 2012-09-25 | 09:02:00 | NULL       | NULL     | Location 1 |
+----+----------+------------+----------+------------+----------+------------+
| 4  | 00001235 | 2012-09-25 | 09:05:00 | NULL       | Event 1  | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 5  | 00001236 | 2012-09-25 | 09:07:00 | NULL       | NULL     | Location 1 |
+----+----------+------------+----------+------------+----------+------------+
| 6  | 00001235 | 2012-09-25 | 09:10:00 | NULL       | Event 1  | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 7  | 00001235 | 2012-09-25 | 09:11:00 | NULL       | Event 1  | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 8  | 00001235 | 2012-09-25 | 09:12:00 | NULL       | Event 1  | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 9  | 00001236 | 2012-09-25 | 09:15:00 | NULL       | NULL     | Location 1 |
+-----+---------+------------+----------+------------+----------+------------+
| 10 | 00001234 | 2012-09-25 | 09:20:00 | John Smith | NULL     | NULL       |
+----+----------+------------+----------+------------+----------+------------+
| 11 | 00001236 | 2012-09-25 | 09:25:00 | NULL       | NULL     | Location 1 |
+----+----------+------------+----------+------------+----------+------------+

查询应基于事件，事件永远不应为NULL。它应该查看行 PRIOR 并将其标记为employee（如果不匹配则为null），它应该查看 AFTER 行并将其标记为位置（如果为没有比赛）。结果应该是什么样的例子，使用上面的数据：

Result should be:
+----------+------------+------------+------------+----------+------------+
| e_serial | date       | time       | employee   | event    | location   |
+----------+------------+------------+------------+----------+------------+
| 00001235 | 2012-09-25 | 09:01:00   | John Smith | Event 1  | Location 1 |
+----------+------------+------------+------------+----------+------------+
| 00001235 | 2012-09-25 | 09:05:00   | NULL       | Event 1  | Location 1 |
+----------+------------+------------+------------+----------+------------+
| 00001235 | 2012-09-25 | 09:10:00   | NULL       | Event 1  | NULL       |
+----------+------------+------------+------------+----------+------------+
| 00001235 | 2012-09-25 | 09:11:00   | NULL       | Event 1  | NULL       |
+----------+------------+------------+------------+----------+------------+
| 00001235 | 2012-09-25 | 09:12:00   | NULL       | Event 1  | Location 1 |
+----------+------------+------------+------------+----------+------------+

我不确定这是否可能通过一些子选择或其他东西，我认为这是我的SQL联盟。目前我使用的是应用程序逻辑，但认为仅使用SQL处理它可能更干净，更快。

任何帮助都会很棒。

Answer 1

SELECT serial_id AS `serial`,
(SELECT `date` FROM History WHERE serial_id= event2.serial_id ) AS `date`,
(SELECT `time` FROM History WHERE serial_id= event2.serial_id ) AS `time`,
(SELECT `name` FROM event1 WHERE serial_id= event2.serial_id ) AS `event1`,`name` AS event2,
(SELECT `name` FROM event3 WHERE serial_id= event2.serial_id ) AS `event1`
FROM event2

您的历史记录表应为History (id, serial_id, date, time)，必须包含serial_id而不是serial name

对于你的场景，希望它能正常工作

SELECT serial_id AS `serial`,
(SELECT `date` FROM History WHERE `serial` = (SELECT `serial` FROM `serial` WHERE id = event2.serial_id LIMIT 0,1)) AS `date`,
(SELECT `time` FROM History WHERE `serial`= (SELECT `serial` FROM `serial` WHERE id = event2.serial_id LIMIT 0,1)) AS `time`,
(SELECT `name` FROM event1 WHERE serial_id= event2.serial_id ) AS `event1`,
`name` AS event2,
(SELECT `name` FROM event3 WHERE serial_id= event2.serial_id ) AS `event1`
FROM event2

Answer 2

如果其他人遇到此问题，请在此处发布解决方案。感谢dianuj引导我朝着正确的方向前进。

SELECT
    serial.serial as 'e_serial',
    @dt:=TIMESTAMP(history.date, history.time) as 'date_time',
    history.date,
    history.time,
    (SELECT
        employee.name
    FROM
        history
        LEFT JOIN serial ON history.serial = serial.serial
        LEFT JOIN employee ON employee.serial_id = serial.id
    WHERE
        TIMESTAMP(history.date, history.time) = (
            SELECT
                max(TIMESTAMP(history.date, history.time))
            FROM
                history
            WHERE TIMESTAMP(history.date, history.time) < @dt
        )
    ) as 'employee',
    event.name as 'event',
    (SELECT
        location.name
    FROM
        history
        LEFT JOIN serial ON history.serial = serial.serial
        LEFT JOIN location ON location.serial_id = serial.id
    WHERE
        TIMESTAMP(history.date, history.time) > @dt
    LIMIT 1
    ) as 'location'
FROM history
LEFT JOIN serial ON history.serial = serial.serial
LEFT JOIN event ON serial.id = event.serial_id
WHERE event.name IS NOT NULL
ORDER BY history.date, history.time

结果

+----------+---------------------+------------+----------+------------+---------+------------+
| e_serial | date_time           | date       | time     | employee   | event   | location   |
+----------+---------------------+------------+----------+------------+---------+------------+
| 00001235 | 2012-09-25 09:01:00 | 2012-09-25 | 09:01:00 | John Smith | Event 1 | Location 1 |
| 00001235 | 2012-09-25 09:05:00 | 2012-09-25 | 09:05:00 | NULL       | Event 1 | Location 1 |
| 00001235 | 2012-09-25 09:10:00 | 2012-09-25 | 09:10:00 | NULL       | Event 1 | NULL       |
| 00001235 | 2012-09-25 09:11:00 | 2012-09-25 | 09:11:00 | NULL       | Event 1 | NULL       |
| 00001235 | 2012-09-25 09:12:00 | 2012-09-25 | 09:12:00 | NULL       | Event 1 | Location 1 |
+----------+---------------------+------------+----------+------------+---------+------------+
5 rows in set (0.01 sec)

MySQL在查询中获取当前行之前和之后的行

2 个答案: