Mysql 如何结合联接高效地预筛选行？
我需要计算表中过滤行之间的时间戳 我使用的联接基于另一个问题的有用答案： 堆栈溢出： 我的问题是，我的表包含来自多个不同对象的混合数据，我希望在执行联接之前，我需要首先对object_id=blah进行筛选，因为如果我不进行预筛选，则联接将使时间戳与不相关的object_id有所不同，因为它只是逐行连续查看 我需要连接根据前面出现的object_id=blah，而不仅仅是前一行，来进行时间戳差异 正在寻找如何最有效地解决此查询。 提前谢谢=D 我已经尝试在连接之后添加WHERE object_id=blah，得到了正确的行数，但是时间戳差异仍然只是基于每个时间戳差异中的前一行 我真的希望WHERE在连接发生之前应用于源表，但显然这不是它的工作方式-Mysql 如何结合联接高效地预筛选行？,mysql,join,Mysql,Join,我需要计算表中过滤行之间的时间戳 我使用的联接基于另一个问题的有用答案： 堆栈溢出： 我的问题是，我的表包含来自多个不同对象的混合数据，我希望在执行联接之前，我需要首先对object_id=blah进行筛选，因为如果我不进行预筛选，则联接将使时间戳与不相关的object_id有所不同，因为它只是逐行连续查看 我需要连接根据前面出现的object_id=blah，而不仅仅是前一行，来进行时间戳差异 正在寻找如何最有效地解决此查询。 提前谢谢=D 我已经尝试在连接之后添加WHERE object_i
SELECT
t1.scanid, t1.event_id, t1.objectect_id, t1.object_timestamp, t1.object_counter,
TIMEDIFF(t1.object_timestamp, t2.object_timestamp) AS diff
FROM event_data AS t1
LEFT JOIN event_data AS t2 ON ( t2.scanid = t1.scanid - 1);
--+---------------+--------------+------+-----+----------------------+----------------+
--| Field | Type | Null | Key | Default | Extra |
--+---------------+--------------+------+-----+----------------------+----------------+
--| scanid | int(11) | NO | PRI | NULL | auto_increment |
--| event_id | int(12) | NO | | NULL | |
--| objectect_id | int(11) | NO | | NULL | |
--| obj_timestamp | timestamp(3) | NO | | CURRENT_TIMESTAMP(3) | |
--| obj_counter | int(11) | YES | | -1 | |
--+---------------+--------------+------+-----+----------------------+----------------+
SELECT
t1.scanid, t1.event_id, t1.objectect_id, t1.object_timestamp, t1.object_counter,
TIMEDIFF(t1.object_timestamp, t2.object_timestamp) AS diff
FROM event_data AS t1
LEFT JOIN event_data AS t2 ON ( t2.scanid = t1.scanid - 1)
WHERE t1.object_id = 2;
+--------+----------+------------+-------------------------+-------------+--------------+
| scanid | event_id | object_id | obj_timestamp | obj_counter | diff |
+--------+----------+------------+-------------------------+-------------+--------------+
| 1 | 1 | 2 | 2019-02-17 13:11:02.425 | 0 | NULL |
| 16 | 1 | 2 | 2019-02-17 13:16:24.326 | 1 | 00:02:56.416 |
| 17 | 1 | 2 | 2019-02-17 13:16:25.362 | 2 | 00:00:01.036 |
| 18 | 1 | 2 | 2019-02-17 13:19:48.318 | 3 | 00:03:22.956 |
| 19 | 1 | 2 | 2019-02-17 13:25:01.604 | 4 | 00:05:13.286 |
| 20 | 1 | 2 | 2019-02-17 13:30:17.024 | 5 | 00:05:15.420 |
| 23 | 1 | 2 | 2019-02-17 13:41:12.324 | 6 | 00:01:51.628 |
| 24 | 1 | 2 | 2019-02-17 13:41:13.349 | 7 | 00:00:01.025 |
| 27 | 1 | 2 | 2019-02-17 13:41:18.467 | 8 | 00:00:01.031 |
| 31 | 1 | 2 | 2019-02-17 13:41:23.591 | 9 | 00:00:01.028 |
| 32 | 1 | 2 | 2019-02-17 13:41:24.619 | 10 | 00:00:01.028 |
+--------+----------+------------+-------------------------+-------------+--------------+
11 rows in set (0.00 sec)
+--------+----------+------------+-------------------------+-------------+
| scanid | event_id | object_id | obj_timestamp | obj_counter |
+--------+----------+------------+-------------------------+-------------+
| 1 | 1 | 2 | 2019-02-17 13:11:02.425 | 0 |
| 2 | 1 | 0 | 2019-02-17 13:11:08.227 | 0 |
| 3 | 1 | 0 | 2019-02-17 13:11:12.303 | 1 |
| 4 | 1 | 0 | 2019-02-17 13:11:31.383 | 2 |
| 5 | 1 | 0 | 2019-02-17 13:11:32.417 | 3 |
| 6 | 1 | 0 | 2019-02-17 13:11:33.451 | 4 |
| 7 | 1 | 0 | 2019-02-17 13:11:34.839 | 5 |
| 8 | 1 | 0 | 2019-02-17 13:11:35.868 | 6 |
| 9 | 1 | 0 | 2019-02-17 13:12:05.143 | 7 |
| 10 | 1 | 0 | 2019-02-17 13:13:08.733 | 8 |
| 11 | 1 | 0 | 2019-02-17 13:13:11.169 | 9 |
| 12 | 1 | 0 | 2019-02-17 13:13:22.239 | 10 |
| 13 | 1 | 0 | 2019-02-17 13:13:24.256 | 11 |
| 14 | 1 | 0 | 2019-02-17 13:13:26.875 | 12 |
| 15 | 1 | 0 | 2019-02-17 13:13:27.910 | 13 |
| 16 | 1 | 2 | 2019-02-17 13:16:24.326 | 1 |
| 17 | 1 | 2 | 2019-02-17 13:16:25.362 | 2 |
| 18 | 1 | 2 | 2019-02-17 13:19:48.318 | 3 |
| 19 | 1 | 2 | 2019-02-17 13:25:01.604 | 4 |
| 20 | 1 | 2 | 2019-02-17 13:30:17.024 | 5 |
| 21 | 1 | 0 | 2019-02-17 13:39:19.664 | 14 |
| 22 | 1 | 0 | 2019-02-17 13:39:20.696 | 15 |
| 23 | 1 | 2 | 2019-02-17 13:41:12.324 | 6 |
| 24 | 1 | 2 | 2019-02-17 13:41:13.349 | 7 |
| 25 | 1 | 0 | 2019-02-17 13:41:14.381 | 16 |
| 26 | 1 | 0 | 2019-02-17 13:41:17.436 | 17 |
| 27 | 1 | 2 | 2019-02-17 13:41:18.467 | 8 |
| 28 | 1 | 0 | 2019-02-17 13:41:20.503 | 18 |
| 29 | 1 | 0 | 2019-02-17 13:41:21.535 | 19 |
| 30 | 1 | 0 | 2019-02-17 13:41:22.563 | 20 |
| 31 | 1 | 2 | 2019-02-17 13:41:23.591 | 9 |
| 32 | 1 | 2 | 2019-02-17 13:41:24.619 | 10 |
+--------+----------+------------+-------------------------+-------------+
32 rows in set (0.00 sec)
+--------+----------+------------+-------------------------+-------------+--------------+
| scanid | event_id | object_id | obj_timestamp | obj_counter | diff |
+--------+----------+------------+-------------------------+-------------+--------------+
| 1 | 1 | 2 | 2019-02-17 13:11:02.425 | 0 | NULL |
| 2 | 1 | 0 | 2019-02-17 13:11:08.227 | 0 | 00:00:05.802 |
| 3 | 1 | 0 | 2019-02-17 13:11:12.303 | 1 | 00:00:04.076 |
| 4 | 1 | 0 | 2019-02-17 13:11:31.383 | 2 | 00:00:19.080 |
| 5 | 1 | 0 | 2019-02-17 13:11:32.417 | 3 | 00:00:01.034 |
| 6 | 1 | 0 | 2019-02-17 13:11:33.451 | 4 | 00:00:01.034 |
| 7 | 1 | 0 | 2019-02-17 13:11:34.839 | 5 | 00:00:01.388 |
| 8 | 1 | 0 | 2019-02-17 13:11:35.868 | 6 | 00:00:01.029 |
| 9 | 1 | 0 | 2019-02-17 13:12:05.143 | 7 | 00:00:29.275 |
| 10 | 1 | 0 | 2019-02-17 13:13:08.733 | 8 | 00:01:03.590 |
| 11 | 1 | 0 | 2019-02-17 13:13:11.169 | 9 | 00:00:02.436 |
| 12 | 1 | 0 | 2019-02-17 13:13:22.239 | 10 | 00:00:11.070 |
| 13 | 1 | 0 | 2019-02-17 13:13:24.256 | 11 | 00:00:02.017 |
| 14 | 1 | 0 | 2019-02-17 13:13:26.875 | 12 | 00:00:02.619 |
| 15 | 1 | 0 | 2019-02-17 13:13:27.910 | 13 | 00:00:01.035 |
| 16 | 1 | 2 | 2019-02-17 13:16:24.326 | 1 | 00:02:56.416 |
| 17 | 1 | 2 | 2019-02-17 13:16:25.362 | 2 | 00:00:01.036 |
| 18 | 1 | 2 | 2019-02-17 13:19:48.318 | 3 | 00:03:22.956 |
| 19 | 1 | 2 | 2019-02-17 13:25:01.604 | 4 | 00:05:13.286 |
| 20 | 1 | 2 | 2019-02-17 13:30:17.024 | 5 | 00:05:15.420 |
| 21 | 1 | 0 | 2019-02-17 13:39:19.664 | 14 | 00:09:02.640 |
| 22 | 1 | 0 | 2019-02-17 13:39:20.696 | 15 | 00:00:01.032 |
| 23 | 1 | 2 | 2019-02-17 13:41:12.324 | 6 | 00:01:51.628 |
| 24 | 1 | 2 | 2019-02-17 13:41:13.349 | 7 | 00:00:01.025 |
| 25 | 1 | 0 | 2019-02-17 13:41:14.381 | 16 | 00:00:01.032 |
| 26 | 1 | 0 | 2019-02-17 13:41:17.436 | 17 | 00:00:03.055 |
| 27 | 1 | 2 | 2019-02-17 13:41:18.467 | 8 | 00:00:01.031 |
| 28 | 1 | 0 | 2019-02-17 13:41:20.503 | 18 | 00:00:02.036 |
| 29 | 1 | 0 | 2019-02-17 13:41:21.535 | 19 | 00:00:01.032 |
| 30 | 1 | 0 | 2019-02-17 13:41:22.563 | 20 | 00:00:01.028 |
| 31 | 1 | 2 | 2019-02-17 13:41:23.591 | 9 | 00:00:01.028 |
| 32 | 1 | 2 | 2019-02-17 13:41:24.619 | 10 | 00:00:01.028 |
+--------+----------+------------+-------------------------+-------------+--------------+
32 rows in set (0.01 sec)
SELECT
t1.scanid, t1.event_id, t1.objectect_id, t1.object_timestamp, t1.object_counter,
TIMEDIFF(t1.object_timestamp, t2.object_timestamp) AS diff
FROM event_data AS t1
LEFT JOIN event_data AS t2 ON ( t2.scanid = t1.scanid - 1)
WHERE t1.object_id = 2;
+--------+----------+------------+-------------------------+-------------+--------------+
| scanid | event_id | object_id | obj_timestamp | obj_counter | diff |
+--------+----------+------------+-------------------------+-------------+--------------+
| 1 | 1 | 2 | 2019-02-17 13:11:02.425 | 0 | NULL |
| 16 | 1 | 2 | 2019-02-17 13:16:24.326 | 1 | 00:02:56.416 |
| 17 | 1 | 2 | 2019-02-17 13:16:25.362 | 2 | 00:00:01.036 |
| 18 | 1 | 2 | 2019-02-17 13:19:48.318 | 3 | 00:03:22.956 |
| 19 | 1 | 2 | 2019-02-17 13:25:01.604 | 4 | 00:05:13.286 |
| 20 | 1 | 2 | 2019-02-17 13:30:17.024 | 5 | 00:05:15.420 |
| 23 | 1 | 2 | 2019-02-17 13:41:12.324 | 6 | 00:01:51.628 |
| 24 | 1 | 2 | 2019-02-17 13:41:13.349 | 7 | 00:00:01.025 |
| 27 | 1 | 2 | 2019-02-17 13:41:18.467 | 8 | 00:00:01.031 |
| 31 | 1 | 2 | 2019-02-17 13:41:23.591 | 9 | 00:00:01.028 |
| 32 | 1 | 2 | 2019-02-17 13:41:24.619 | 10 | 00:00:01.028 |
+--------+----------+------------+-------------------------+-------------+--------------+
11 rows in set (0.00 sec)
从您的数据来看，表的合适连接条件实际上是t2.obj_counter=t1.obj_counter-1和t2.object_id=t1.object_id；这将确保只比较与给定对象相关的时间戳。因此，您的查询将基于示例数据：
SELECT
t1.scanid, t1.event_id, t1.object_id, t1.obj_timestamp, t1.obj_counter,
TIMEDIFF(t1.obj_timestamp, t2.obj_timestamp) AS diff
FROM event_data AS t1
LEFT JOIN event_data AS t2 ON t2.obj_counter = t1.obj_counter - 1 AND t2.object_id = t1.object_id
WHERE t1.object_id = 2
ORDER BY t1.obj_counter
scanid event_id object_id obj_timestamp obj_counter diff
1 1 2 2019-02-17 13:11:02 0 null
16 1 2 2019-02-17 13:16:24 1 00:05:22
17 1 2 2019-02-17 13:16:25 2 00:00:01
18 1 2 2019-02-17 13:19:48 3 00:03:23
19 1 2 2019-02-17 13:25:02 4 00:05:14
20 1 2 2019-02-17 13:30:17 5 00:05:15
23 1 2 2019-02-17 13:41:12 6 00:10:55
24 1 2 2019-02-17 13:41:13 7 00:00:01
27 1 2 2019-02-17 13:41:18 8 00:00:05
31 1 2 2019-02-17 13:41:24 9 00:00:06
32 1 2 2019-02-17 13:41:25 10 00:00:01
非常感谢，尼克！你的解决方案很棒=D@LeonShaner别担心。很高兴我能提供帮助。快速跟进，因为所有的上下文都已经在这里了=D如果我想将TIMEDIFF结果存储到一个本地变量中，除了存储为diff，这样我就可以在计算连接中的另一列时使用本地变量，那该怎么办？我搞不懂语法。我可以重复TIMEDIFF而不是引用本地var，但这似乎效率低下。我将在下一条注释中发布我的低效代码。@LeonShaner除非您将此SELECT转换为子查询，否则您不能在同一SELECT中使用别名，因此您必须重复此操作TIMEDIFF@LeonShaner除非您编写子查询，否则必须这样做，例如选择scanid、event_id、object_id、obj_时间戳、obj_计数器、diff、，diff/2作为hdiff从中选择t1.scanid、t1.event_id、t1.object_id、t1.obj_时间戳、t1.obj_计数器、TIMEDIFFt1.obj_时间戳、t2.obj_时间戳作为diff从事件_数据作为t1左连接事件_数据作为t2。obj_计数器=t1.obj_计数器-1和t2.object_id=t1.object_计数器，其中t1.object_id=2按t1.obj_计数器作为diffs