Oracle -删除重叠记录

Question

由于使用数据填充表时出错，表中现在出现了错误记录。特别是有重叠的时间戳。现在，我要删除这些记录，这些记录具有如下示例中所示的特征：

​

​

Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('1','3',to_timestamp('27.05.2016 17:27:08','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('31.05.2016 18:34:16','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('2','3',to_timestamp('31.05.2016 18:34:16','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('31.05.2016 18:34:26','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('3','3',to_timestamp('31.05.2016 18:34:26','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('01.06.2016 08:56:51','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('4','3',to_timestamp('01.06.2016 08:56:51','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('01.06.2016 18:38:30','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('5','3',to_timestamp('01.06.2016 18:38:30','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('08.06.2016 15:23:22','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('6','3',to_timestamp('08.06.2016 15:23:22','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('08.06.2016 21:02:21','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('7','3',to_timestamp('08.06.2016 17:09:02','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('08.06.2016 19:04:05','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('8','3',to_timestamp('08.06.2016 19:04:05','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('08.06.2016 21:02:21','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('9','3',to_timestamp('08.06.2016 21:02:21','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('09.09.9999 00:00:00','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('10','6',to_timestamp('20.06.2016 21:23:48','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('20.06.2016 21:23:56','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('11','6',to_timestamp('20.06.2016 21:23:56','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('21.06.2016 08:59:16','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('12','6',to_timestamp('21.06.2016 08:22:35','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('21.06.2016 08:23:02','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('13','6',to_timestamp('21.06.2016 08:23:02','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('21.06.2016 08:59:16','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('14','6',to_timestamp('21.06.2016 08:59:16','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('21.06.2016 12:06:49','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('15','6',to_timestamp('21.06.2016 12:06:49','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('22.06.2016 06:29:51','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('16','6',to_timestamp('22.06.2016 06:29:51','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('22.06.2016 08:18:18','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('17','6',to_timestamp('22.06.2016 08:18:18','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('22.06.2016 08:24:58','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('18','6',to_timestamp('22.06.2016 08:24:58','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('22.06.2016 08:38:18','DD.MM.RR HH24:MI:SSXFF'));
Insert into TEST_OVERLAP (TICKET_SK,TICKET_ID,VALID_FROM,VALID_TO) values ('19','6',to_timestamp('22.06.2016 08:38:18','DD.MM.RR HH24:MI:SSXFF'),to_timestamp('09.09.9999 00:00:00','DD.MM.RR HH24:MI:SSXFF'));

要删除的记录数量也可能大于每个ticket_id的2条。

这是怎么做到的呢？

Answer 1

如果我正确理解，您希望删除所有具有相同ticket_id记录和完全包含该记录的时间框架的记录。

下面是一个执行此任务的查询：

DELETE tickets t
WHERE EXISTS (
    SELECT *
    FROM tickets t2
    WHERE t2.ticket_id  =  t.ticket_id
      AND t2.valid_from <  t.valid_from
      AND t2.valid_to   >= t.valid_to
);