Question

嗨我在postgres中有两个表，它们具有如下所示的相似结构

表A

╔═══════════════════════════════════════════════════╗
║ id | timestamp                |  type | category  ║
╠═══════════════════════════════════════════════════╣
║ 1  | 2014-02-11 01:10:34.1234 | A  | 1            ║
║ 2  | 2014-02-11 01:10:36.7533 | A  | 2            ║
║ 3  | 2014-02-11 01:10:39.914  | E  | 1            ║
║ 4  | 2014-02-11 01:10:42.6712 | E  | 3            ║
║ 5  | 2014-02-11 01:10:34.9123 | E  | 1            ║
║ 6  | 2014-02-11 01:10:36.0313 | C  | 2            ║
║ 7  | 2014-02-11 01:10:39.1234 | X  | 1            ║
║ 7  | 2014-02-11 01:10:40.6743 | X  | 1            ║
║ 8  | 2014-02-11 01:10:42.9092 | B  | 3            ║
║ 9  | 2014-02-11 01:10:43.8234 | T  | 1            ║
║ 10 | 2014-02-11 01:10:45.1566 | T  | 1            ║
║ 11 | 2014-02-11 01:10:58.7344 | T  | 1            ║
║ 12 | 2014-02-11 01:10:59.9232 | T  | 1            ║
║ 13 | 2014-02-11 01:10:59.9232 | T  | 3            ║
║ 14 | 2014-02-11 01:10:59.9232 | T  | 2            ║
║ 15 | 2014-02-11 01:10:59.9232 | T  | 2            ║
╚═══════════════════════════════════════════════════╝

表B

╔═══════════════════════════════════════════════════╗
║ id | timestamp                |  type | category  ║
╠═══════════════════════════════════════════════════╣
║ 1  | 2014-02-11 01:10:34.123  | A  | 1            ║
║ 2  | 2014-02-11 01:10:35.9092 | A  | 2            ║
║ 3  | 2014-02-11 01:10:36.1234 | E  | 1            ║
║ 4  | 2014-02-11 01:10:40.0100 | E  | 3            ║
║ 5  | 2014-02-11 01:10:51.1234 | E  | 2            ║
║ 7  | 2014-02-11 01:10:54.5347 | X  | 1            ║
║ 8  | 2014-02-11 01:11:02.7914 | B  | 3            ║
║ 9  | 2014-02-11 01:11:03.9000 | T  | 1            ║
║ 10 | 2014-02-11 01:11:05.7829 | T  | 1            ║
║ 11 | 2014-02-11 01:11:06.125  | T  | 1            ║
║ 12 | 2014-02-11 01:11:10.0000 | T  | 1            ║
║ 13 | 2014-02-11 01:10:59.9232 | T  | 3            ║
║ 14 | 2014-02-11 01:10:59.9232 | T  | 2            ║
╚═══════════════════════════════════════════════════╝

如果最后5条记录与类别字段匹配，那么我希望在存储过程中比较给定类型的最后5条记录 如果即使一个人没有多少，那么这一切都是失败的。

在类型和时间戳上进行连接是很诱人的，但是表上的时间戳永远不会精确，因为第二个表数据来自外部系统。（在这种情况下不可能使用NTP）

e.g。

在上表中，类型A和B是成功的，因为它们具有完全相同的最后5条记录，这些记录在相同的顺序中属于同一类别，即使它们中的非共有总共5个。
E失败，因为他们都有三个记录，但第二个表的最后一个类别是2而不是1。
C失败，因为表B中不存在
T失败，因为最后5个不匹配。

我希望将两个数据集作为独立数组返回，而不是通过for循环和一个索引来比较各个字段，但是没有看到在postgres中以这种方式使用的数组，请指出我正确的方向。 / p>

所有postgres示例似乎遍历单个表的记录。

example 1

example 2

postgres doc

Answer 1

这是一个SQL查询，它为您提供没有存储过程的结果：

select ta.type, case when ta.str_agg = tb.str_agg then 'success' else 'failed' end
from
(select type, array_to_string(array_agg(category order by timestamp desc), ',') as str_agg
from (select * from table_a as t where t.id in (select id from table_a WHERE type = t.type order by timestamp desc  LIMIT 5)) as t1
group by type) as ta
LEFT JOIN
(select type, array_to_string(array_agg(category order by timestamp desc), ',') as str_agg
from (select * from table_b as t where t.id in (select id from table_b WHERE type = t.type order by timestamp desc  LIMIT 5)) as t1
group by type) as tb
ON (ta.type = tb.type)

test here

对于多个表的循环postgres

1 个答案: