如何修复具有多个连接的计数的物化视图？

Question

我正在 PostgreSQL 中使用物化视图优化查询，但查询逻辑在 mat 视图中不起作用

我想优化涉及多个连接的查询，它的执行时间也更长，所以我使用 PostgreSQL 在物化视图中尝试了相同的查询，但是当涉及到 mat 视图时，查询逻辑出错了。

我已经尝试在 PostgreSQL 11 中创建这个 mat 视图。

在下面的代码中，三个键 tables

1.posts 
2.Topics 
3.Post_topics.

Post_topics tables 持有 post_id 和 topic_id.

Topics tables 是主题列表（每个主题都有多个值；

example if topic is egg, values associated with egg were 'breakfast','dinner','cheese' etc)

和 posts tables 持有与主题相关的帖子。每个主题可能有多个帖子。

我想从 topic table 中获取保存值的 count。那么当主题 id 是鸡蛋时，早餐、晚餐、奶酪的数量是多少。主题 table 持有 8000 个，其中早餐、晚餐、奶酪也在值列表中。如果我输入奶酪，鸡蛋、早餐、奶酪的计数应该来了。在 Normal 查询中，我已经这样做了，但在 mat view 中，我正在努力获得这个逻辑。

原始查询：

SELECT t.id as topic_id, t.value as value, t.topic as topic, COUNT(c.topic_id) as count 
FROM post_topics a 
JOIN posts b ON a.post_id = b.id 
JOIN post_topics c on a.post_id = c.post_id 
--JOIN post_locations pl ON pl.post_id = c.post_id 
JOIN topics t on c.topic_id = t.id --AND t.topic = 'cuisine' 
WHERE a.topic_id = '1234547hnunsdrinfs' 
AND t.id != '1234547hnunsdrinfs' 
AND b.date_posted BETWEEN ('2019-06-17'::date - interval '6 month') AND '2019-06-17'::date 
     GROUP BY t.id, c.topic_id 
     ORDER BY count DESC 
     LIMIT 20

我已经使用主题 ID 列表编辑了垫子视图查询（主题 table 有 8000 个值）。在最初的查询中，我只提到了 1 个主题 ID，但我需要整个 8000 个的结果。对于下面的垫子视图，我得到了每个 ID 的热门主题，但计数不匹配。

**Edited Mat View
      Create materialized view top_associations_mv AS
        SELECT a.topic_id, t.id as topic_id, t.value as value, t.topic as topic, COUNT(c.topic_id) as count 
        FROM post_topics a 
        JOIN posts b ON a.post_id = b.id 
        JOIN post_topics c on a.post_id = c.post_id 
        JOIN topics t on c.topic_id = t.id 
        WHERE a.topic_id in ('c108200f-e4dc-415e-9150-3f6c74b879e2', '107f8cad-75b3-43fb-9b2f-f7914bf45155') -- here all 8000 topic id should be placed
        --AND t.id <>  ( 'c108200f-e4dc-415e-9150-3f6c74b879e2', '107f8cad-75b3-43fb-9b2f-f7914bf45155')--,'c348af9d-dd98-49f1-b6c2-8ea36b404ffa')
         and (b.date_posted > (('now'::text)::date - '6 mons'::interval))
        GROUP BY t.id, c.topic_id,a.topic_id 
        ORDER BY count DESC limit 10 ;

物化视图：

Create MATERIALIZED VIEW top_associations_mv as 
SELECT t.id as topic_id, t.value as value, t.topic as topic, COUNT(c.topic_id) as count 
FROM post_topics a 
JOIN posts b ON a.post_id = b.id 
JOIN post_topics c on a.post_id = c.post_id
JOIN topics t on c.topic_id = t.id 
WHERE t.id != c.post_id and (b.date_posted > (('now'::text)::date - '6mons'::interval)) 
GROUP BY t.id, c.topic_id 
ORDER BY count DESC

我的预期结果是：

我想从包含值的主题 table 中获取计数。那么当主题 id 是鸡蛋时，早餐、晚餐、奶酪的数量是多少。但实际结果计数是错误的。

我真的需要帮助！！！

Answer 1

您应该在连接上使用子查询，并且子查询还应该包含 distinct on 子句。你可以使用它：

create materialized view top_associations_mv as select
    t.id as topic_id,
    t.value as value,
    t.topic as topic,
    count(c.topic_id) as count
from
    post_topics a
join (
    select
        distinct on
        (id) id,
        date_posted
    from
        posts ) as b on
    a.post_id = b.id
and
    (b.date_posted > (('now'::text)::date - '6mons'::interval)) (
    select
        distinct on
        (post_id) topic_id
    from
        post_topics ) as c on
    a.post_id = c.post_id (
    select
        distinct on
        (id) id
    from
        topics ) as t on
    c.topic_id = t.id
    and t.id != c.post_id
group by
    t.id,
    c.topic_id
order by
    count desc

Answer 2

我修正了查询。它只是另一个主题 table.

SELECT a.topic_id as main_topic, t.id as topicid, t.value as value, t.topic as topic, COUNT(c.topic_id) as count
FROM post_topics a
JOIN posts b ON a.post_id = b.id
JOIN post_topics c on a.post_id = c.post_id
JOIN topics t on c.topic_id = t.id
**JOIN topics t2 on t2.id = a.topic_id and (t.id) <> (t2.id)**
WHERE (b.date_posted > (('now'::text)::date - '6 mons'::interval))
AND LOWER(b.source) = 'instagram'
GROUP BY t.id, c.topic_id,a.topic_id
ORDER BY c.topic_id,count DESC;

如何修复具有多个连接的计数的物化视图？

How to fix the materialized view on counts with multiple joins?

postgresql

materialized-views