Postgresql 循环通过一个查询为另一个查询创建结果

Question

我正在使用 Postgresql 9.6，想知道这是否可行。

我有一个类似的查询：

with mkt_types as (
    select distinct(mkt_medium)
    from sessions
), sessions_all as (
    select *
    from sessions
)
select
        count(visitor_id) as total_visits, 
            count(1) FILTER (where mkt_medium = 'seo') as seo_visitors, 
            count(1) FILTER (where mkt_medium = 'seo' and domain_sessionidx = 1) as seo_new_visitors,
            count(1) FILTER (where mkt_medium = 'seo' and conversion_visit = 'true') as seo_converters,
            count(1) FILTER (where mkt_medium = 'seo' and domain_sessionidx = 1 and conversion_visit = 'true') as seo_new_converters
    from sessions_all;

是否可以从 mkt_types 获取值并遍历结果并替换查询的 'seo' 部分。

示例如果我有 mkt_types 'seo'、'email'、'banner' 那么我的最终查询将以我输入的结果结束：

with mkt_types as (
    select distinct(mkt_medium)
    from sessions
), sessions_all as (
    select *
    from sessions
)
select
        count(visitor_id) as total_visits, 
            count(1) FILTER (where mkt_medium = 'seo') as seo_visitors, 
            count(1) FILTER (where mkt_medium = 'seo' and domain_sessionidx = 1) as seo_new_visitors,
            count(1) FILTER (where mkt_medium = 'seo' and conversion_visit = 'true') as seo_converters,
            count(1) FILTER (where mkt_medium = 'seo' and domain_sessionidx = 1 and conversion_visit = 'true') as seo_new_converters,
            count(1) FILTER (where mkt_medium = 'email') as email_visitors, 
            count(1) FILTER (where mkt_medium = 'email' and domain_sessionidx = 1) as email_new_visitors,
            count(1) FILTER (where mkt_medium = 'email' and conversion_visit = 'true') as email_converters,
            count(1) FILTER (where mkt_medium = 'email' and domain_sessionidx = 1 and conversion_visit = 'true') as email_new_converters,
            count(1) FILTER (where mkt_medium = 'banner') as banner_visitors, 
            count(1) FILTER (where mkt_medium = 'banner' and domain_sessionidx = 1) as banner_new_visitors,
            count(1) FILTER (where mkt_medium = 'banner' and conversion_visit = 'true') as banner_converters,
            count(1) FILTER (where mkt_medium = 'banner' and domain_sessionidx = 1 and conversion_visit = 'true') as banner_new_converters
    from sessions_all;

谢谢

Answer 1

在我看来，我们似乎不能那样做。当然，您可以放入函数并构建查询。

如果您可以垂直而不是水平获得结果，我可以建议另一个这样的查询：

(
 select 'all' as mkt_medium,
    count(1) as total_visits, 
    count(1) FILTER (where domain_sessionidx = 1) as new_visitors, 
    count(1) FILTER (where conversion_visit = 'true') as converters, 
    count(1) FILTER (where domain_sessionidx = 1 and conversion_visit = 'true') as new_converters
   FROM sessions_all
) UNION ALL (
 SELECT mkt_medium,
            count(1) as total_visits, 
    count(1) FILTER (where domain_sessionidx = 1) as new_visitors, 
    count(1) FILTER (where conversion_visit = 'true') as converters, 
    count(1) FILTER (where domain_sessionidx = 1 and conversion_visit = 'true') as new_converters
    from sessions_all
   GROUP BY mkt_medium
)

Postgresql 循环通过一个查询为另一个查询创建结果

Postgresql loop through onequery to create results for another query

sql

postgresql

pivot