sql - postgreSQL 中的循环表

Question

我需要创建一个物化 View 来循环 pg_tables 中的所有工作列表。

首先我创建一个查询，然后尝试创建一个 for 循环。

这是我的第一个代码没有 for 循环。

SELECT 
count(visit_num) AS total_claim_count,
count(user_id) as assigned_count,
(count(visit_num)-count(user_id)) AS not_assigned_count,
count(case when doe <= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end ) AS Assigned_Last_30_Days,
(count(user_id)-count(case when doe <= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end )) AS not_Assigned_last_30_days,
count(case when doe >= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end ) AS Assigned_before_30_Days,
count(case when doe > (SELECT doe - INTERVAL '30 days') and user_id is null then 1 end ) AS not_Assigned_before_30_days,
count(CASE WHEN date_worked is not null then 1 end) followed_up,
count(CASE WHEN last_denied_date is not null then 1 end) denials,
count(case when status_and_action_code='No Status - Need to follow-up' then 1 end) AS no_status
FROM Worklist_123456

这是我尝试过的代码带有for循环。但它不起作用。

DO $$

DECLARE
    pg_tables varchar(1000000);
    tbl_name record;
    tablename varchar(1000000);
    
begin
For tbl_name IN SELECT tablename FROM pg_tables WHERE schemaname = 'public' AND tablename LIKE 'worklist¥_%' ESCAPE '¥'

loop

    tablename := tbl_name.tablename;

    SELECT 
    count(visit_num) AS total_claim_count,
    count(user_id) as assigned_count,
    (count(visit_num)-count(user_id)) AS not_assigned_count,
    count(case when doe <= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end ) AS Assigned_Last_30_Days,
    (count(user_id)-count(case when doe <= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end )) AS not_Assigned_last_30_days,
    count(case when doe >= (SELECT doe - INTERVAL '30 days') and user_id is not null then 1 end ) AS Assigned_before_30_Days,
    count(case when doe > (SELECT doe - INTERVAL '30 days') and user_id is null then 1 end ) AS not_Assigned_before_30_days,
    count(CASE WHEN date_worked is not null then 1 end) followed_up,
    count(CASE WHEN last_denied_date is not null then 1 end) denials,
    count(case when status_and_action_code='No Status - Need to follow-up' then 1 end) AS no_status
    FROM tablename;

END LOOP;

END; $$;

它给出这样的错误。

ERROR:  column reference "tablename" is ambiguous
LINE 1: SELECT tablename FROM pg_tables WHERE schemaname = 'public' ...
               ^
DETAIL:  It could refer to either a PL/pgSQL variable or a table column.
QUERY:  SELECT tablename FROM pg_tables WHERE schemaname = 'public' AND tablename LIKE 'worklist¥_%' ESCAPE '¥'
CONTEXT:  PL/pgSQL function inline_code_block line 9 at FOR over SELECT rows
SQL state: 42702

任何人都可以帮我解决这个问题。谢谢。

Answer 1

如pointed out通过 @Frank Heikens ，您正在考虑 dynamic SQL条款: demo

DO $proc$
DECLARE tbl_name record;
BEGIN
FOR tbl_name IN SELECT tablename 
                FROM  pg_tables 
                WHERE schemaname = 'public' 
                AND   tablename LIKE 'worklist¥_%' ESCAPE '¥'
LOOP execute format($dynamic_sql_query$
        INSERT INTO anon_proc_output
        SELECT 
            %1$L             AS source_table --first argument (1$), as a string Literal (L)
          , count(visit_num) AS total_claim_count
          , count(user_id)   AS assigned_count
          ,(count(visit_num) 
           -count(user_id) ) AS not_assigned_count
          , count(user_id)filter(where doe <= (SELECT doe - INTERVAL '30 days') )   AS Assigned_Last_30_Days
          ,(count(user_id) 
           -count(user_id)filter(where doe <= (SELECT doe - INTERVAL '30 days') ) ) AS not_Assigned_last_30_days
          , count(user_id)filter(where doe >= (SELECT doe - INTERVAL '30 days') )   AS Assigned_before_30_Days
          , count(*)filter(where doe > (SELECT doe - INTERVAL '30 days') 
                           and user_id is null)                                     AS not_Assigned_before_30_days
          , count(date_worked)      AS followed_up
          , count(last_denied_date) AS denials
          , count(*)filter(where status_and_action_code='No Status - Need to follow-up') AS no_status
        FROM %1$s; --still the same first argument, hence the 1$, but this time as-is, hence the s
        $dynamic_sql_query$
      , tbl_name.tablename);--this is used to replace %1$L %1$s
END LOOP;
END $proc$;

1. count(case...) 正在模拟聚合 FILTER条款。
2. doe>=(SELECT doe-INTERVAL '30 days') 毫无意义。您正在检查该值是否大于其自身，是否减少。这实际上是 some_number>=(some_number-1)。
3. count(arg) doesn't count rows where arg is null :

count ( "any" ) → bigint Computes the number of input rows in which the input value is not null.

• so count(case when arg is not null then 1 end) 与 count(arg) 相同，

• count(case when (condition) and arg is not null then 1 end) 与 count(arg)filter(where (condition)) 相同.

  您可能需要重新思考在何处以及为何使用 count(arg)，而不是 count(*) 和 count(case when arg is不为空则 1 结束)。