函数循环并从多个表中选择数据 [英] Function to loop through and select data from multiple tables

问题描述

我是Postgres的新手，并且拥有一个具有相同结构的多个表的数据库。我需要从每个表中选择符合特定条件的数据。

我可以用一堆 UNION 查询，但是我需要搜索的表的数量会随着时间的推移而改变，所以我不想像这样对其进行硬编码。我一直在试图开发一个函数，将循环通过特定的表（他们有一个共同的命名约定），并返回一个记录表，但我没有得到任何结果，当我查询函数。函数代码如下：

pre $ CREATE OR REPLACE FUNCTION public.internalid_formaltable_name_lookup（）
RETURNS TABLE（natural_id text，name text ，natural_id_numeric文本）AS
$ BODY $
DECLARE
formal_table text;
begin
for formal_table IN
从information_schema.tables中选择table_name
其中table_schema ='public'和table_name like'formaltable％'
LOOP
EXECUTE' SELECT natural_id，name，natural_id_numeric
FROM'|| formal_table ||
'WHERE natural_id_numeric IN（
从internal_idlookup中选择natural_id_numeric
where internal_id =''7166571''）';
RETURN NEXT;
END LOOP;
返回;
END;
$ BODY $
语言plpgsql;

当我尝试使用该函数时，没有收到任何错误，但没有返回任何行：

  SELECT * from internalid_formaltable_name_lookup（）; 
  

有什么想法我错了？

RETURNS TABLE（natural_id text，name text，natural_id_numeric text）AS
$ func $
DECLARE
formal_table text;
BEGIN
for formal_table IN
SELECT quote_ident（table_name）
FROM information_schema.tables
WHERE table_schema ='public'
AND table_name LIKE'formaltable％'
LOOP
RETURN QUERY EXECUTE
'SELECT t.natural_id，t.name，t.natural_id_numeric
FROM internal_idlookup i
JOIN public。 formal_table || 't USING（natural_id_numeric）
WHERE i.internal_id = 7166571'; - 假设internal_id是数字
END LOOP;
END
$ func $ LANGUAGE plpgsql;

要点：

• 您必须使用 返回QUERY EXECUTE 返回每个行集。
  EXECUTE ，然后是 RETURN NEXT ，完全不符合你期望的 。




• 您需要清理标识符。我在这里使用 quote_ident（）。或者你的查询将打破非标准的标识符，并允许SQL注入！
  转换你的 col IN（子选择）

code>转换为更高效的 JOIN 。


• 这与使用一堆UNION查询。它不会删除重复的行，并且实际上像 UNION ALL 。




就个人而言，我宁愿在系统目录 pg_class 上构建它。详细信息：




• 
  




然后，您可以使用 pg_class.oid :: regclass 自动转义和模式化表名。详细信息：





• 表名作为PostgreSQL函数参数


• 
  




但这取决于您的要求和...味道的细节。


I'm new to Postgres and have a database with multiple tables of the same structure. I need to select data from each table that matches certain criteria.

I could do this with a bunch of UNION queries, but the number of tables I need to search can change over time, so I don't want to hard code it like that. I've been trying to develop a function that will loop through specific tables (they have a common naming convention) and return a table of records, but I'm not getting any results when I query the function. Function code is below:

CREATE OR REPLACE FUNCTION public.internalid_formaltable_name_lookup()
  RETURNS TABLE(natural_id text, name text, natural_id_numeric text) AS
$BODY$
DECLARE
    formal_table text;
begin
  FOR formal_table IN
    select table_name from information_schema.tables
    where table_schema = 'public' and table_name like 'formaltable%'
  LOOP
    EXECUTE 'SELECT natural_id, name, natural_id_numeric
             FROM ' || formal_table || 
           ' WHERE natural_id_numeric IN (
                select natural_id_numeric from internal_idlookup
                where internal_id = ''7166571'')';
    RETURN NEXT;
 END LOOP;
 Return;
END;
$BODY$
  LANGUAGE plpgsql;

I am not getting any errors when I try to use the function, but it's not returning any rows:

SELECT * From internalid_formaltable_name_lookup();

Any idea where I went wrong?

解决方案

CREATE OR REPLACE FUNCTION public.internalid_formaltable_name_lookup()
  RETURNS TABLE(natural_id text, name text, natural_id_numeric text) AS
$func$
DECLARE
   formal_table text;
BEGIN
   FOR formal_table IN
      SELECT quote_ident(table_name)
      FROM   information_schema.tables
      WHERE  table_schema = 'public'
      AND    table_name LIKE 'formaltable%'
   LOOP
      RETURN QUERY EXECUTE
      'SELECT t.natural_id, t.name, t.natural_id_numeric
       FROM   internal_idlookup i 
       JOIN   public.' || formal_table || ' t USING (natural_id_numeric)
       WHERE  i.internal_id = 7166571';   -- assuming internal_id is numeric
   END LOOP;
END
$func$  LANGUAGE plpgsql;

Major points:

• You have to use RETURN QUERY EXECUTE to return each set of rows.
  EXECUTE, followed by RETURN NEXT, does not do what you seem to expect at all.

• You need to sanitize identifiers. I am using quote_ident() here. Or your query will break with non-standard identifiers and allow SQL injection!

• Converted your col IN (sub-select) to a more efficient JOIN.

• This is subtly different from using a bunch of UNION queries. It does not remove duplicate rows, and actually works like UNION ALL.

Personally, I'd rather build this on the system catalog pg_class. Details:

• How to check if a table exists in a given schema

Then you can work with the pg_class.oid::regclass to escape and schema-qualify table names automatically. Details:

• Table name as a PostgreSQL function parameter
• Search across multiple tables and also display table name in resulting rows

But that depends on the details of your requirements and ... taste.

这篇关于函数循环并从多个表中选择数据的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！