Question

我在Heroku（PostgreSQL）中有许多数据字段，可以通过多个关系调用数据范围。对于过去一周的数据，我有类似以下的查询：

SELECT x, y, ts FROM (
  SELECT x1 as x, y1 as y, t1.mytimestamp as ts  
  FROM t1, t2
  WHERE ... t1.mytimestamp::date > (CURRENT_TIMESTAMP - interval '8' day)::date
 UNION
  SELECT x2 as x, y2 as y, t3.mytimestamp as ts  
  FROM t3, t4
  WHERE ... t3.mytimestamp::date > (CURRENT_TIMESTAMP - interval '8' day)::date
 UNION
  ...(etc)...
) ORDER BY ts DESC

有没有办法将像这样的字符串保存为变量，所以要更改所有查询的日期范围（即“'8'日” - ＆gt;“'366'日”）我只需要改变字符串一次而不是每次发生？我发现一些线程说在一个函数中声明变量，但由于某种原因，似乎没有任何工作。

在heroku的文档中它只是说'使用标准SQL' - 函数/变量不是标准的sql特性吗？任何建议都赞赏，谢谢！

Answer 1

我会使用CTE做类似的事情：

WITH vars AS (
  SELECT 
     interval '8' day AS "range"
)

SELECT x, y, ts FROM (
  SELECT x1 as x, y1 as y, t1.mytimestamp as ts  
  FROM t1, t2, vars
  WHERE ... t1.mytimestamp::date > (CURRENT_TIMESTAMP - vars.range)::date
 UNION
  SELECT x2 as x, y2 as y, t3.mytimestamp as ts  
  FROM t3, t4
  WHERE ... t3.mytimestamp::date > (CURRENT_TIMESTAMP - vars.range)::date
 UNION
  ...(etc)...
) ORDER BY ts DESC

Answer 2

将其包装在SQL函数中：

CREATE OR REPLACE FUNCTION my_whatever_func(integer)
returns table(x integer, y integer, ts timestamp)
as $$
SELECT x, y, ts FROM (
  SELECT x1 as x, y1 as y, t1.mytimestamp as ts  
  FROM t1, t2
  WHERE ... t1.mytimestamp::date > (CURRENT_TIMESTAMP - interval $1 day)::date
 UNION
  SELECT x2 as x, y2 as y, t3.mytimestamp as ts  
  FROM t3, t4
  WHERE ... t3.mytimestamp::date > (CURRENT_TIMESTAMP - interval $1 day)::date
) ORDER BY ts DESC
$$ language sql;

并使用：

调用

SELECT * FROM my_whatever_func(8);

这种查询模式表明您可能希望使用表继承。

在Heroku Dataclips上使用PostgreSQL的变量

2 个答案: