合并重复的相同子查询
Posted
技术标签:
【中文标题】合并重复的相同子查询【英文标题】：Consolidate repeated identical subquery 【发布时间】：2018-12-28 03:54:45 【问题描述】：如何在下面的查询中删除或合并三个相同的子查询？
用例详细信息：我使用的 Projects 表列出了数百万个项目。每个项目记录都有一个创建者、管理员和编辑者，由系统 ID 指示。我想用 Employees 表中保存的相应名称替换这些系统 ID。要交叉引用系统 ID 和名称，需要名为 Users 的第三个表及其 hr_id 字段。
Projects:
--prj-- --name-- -owner- -creator- -editor- --many more columns...
001 alpha 001Z 300Z 304Z ...
002 beta 020Z 350Z 600Z ...
003 charlie 600Z 020Z 001Z ...
Employees: Users:
--hr_id-- --name-- -hr_id- -sys_id-
A01 john A01 001Z
A02 susan A02 020Z
A03 ryan A03 300Z
A04 kelly A04 304Z
A05 matt A05 350Z
A06 bert A06 600Z
Desired output:
--prj-- --name-- -owner- -creator- -editor- --adt'l cols...
001 alpha john ryan kelly ...
002 beta susan matt bert ...
003 charlie bert susan john ...
以下是我的代码，包括一些必须保留的不相关连接。我的查询按预期运行，但效率不高，我将不胜感激。另外，（根据我的谷歌搜索，我认为这是相关的）我在一个不适合 CTE 的环境中工作。
SELECT projects.prj As project_id,
projects.name As project_name,
owner.name As owner_name,
creator.name As creator_name,
editor.name As editor_name,
stats.stat1 As stat_1,
actuals.stat2 As stat_2
FROM "dbconnect"."projects" As projects
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS owner ON projects.owner = owner.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS creator ON projects.creator = creator.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS editor ON projects.editor = editor.sys_id
LEFT JOIN "dbconnect"."prjstats" As stats ON projects.prj = prjstats.prj_id
LEFT JOIN "dbconnect"."prjactuals" As actuals ON projects.prj = prjactuals.prj_id
【问题讨论】：
您是否尝试过使用STABLE 函数（输入 = 用户的 sys_id，输出 = 员工姓名）？见gpdb.docs.pivotal.io/5150/ref_guide/sql_commands/… + postgresql.org/docs/current/sql-createfunction.html + postgresql.org/docs/current/xfunc-volatility.html
【参考方案1】：
您可以为子查询创建标量值函数，然后像这样重写查询。
CREATE FUNCTION dbo.getName (@id varchar(30))
RETURNS varchar(128)
AS
BEGIN
DECLARE @v_name varchar(128)
SELECT @v_name=emps.name
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users ON emps.hr_id = users.hr_id
WHERE users.sys_id=@id
RETURN @v_name
END
--Query
SELECT projects.prj As project_id,
projects.name As project_name,
/*
owner.name As owner_name,
creator.name As creator_name,
editor.name As editor_name,
*/
dbo.getName(projects.owner) as owner_name,
dbo.getName(projects.creator) as creator_name,
dbo.getName(projects.editor) as editor_name,
stats.stat1 As stat_1,
actuals.stat2 As stat_2
FROM "dbconnect"."projects" As projects
/*
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS owner ON projects.owner = owner.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS creator ON projects.creator = creator.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS editor ON projects.editor = editor.sys_id
*/
LEFT JOIN "dbconnect"."prjstats" As stats ON projects.prj = prjstats.prj_id
LEFT JOIN "dbconnect"."prjactuals" As actuals ON projects.prj = prjactuals.prj_id
或者您可以创建表值函数，然后使用 APPLY 运算符加入该函数。
CREATE FUNCTION dbo.getName (@id varchar(30))
RETURNS TABLE
AS
RETURN
(
SELECT emps.name
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users ON emps.hr_id = users.hr_id
WHERE users.sys_id=@id
)
--Query
SELECT projects.prj As project_id,
projects.name As project_name,
owner.name As owner_name,
creator.name As creator_name,
editor.name As editor_name,
stats.stat1 As stat_1,
actuals.stat2 As stat_2
FROM "dbconnect"."projects" As projects
OUTER APPLY dbo.getName(projects.owner) as owner
OUTER APPLY dbo.getName(projects.creator) as creator
OUTER APPLY dbo.getName(projects.editor) as editor
/*
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS owner ON projects.owner = owner.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS creator ON projects.creator = creator.sys_id
LEFT JOIN (
SELECT emps.name,
users.hr_id,
users.sys_id
FROM "dbconnect"."employees" AS emps
RIGHT JOIN "dbconnect"."users" AS users
ON emps.hr_id = users.hr_id
) AS editor ON projects.editor = editor.sys_id
*/
LEFT JOIN "dbconnect"."prjstats" As stats ON projects.prj = prjstats.prj_id
LEFT JOIN "dbconnect"."prjactuals" As actuals ON projects.prj = prjactuals.prj_id
【讨论】：
【参考方案2】：只需使用 CTE。我更喜欢left joins，所以我会这样写：
WITH eu as (
SELECT e.name, u.hr_id, u.sys_id
FROM "dbconnect"."users" u LEFT JOIN
"dbconnect"."employees" e
ON e.hr_id = u.hr_id
)
SELECT p.prj As project_id, p.name As project_name,
euo.name As owner_name, euc.name As creator_name,
eue.name As editor_name,
ps.stat1 As stat_1,
pa.stat2 As stat_2
FROM "dbconnect"."projects" p LEFT JOIN
eu euo
ON p.owner = euo.sys_id LEFT JOIN
eu euc
ON p.creator = euc.sys_id LEFT JOIN
eu eue
ON p.editor = eue.sys_id LEFT JOIN
"dbconnect"."prjstats" ps
ON p.prj = ps.prj_id LEFT JOIN
"dbconnect"."prjactuals" pa
ON p.prj = pa.prj_id;
【讨论】：
以上是关于合并重复的相同子查询的主要内容，如果未能解决你的问题，请参考以下文章