如何避免 xml 路径的多次使用？

Question

我在使用 sql 查询时遇到问题。 我要做的是获取商店列表，其中包含分配给用户的任务。 用用户名分隔列中的每个角色。 我正在使用 STUFF 来获取连接行，但对于表中的 100K+ 行来说，这似乎是个坏主意。

这是简化的结构：

用户表

CREATE TABLE #temp_Users(
    [id] [int] IDENTITY(1,1) NOT NULL,  
    [user_name] [nvarchar](250) NULL,
 CONSTRAINT [PK_Users] PRIMARY KEY CLUSTERED ([id] ASC)
 )

 insert into #temp_Users (user_name) values ('Joe'),('Jeff'),('Jimm')

任务表

create table #temp_Tasks (
    [id] [int] IDENTITY(1,1) NOT NULL,
    [id_user] [int] NULL,
    [id_store] [int] NULL,
    [id_role] [int] NULL,
    CONSTRAINT [PK_Tasks] PRIMARY KEY CLUSTERED  ([id] ASC)
) 

insert into #temp_Tasks ([id_user],[id_store],[id_role])
       values (1,1,0),(1,2,0),(2,1,0),(2,2,0),(1,1,1),(2,2,1),(3,1,0),(3,2,0),(3,2,1)

Select

SELECT distinct t.id_store,
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 0 FOR XML PATH('')),1,2,'' ) as 'role_0',
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 1 FOR XML PATH('')),1,2,'' )  as 'role_1' 
FROM #temp_Tasks t

结果

问题是 select 越来越慢，因为每个 STUFF 都是另一个嵌套循环，所以当表格填充得越来越多时。 如果我需要在这个 select 中添加一些“角色”，我必须添加另一个东西，如下所示：

SELECT distinct t.id_store,
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 0 FOR XML PATH('')),1,2,'' ) as 'role_0',
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 1 FOR XML PATH('')),1,2,'' )  as 'role_1' 
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 2 FOR XML PATH('')),1,2,'' )  as 'role_2' 
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 3 FOR XML PATH('')),1,2,'' )  as 'role_3' 
  stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 4 FOR XML PATH('')),1,2,'' )  as 'role_4' 
FROM #temp_Tasks t

问题是我可以避免使用这个多个 STUFF 函数吗？ 或者我可能需要在表任务上创建非聚集索引？ 也许每个角色都有过滤索引？ 提前致谢！

编辑：我正在使用 MSSQL 2016。

正如 Gordon Linoff 建议的那样，我已经将 Select 从 Distinct 更改为 Group By ，如下所示：

SELECT t.id_store,
      stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 0 FOR XML PATH('')),1,2,'' ) as 'role_0',
      stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 left outer join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 1 FOR XML PATH('')),1,2,'' )  as 'role_1' 
    FROM #temp_Tasks t
group by t.id_store

执行时间从 20 秒减少到 2 秒。 所以问题不在于 Stuff，而在于 Distinct。

Answer 1

自 SQL Server 2017 以来，您使用string_agg()吗？

SELECT t.id_store,
       STRING_AGG(CASE WHEN t.id_role = 0 THEN u.user_name END, ',') WITHIN GROUP (ORDER BY u.user_name) as role_0,
       STRING_AGG(CASE WHEN t.id_role = 1 THEN u.user_name END, ',') WITHIN GROUP (ORDER BY u.user_name) as role_1
FROM #temp_Tasks t JOIN
     #temp_Users u
     ON u.id = t2.id_users
GROUP BY t.id_store;

正如Aaron Bertrand 解释的那样，在旧版本中，您可能会受益于使用GROUP BY而不是SELECT DISTINCT 。 我还将用INNER JOIN替换子查询中的LEFT JOIN JOIN ：

SELECT t.id_store,
       stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 0 FOR XML PATH('')),1,2,'' ) as 'role_0',
       stuff( (SELECT ', ' + u.[user_name] FROM #temp_Tasks t2 join #temp_Users u on u.id = t2.id_user WHERE t2.id_store = t.id_store and t2.id_role = 1 FOR XML PATH('')),1,2,'' )  as 'role_1' 
FROM #temp_Tasks t
GROUP BY t.id_store;