如何正确指定条件：sum count = column value？

Question

我正在更新系统，尝试创建原始表的“影子”表，通过触发器接收更改，但不使用原始表，因为我真的不喜欢原始程序的设计方式。

其中一个是团队配置表，其中包含所有用户及其团队 ID。具有相同团队 id 的用户属于同一团队。我正在尝试在此处创建触发器以将数据截获到新的团队历史日志中。 （1 行 / 1 个团队 ID） 但是由于在团队设置之后，表格的多行会被更新，并且更新不是批量完成的，因此很难判断新的团队设置何时结束。幸运的是有这个 member_count 列，可以用来统计成员数量，并确定与这个新的 team_id 相关的所有必要记录何时更新。

| team id     | uid | member_count |
|201701010800A| 1   | 3            |
|201701010800A| 2   | 3            |
|201701010800A| 3   | 3            |

我的问题是我应该如何在更新后触发器中指定这个条件。 这就是我现在要做的。但这似乎很hacky。我想还有更明智的表达方式。

if exists(
    select 1 from ( values (1) ) t(c) 
    outer apply (
        select count(team_id) tc, max(member_count) mx, min(member_count) mm
        from old_teamtable 
        where team_id = ( select top 1 team_id from inserted )
    ) e
    where tc = mx and mx = mm
)

Answer 1

这对你有用吗？

if exists (
    select  1 
    from    old_teamtable 
    where   team_id = (select top 1 team_id from inserted) 
    having  count(*) = max(member_count) 
    and     max(member_count) = min(member_count)
)

Answer 2

以下代码演示了一种基于对团队和用户表的插入来管理团队历史记录表的方法。触发器在每个insert 到TeamsAndUsers 之后触发。对于尚未在TeamHistory 中的新团队，将添加适当的行并标记为IsComplete = 0。然后通过检查受insert 影响的所有团队来更新TeamHistory，以查看他们是否拥有正确数量的用户。如果是这样，IsComplete 位将设置为 1，用于 TeamHistory 表中的相应行。

-- Create sample tables.
create table TeamsAndUsers ( Id Int Identity, TeamId Int, UserId Int, TeamMembers Int );

create table TeamHistory ( Id Int Identity, TeamId Int, IsComplete Bit );
go

-- Create the trigger.
create trigger UpdateTeamHistory
  on TeamsAndUsers
  after Insert
as
  set nocount on

  -- Handle any teams we've not heard of before.
  insert into TeamHistory ( TeamId, IsComplete )
    select distinct TeamId, 0
      from inserted
      where TeamId not in ( select TeamId from TeamHistory );

  -- See if any teams have been completed, i.e. all of their members are present.
  update TeamHistory
    set IsComplete = 1
    where TeamId in (
      select TAU.TeamId from (
        -- TeamId   values touched by the INSERT ...
        select distinct TeamId from inserted ) as i inner join
          -- ... matched with corresponding incomplete teams in the team history ...
          TeamHistory as TH on TH.TeamId = i.TeamId and TH.IsComplete = 0 inner join
          -- ... matched with users currently assigned to the teams.
          TeamsAndUsers as TAU on TAU.TeamId = i.TeamId
    group by TAU.TeamId
    having Count( TAU.UserId ) = Max( TAU.TeamMembers ) );
go

-- Test the trigger.
insert into TeamsAndUsers ( TeamId, UserId, TeamMembers ) values
  ( 101, 1, 3 ), ( 101, 2, 3 );
select * from TeamsAndUsers;
select * from TeamHistory;

insert into TeamsAndUsers ( TeamId, UserId, TeamMembers ) values
  ( 101, 3, 3 );
select * from TeamsAndUsers;
select * from TeamHistory;

insert into TeamsAndUsers ( TeamId, UserId, TeamMembers ) values
  ( 102, 4, 1 );
select * from TeamsAndUsers;
select * from TeamHistory;

insert into TeamsAndUsers ( TeamId, UserId, TeamMembers ) values
  ( 103, 10, 2 );
select * from TeamsAndUsers;
select * from TeamHistory;

insert into TeamsAndUsers ( TeamId, UserId, TeamMembers ) values
  ( 103, 11, 2 );
select * from TeamsAndUsers;
select * from TeamHistory;

除此之外：update 中的 select 看起来相当笨拙，但这里已经很晚了。大概可以整理一下。