Sql 合并并求和加倍结果

我有两张桌子，t_tran_log和t_item_master

当我对t_tran_log运行此查询时，我得到了准确的结果：

declare @startdate as datetime; declare @enddate as datetime
set @startdate = '9/13/20 5am'; set @enddate = dateadd(hour,24,@startdate)

select
cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date) tran_date
,t.item_number
,sum(t.tran_qty) qty

from t_tran_log t

where t.start_tran_date+t.start_tran_time >= @startdate
and t.start_tran_date+t.start_tran_time < @enddate
and t.tran_type = '750'  and t.description like 'Customer Return' and t.hu_id not like '1Z%'

group by cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date), t.item_number

但是，当我试图通过左联接从另一个表中引入额外的项目详细信息时，我突然得到了项目数量的两倍结果！我做错了什么

declare @startdate as datetime; declare @enddate as datetime
set @startdate = '9/13/20 5am'; set @enddate = dateadd(hour,24,@startdate)

select
cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date) Date
,t.item_number
,i.dept_name
,i.class_name
,i.subclass_name
,sum(t.tran_qty) qty

from t_tran_log t

left join t_item_master i on t.item_number = i.item_number

where t.start_tran_date+t.start_tran_time >= @startdate
and t.start_tran_date+t.start_tran_time < @enddate
and t.tran_type = '750'  and t.description like 'Customer Return' and t.hu_id not like '1Z%'

group by cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date), t.item_number, i.dept_name, i.class_name, i.subclass_name

---

在item_master中可能有多行与联接条件匹配。结果，行被相乘，聚合函数出错

首先考虑在子查询中预聚合：

select t.*, i.dept_name, i.class_name, i.subclass_name
from (
    select
        cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date) tran_date
        ,t.item_number
        ,sum(t.tran_qty) qty
    from t_tran_log t
    where 
        t.start_tran_date + t.start_tran_time >= @startdate
        and t.start_tran_date+t.start_tran_time < @enddate
        and t.tran_type = '750'  
        and t.description like 'Customer Return' 
        and t.hu_id not like '1Z%'
    group by 
        cast(t.start_tran_date+t.start_tran_time  - '1900-01-01 05:00' as date), 
        t.item_number
) t
left join t_item_master i on t.item_number = i.item_number

---

简短回答：您执行多对多连接而不是1对多连接，即“t_item_master”。item_number'不是唯一的抱歉，这不太有效。现在我得到了每个项目编号的两行结果？