SQL Server 在日期范围内聚合答案

【问题标题】：SQL Server aggregate over range of datesSQL Server 在日期范围内聚合
【发布时间】：2016-07-29 11:43:29
【问题描述】：

我使用的是 SQL Server 2014。我需要汇总按客户和位置划分或分组的日期范围内的总计（总和）。关键是获取所有调整金额并在它们适用于开票交易日期时对其进行汇总。

因此，在最后一个账单日期之后但小于下一个账单日期的所有调整都需要汇总并与账单金额一起很好地呈现。

查看示例：

+------------------+------------+------------+------------------+--------------------+
| TRANSACTION_TYPE | CUSTOMERID | LOCATIONID | TRANSACTION DATE | TRANSACTION AMOUNT |
+------------------+------------+------------+------------------+--------------------+
| bill             | 215        | 102        | 7/7/2016         | $100.00            |
| bill             | 215        | 102        | 6/6/2016         | $121.00            |
| adj              | 215        | 102        | 6/1/2016         | $22.00             |
| adj              | 215        | 102        | 5/8/2016         | $0.35              |
| adj              | 215        | 102        | 5/7/2016         | $5.00              |
| bill             | 215        | 102        | 5/6/2016         | $115.00            |
| bill             | 215        | 102        | 4/7/2016         | $200.00            |
| adj              | 215        | 102        | 4/2/2016         | $4.35              |
| adj              | 215        | 102        | 4/1/2016         | $(0.50)            |
| adj              | 215        | 102        | 3/28/2016        | $33.00             |
| bill             | 215        | 102        | 3/28/2016        | $75.00             |
| adj              | 215        | 102        | 3/5/2016         | $0.33              |
| bill             | 215        | 102        | 3/3/2016         | $99.00             |
+------------------+------------+------------+------------------+--------------------+

我想看到的是以下内容：

    +------------------+------------+------------+------------------+-------------+-------------------+
| TRANSACTION_TYPE | CUSTOMERID | LOCATIONID | TRANSACTION DATE | BILL AMOUNT | ADJUSTMENT AMOUNT |
+------------------+------------+------------+------------------+-------------+-------------------+
| bill             | 215        | 102        | 7/7/2016         | $100.00     | $-                |
| bill             | 215        | 102        | 6/6/2016         | $121.00     | $27.35            |
| bill             | 215        | 102        | 5/6/2016         | $115.00     | $-                |
| bill             | 215        | 102        | 4/7/2016         | $200.00     | $36.85            |
| bill             | 215        | 102        | 3/28/2016        | $75.00      | $0.33             |
| bill             | 215        | 102        | 3/3/2016         | $99.00      | $-                |
+------------------+------------+------------+------------------+-------------+-------------------+

【问题讨论】：

标签： sql sql-server-2014 aggregates

【解决方案1】：

你需要：

首先将表设想为 TransactionType 上的两个（虚拟）子表；
然后使用 LEAD 函数获取要应用的调整的日期范围；和
最终执行 eft 连接。

以下未经测试的 SQL：

with
BillData as (
    select
        TransactionType,
        CustomerID,
        LocationID,
        TransactionDate,
        TransactionAmount,
        lead(TransactionDate, 1) over (partition by CustomerID 
                                       order by TransactionDate) as NextDate
    from @data bill
    where TransactionType = 'bill'
),
AdjData as (
    select
        CustomerID,
        TransactionDate,
        sum(TransactionAmount) as AdjAmount
    from @data adj
    where TransactionType = 'adj'
)
select
    bill.TransactionType,
    bill.CustomerID,
    bill.LocationID,
    bill.TransactionDate,
    sum(TransactionAmount)  as BillAmount,
    sum(AdjAmount)          as AdjAmount
from BillData bill
left join AdjData adj
    on adj.CustomerID = bill.CustomerID
   and bill.TransactionDate <= adj.TransactionDate
   and adj.TransactionDate < bill.NextDate
group by
    bill.TransactionType,
    bill.CustomerID,
    bill.LocationID,
    bill.TransactionDate
;

【讨论】：

感谢您的帮助。该查询并不完全有效，但它确实为我指明了正确的方向，所以我会接受这个作为正确答案。再次感谢！

【解决方案2】：

这就是我最终做的：

        select
        bill.TransactionType,
        bill.CustomerID,
        bill.LocationID,
        bill.TransactionDate,
        TransactionAmount  as BillAmount,
        sum(AdjAmount)          as AdjAmount
    from 
    (
        select
            TransactionType,
            CustomerID,
            LocationID,
            TransactionDate,
            TransactionAmount,
            lag(TransactionDate, 1) over (partition by CustomerID, LocationID
                                           order by TransactionDate) as PreviousDate --NextDate
        from test1
        where TransactionType = 'bill'
    ) as bill
    left join
    (
        select
            CustomerID,
            LocationID,
            TransactionDate,
            TransactionAmount as AdjAmount
        from test1
        where TransactionType = 'adj'
    ) as adj
    ON 
        adj.CustomerID = bill.CustomerID
        and adj.LocationID = bill.LocationID
    and adj.TransactionDate >= bill.PreviousDate
    and adj.TransactionDate < bill.TransactionDate
    group by
        bill.TransactionType,
        bill.CustomerID,
        bill.LocationID,
        bill.TransactionDate,
        bill.TransactionAmount
    order by 4 desc

【讨论】：