您可以在期初和期末查看情况，如果没有发现期初情况，则假设没有库存。如果没有发现最终情况，这意味着在此期间没有销售。
例如，开始的时段 2018-05-13 和结束 2018-05-14 :

select  parts.PartNo
,       coalesce(FirstSale.Total, 0) - coalesce(LastSale.Total, FirstSale.Total, 0) as Sold
,       (coalesce(FirstSale.Total, 0) - coalesce(LastSale.Total, FirstSale.Total, 0)) *
           coalesce(LastSale.Retail - LastSale.Wholesale, 0) as Profit
from    (
        select  PartNo
        ,       max(case when Date < '2018-05-13' then Date end) as FirstEntry
        ,       max(case when Date <= '2018-05-14' then Date end) as LastEntry
        from    Sales
        group by
                PartNo
        ) parts
left join    
        Sales FirstSale
on      FirstSale.PartNo = parts.PartNo
        and FirstSale.Date = parts.FirstEntry
left join    
        Sales LastSale
on      LastSale.PartNo = parts.PartNo
        and LastSale.Date = parts.LastEntry

例如sql fiddle。

SELECT c.partno as partno,MAX(c.inv)-MIN(c.inv) as sold,SUM(CASE WHEN c.date = c.last_date THEN profit else 0 END)*(MAX(c.inv)-MIN(c.inv)) as profit
FROM (SELECT partno,date,inv,retail-wholesale as profit,MAX(date) OVER (partition by partno) AS last_date FROM test1)c
GROUP BY c.partno
ORDER BY c.partno;

使用window函数，首先添加一个新列来跟踪每个partno的最大日期。所以from中的内部查询将生成这样的行，其中一列添加到原始数据集中，

| 1      | 2018-05-12 00:00:00 | 15    | $100   | $90       |**2018-05-13 00:00:00**|

突出显示的字段是添加到数据集的字段，它是该零件号日期范围内的最后一个日期！
现在从这个结果中，我们可以通过检查date列与我们添加的新列相等的行来提取利润，这实际上是通过从零售中减去批发，再乘以售出的商品来计算最后一个日期的利润。
ps：销售商品的逻辑是按零件号分组，并从最大值（inv）中减去最小值（inv）
链接到sql fiddle