Question

我正在尝试创建一个手动计算的列，以跟踪当前库存。

当前，我有一个看起来像这样的表：

| Group        | Part | Operation Type | Transaction Amount |
|--------------|------|----------------|--------------------|
| Concrete     | A    | STOCK          | 100                |
| Concrete     | A    | Buy            | 25                 |
| Concrete     | A    | Make           | -10                |
| Concrete     | A    | Make           | -10                |
| Concrete     | A    | Make           | -10                |
| Concrete     | A    | Make           | -10                |
| Concrete     | A    | Make           | -10                |
| Concrete     | B    | STOCK          | -10                |
| Concrete     | B    | Make           | -10                |
| Concrete     | B    | Make           | -10                |
| Concrete     | B    | Make           | -10                |
| Concrete     | B    | Make           | -10                |
| Concrete     | B    | Make           | 150                |
| Construction | C    | STOCK          | 10                 |
| Construction | C    | Make           | -1                 |
| Construction | C    | Make           | -1                 |
| Construction | C    | Make           | -1                 |
| Construction | C    | Make           | -1                 |
| Construction | D    | STOCK          | 5                  |
| Construction | D    | Make           | -5                 |

该表首先按group排序，然后按part排序，然后STOCK始终显示为第一个值。这个想法是创建一个新的手动计算的列curr_inventory，该列使我们可以跟踪当前库存，并查看给定零件，给定组的库存是否或何时下降到0以下。

理想情况下，最终结果将如下所示：

|     Group    | Part | Operation Type | Transaction Amount | New_Inventory_Column |
|:------------:|:----:|:--------------:|:------------------:|:--------------------:|
|   Concrete   |   A  |      STOCK     |         100        |          100         |
|   Concrete   |   A  |       Buy      |         25         |          125         |
|   Concrete   |   A  |      Make      |         -10        |          115         |
|   Concrete   |   A  |      Make      |         -10        |          105         |
|   Concrete   |   A  |      Make      |         -10        |          95          |
|   Concrete   |   A  |      Make      |         -10        |          85          |
|   Concrete   |   A  |      Make      |         -10        |          75          |
|   Concrete   |   B  |      STOCK     |         10         |          10          |
|   Concrete   |   B  |      Make      |         -10        |           0          |
|   Concrete   |   B  |      Make      |         -10        |          -10         |
|   Concrete   |   B  |      Make      |         -10        |          -20         |
|   Concrete   |   B  |      Make      |         -10        |          -30         |
|   Concrete   |   B  |      Make      |         150        |          120         |
| Construction |   C  |      STOCK     |         10         |          10          |
| Construction |   C  |      Make      |         -1         |           9          |
| Construction |   C  |      Make      |         -1         |           8          |
| Construction |   C  |      Make      |         -1         |           7          |
| Construction |   C  |      Make      |         -1         |           6          |
| Construction |   D  |      STOCK     |          5         |           5          |
| Construction |   D  |      Make      |         -5         |           0          |

最终结果将是一列，该列在零件号更改且操作类型为STOCK时启动，然后开始（使用交易金额）计算当前库存量。我不确定从哪里开始执行SQL查询。直观地，伪代码看起来像：

for each row in table:
    if operation_type == "stock":
        curr_inv = stock.value
    else:
        curr_inv = previous_curr_inv + transaction_amount

但是，我不确定如何开始为此编写SQL。我通常尝试发布正在使用的SQL，但我什至不知道从哪里开始。我在网上在线浏览了各种帖子，包括帖子like this，and this，and this，但是我看不到如何将所选答案用作解决方案。

Answer 1

我使用了窗口函数来计算运行总计。

我在子查询中添加了row_number列。

尝试一下：

select t1."Group",t1."Part",t1."Operation Type", t1."Transaction Amount",
sum(t1."Transaction Amount") over (partition by t1."Group",t1."Part" order by t1.rownumber)
from (
select row_number() over (order by null) as rownumber, t.*
from test t ) t1

测试结果：

DB<>Fiddle

Answer 2

SQL表表示无序集。为了使事情井井有条，您需要一列排序。在下面的代码中，我将在此列中使用?。

然后，您需要累计金额：

select t.*,
       sum(case when operation_type in ('STOCK'),
                when operation_type in ('Make') then - amount
                when operation_type in ('Buy') then - amount
                else 0
           end) over (partition by group, part order by ?)
from t;

根据先前的行创建计算字段Oracle SQL

2 个答案: