SQL 查询列，如果在 "chain" 行内满足条件则为真

Question

这是的跟进。这会变得复杂。

我有一个如下所示的 table 用于跟踪特定类型事件的开始和结束：

+----------+-------------+------------------+------------------+
| PersonID | Weapon Used |     StartDate    |     EndDate      |
+----------+-------------|------------------+------------------+
|      006 | Knife       | 2019-12-01 09:30 | 2019-12-02 06:15 |
|      007 | Grenade     | 2019-12-01 12:15 | 2019-12-01 12:16 |
|      006 | Bat         | 2019-12-02 06:15 | 2019-12-02 15:15 |
|      006 | Pistol      | 2019-12-02 15:15 | 2019-12-02 23:01 |
|      007 | Pistol      | 2019-12-02 07:23 | 2019-12-04 08:30 |
+----------+-------------+------------------+------------------+

我正在尝试创建一个在以下条件下returns TRUE 的计算列：

行事件是 "chain" 1+ 背靠背事件的一部分。
AND 手枪在事件链中使用了 1 次以上。
AND 行事件具有链中最早的 StartDate。

像这样创建 table：

+----------+-------------+------------------+------------------+---------+
| PersonID | Weapon Used |     StartDate    |     EndDate      | Pistol+ |
+----------+-------------|------------------+------------------+---------+
|      006 | Knife       | 2019-12-01 09:30 | 2019-12-02 06:15 | TRUE    |
|      007 | Grenade     | 2019-12-01 12:15 | 2019-12-01 12:16 | FALSE   |
|      006 | Bat         | 2019-12-02 06:15 | 2019-12-02 15:15 | FALSE   |
|      006 | Pistol      | 2019-12-02 15:15 | 2019-12-02 23:01 | FALSE   |
|      007 | Pistol      | 2019-12-02 07:23 | 2019-12-04 08:30 | TRUE    |
+----------+-------------+------------------+------------------+---------+

这确实是一项艰巨的任务，但我必须为大量的行执行此任务，因此我拼命地避免必须用手和眼球来计算它。这可能吗？

Answer 1

您将需要 recursive cte 找到 grp 相关的连续行

seq是识别第一个的group

为了识别组是否包含 pistol，它使用 window 函数 sum() 和条件 case

with
cte as
(
    select  *, rn = row_number() over (partition by PersonID order by StartDate)
    from    yourtable
),
rcte as
(
    select  c.PersonID, c.[Weapon Used], c.StartDate, c.EndDate, c.rn, 
            grp = 1, seq = 1
    from    cte c
    where   rn  = 1

    union all

    select  c.PersonID, c.[Weapon Used], c.StartDate, c.EndDate, c.rn,
            grp   = case    when    r.EndDate = c.StartDate
                            then    r.grp
                            else    r.grp + 1
                            end,
            seq   = case    when    r.EndDate = c.StartDate
                            then    r.seq + 1
                            else    1
                            end
    from    cte c
            inner join rcte r   on  c.PersonID  = r.PersonID
                               and  c.rn        = r.rn + 1
)
select  *,
        case    when    seq = 1
                and     sum(case when [Weapon Used] = 'Pistol' then 1 else 0 end) 
                        over (partition by PersonID, grp) >= 1
                then    'TRUE'
                else    'FALSE'
                end
from    rcte;

SQL 查询列，如果在 "chain" 行内满足条件则为真

TSQL Query column which is true if condition is met within "chain" of rows

tsql

sql-server

sorting

calculated-columns