Question

我需要一种使用SQL-Server来计数当前行之后具有特定类型的行数的方法，但是只能计数到特定类型的下一行，此时计数将重新开始。在下面的示例中，我需要在“时间表”之后的“子条款”或“时间表”之后的“时间表部分”进行计数。

示例数据：

Type                  Title
--------------------------------------------------
Clause                20
Sub-Clause            20.1 A Sub-Clause
Sub-Clause Section    20.1-1 A Sub-Clause Section
Schedule              Schedule 1 to Blah
Schedule Section      1.0 X
Schedule Section      2.0 X
Schedule              Schedule 2 to Blah
Sub-Clause            20.2 A Sub-Clause
Clause                21
Schedule              Schedule 1 to Clause 21
Schedule Section      1.0 X
Schedule Section      2.0 X

我需要的输出将是这样的：

Type                  Title                          Count
---------------------------------------------------------------
Clause                20                             0
Sub-Clause            20.1 A Sub-Clause              1
Schedule              Schedule 1 to Blah             2
Schedule              Schedule 2 to Blah             0
Sub-Clause            20.2 A Sub-Clause              0
Clause                21                             0
Schedule              Schedule 1 to Clause 21        2

我已经接近使用row_number（）标识每一行，然后计算出当前行与下一行之间的差异以及过滤出标准行的解决方案。这是一种很粗糙的方法，我相信有更好的方法可以实现这一目标。另外，我的方法不适用于最后一行，因为它将为计数返回NULL。我的代码的简化版本是这样的：

SELECT
    Type
  , Title
  , LEAD(myTable.Row#, 1) OVER(ORDER BY Row#) - Row# -1 AS 'Count'
FROM
  (
  SELECT
    ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) AS Row#
  , Type
  , Title
  FROM Content      
  ORDER BY Row# 
  ) AS myTable
WHERE myTable.Type in ('Clause', 'Sub-Clause', 'Schedule')

关于可以实现我所需要的更好方法的任何建议吗？

Answer 1

我做一些假设。有一个ID列可以定义行的顺序

-- Schema
declare @Content table
(
    ID  int identity,
    Type    varchar(20),
    Title   varchar(50)
)

-- Sample Data
insert into @Content
SELECT  'Clause',                '20'                       union all
SELECT  'Sub-Clause',            '20.1 A Sub-Clause'        union all
SELECT  'Sub-Clause Section',    '20.1-1 A Sub-Clause Section'  union all
SELECT  'Schedule',              'Schedule 1 to Blah'       union all
SELECT  'Schedule Section',      '1.0 X'                    union all
SELECT  'Schedule Section',      '2.0 X'                    union all
SELECT  'Schedule',              'Schedule 2 to Blah'       union all
SELECT  'Sub-Clause',            '20.2 A Sub-Clause'        union all
SELECT  'Clause',                '21'                       union all
SELECT  'Schedule',              'Schedule 1 to Clause 21'  union all
SELECT  'Schedule Section',      '1.0 X'                    union all
SELECT  'Schedule Section',      '2.0 X'

-- Query
; with 
cte as
(
    -- grp is to identify group by rows with same Type
    -- first set of `Schedule Section` 
    --     and last set of `Schedule Section` will have diff `grp`
    SELECT  *, grp = ID - row_number() over (PARTITION BY Type ORDER BY ID)
    FROM    @Content c
),
type_cnt as
(
    -- count no of rows for each type
    SELECT  ID = min(ID), Type, cnt = count(*)
    FROM    cte c
    WHERE   c.Type not in ('Clause', 'Sub-Clause', 'Schedule')
    GROUP BY Type, grp
)
SELECT  c.ID, c.Type, c.Title, cnt = isnull(n.cnt, 0)
FROM    cte c
        outer apply
        (
            SELECT  TOP 1 x.cnt
            FROM    type_cnt x
            WHERE   x.ID    = c.ID + 1
        ) n
WHERE   c.Type in ('Clause', 'Sub-Clause', 'Schedule')
ORDER BY ID

/*  Result
1   Clause      20          0
2   Sub-Clause  20.1 A Sub-Clause   1
4   Schedule    Schedule 1 to Blah  2
7   Schedule    Schedule 2 to Blah  0
8   Sub-Clause  20.2 A Sub-Clause   0
9   Clause      21          0
10  Schedule    Schedule 1 to Clause 21 2
*/

SQL：计算行数，直到满足条件

1 个答案: