Question

我在表格中有一列包含以下示例数据：

20170426-31974-B
20170426-31980
20170426-32014
20170426-32609
20170426-32069-B
20170426-32649
20170426-32659
20170426-32669-B
20170426-32849

-B的值是参考值。所有其他文件（不带-B）应该有一个与之关联的引用文件。因此，引用（-B）之间的所有差距应该与之前的＆＃34; -B＆＃34;相关联。

无论如何，通过查询，将行与引用相关联＆＃34; -B＆＃34;？

我的意思是：

20170426-31980 -> reference to 20170426-31974-B
20170426-32014 -> reference to 20170426-31974-B
20170426-32609 -> reference to 20170426-31974-B
20170426-32649 -> reference to 20170426-32069-B
20170426-32659 -> reference to 20170426-32069-B
20170426-32849 -> refence to 20170426-32669-B

换句话说，我需要在-B之间获取所有文件。所有这些文件都应与之前的-B。

相关联

选择应返回类似

的内容

20170426-31980 || 20170426-31974-B
20170426-32014 || 20170426-31974-B
20170426-32609 || 20170426-31974-B
20170426-32649 || 20170426-32069-B
20170426-32659 || 20170426-32069-B
20170426-32849 || 20170426-32669-B

我尝试了this idea：

SELECT  FILE_NAME
FROM TBL_FILE_NAMES
CROSS APPLY(SELECT FILE_NAME) FROM TBL_FILE_NAMES WHERE file_name LIKE '%-B') CA_min(file_name)
CROSS APPLY(SELECT FILE_NAME) FROM TBL_FILE_NAMES WHERE file_name LIKE '%-B') CA_max(file_name)
WHERE [DATE] BETWEEN start_dt AND end_dt

但我相信我的逻辑错了。我应该尝试使用光标而不是查询吗？

经过一些评论后，订单很重要。这些名称是关于时间创建的。该文件的第一部分是yearmonthday-seconds。我的问题是每个文件都应该有一个引用（-B）。如何知道哪些文件与-B相关联？它是参考文件之间的日期。

20170426-31974-B.csv
20170426-31980.csv
20170426-31987.csv
20170426-31994.csv
20170426-32004.csv
20170426-32014.csv
20170426-32069-B.csv
20170426-32073.csv
20170426-32079.csv
20170426-32639.csv
20170426-32659.csv
20170426-32669-B.csv
20170426-32674.csv

在这种情况下，我有我的第一个参考号20170426-31974-B.csv。所有即将到来的文件（按时间顺序）直到我们达到第二个参考应该指向第一个参考。

参考文献1 = 20170426-31974-B.csv

与

20170426-31980.csv
20170426-31987.csv
20170426-31994.csv
20170426-32004.csv
20170426-32014.csv

参考文献2 = 20170426-32069-B.csv

与

20170426-32073.csv
20170426-32079.csv
20170426-32639.csv
20170426-32659.csv

参考文献3：20170426-32669-B.csv

与

20170426-32674.csv

Answer 1

尝试

declare @t table(col varchar(100));
insert @t(col)
values
('20170426-31974-B')
,('20170426-31980')
,('20170426-32014')
,('20170426-32609')
,('20170426-32069-B')
,('20170426-32649')
,('20170426-32659')
,('20170426-32669-B')
,('20170426-32849');

with parent as (
    select col, strt = left(col,14), nxt = coalesce(left(lead(col) over(order by col), 14), '99999999-99999')
    from @t
    where right(col,2) ='-B' 
)
select parent.col, t1.col
from parent
-- left -- probably
join @t t1 on len(t1.col) = 14 and t1.col between parent.strt and parent.nxt
order by parent.col, t1.col

Answer 2

根据描述，我认为这正是你要找的......

IF OBJECT_ID('tempdb..#TestData', 'U') IS NOT NULL 
DROP TABLE #TestData;

CREATE TABLE #TestData (
    SomeValue VARCHAR(20) NOT NULL PRIMARY KEY CLUSTERED
    );
INSERT #TestData (SomeValue) VALUES 
    ('20170426-31974-B.csv'),
    ('20170426-31980.csv'),
    ('20170426-32014.csv'),
    ('20170426-32609.csv'),
    ('20170426-32069-B.csv'),
    ('20170426-32649.csv'),
    ('20170426-32659.csv'),
    ('20170426-32669-B.csv'),
    ('20170426-32849.csv');

-- SELECT * FROM #TestData td;

--==================================

WITH 
    cte_BaseStarts AS (
        SELECT 
            td.SomeValue,
            BaseValue = IIF(RIGHT(td.SomeValue, 5) = 'B.csv', td.SomeValue, NULL)
        FROM 
            #TestData td
        ),
    cte_FillAssocBase AS (
        SELECT 
            bs.SomeValue,
            AssocBaseValue = MAX(bs.BaseValue) OVER (ORDER BY bs.SomeValue ASC)
        FROM 
            cte_BaseStarts bs
        )
SELECT 
    fab.SomeValue,
    fab.AssocBaseValue
FROM 
    cte_FillAssocBase fab
WHERE 
    fab.SomeValue NOT LIKE '%B.csv';

选择特定值

2 个答案: