计算某些字段后如何从另一个表创建新表
Posted
技术标签:
【中文标题】计算某些字段后如何从另一个表创建新表【英文标题】:How can create a new table from another after Calculate Some Field 【发布时间】:2019-11-02 09:01:05 【问题描述】:我有一个名为 WorkSpace
的表,有 4 列。
此表中的每条记录通过Parent_id
列引用它们的父记录。
您应该继续这个循环,直到您到达主要父级以及View_Parent_id
列中指定的所有这些步骤。
主父编号的ID由1指定。
所以现在我想创建一个NEW_WorkSpace
表并将View_Parent_id
列中的每个值分隔到一个单独的列中。
View_Parent_id
列最终值为 15,因此我们需要新表中的 15 列 (Level_1...Level_15)。
我在查询中使用了循环,但速度很慢。
WorkSpace
表有 150 万行。
WorkSpace
表:
+-----+-------+-----------+------------------+
| W_ID| Title | Parent_id | View_Parent_id |
+-----+-------+-----------+------------------+
| 1 | AAA | null | null |
| 2 | BV | 1 | 1 |
| 3 | CX | 2 | 1+2 |
| 4 | DSO | 2 | 1+2 |
| 5 | ER | 3 | 1+2+3 |
| 6 | ER | 5 | 1+2+3+5 |
| ... | ... | ... | ... |
| 1000| MNV | 1 | 1 |
| 1001| SF | 1000 | 1+1000 |
| 1002| EDD | 1000 | 1+1000 |
| 1003| YSG | 1001 | 1+1000+1001 |
| 1004| RPO | 1003 | 1+1000+1001+1003 |
+-----+-------+-----------+------------------+
NEW_WorkSpace
表:
+-----+-------+-----------+---------+---------+---------+-----+----------+
| ID | W_id | Parent_id | Level_1 | Level_2 | Level_3 | ... | Level_15 |
+-----+-------+-----------+---------+-------- +---------+-----+----------+
| 100 | 1 | null | AAA | | | ... | |
| 101 | 2 | 1 | AAA | BV | | ... | |
| 102 | 3 | 2 | AAA | BV | | ... | |
| 103 | 4 | 2 | AAA | BV | CX | ... | |
| 104 | 5 | 3 | AAA | BV | CX | ... | |
| ... | ... | ... | ... | ... | ... | ... | ... |
+-----+-------+-----------+---------+---------+---------+-----+----------+
我的代码:
BEGIN
DECLARE @W_ID decimal(20, 0);
DECLARE @parent_id decimal(20, 0);
DECLARE @Level1 nvarchar(MAX);
DECLARE @Level2 nvarchar(MAX);
DECLARE @Level3 nvarchar(MAX);
DECLARE @Level4 nvarchar(MAX);
DECLARE @Level5 nvarchar(MAX);
DECLARE @Level6 nvarchar(MAX);
DECLARE @Level7 nvarchar(MAX);
DECLARE @Level8 nvarchar(MAX);
DECLARE @Level9 nvarchar(MAX);
DECLARE @Level10 nvarchar(MAX);
DECLARE @Level11 nvarchar(MAX);
DECLARE @Level12 nvarchar(MAX);
DECLARE @Level13 nvarchar(MAX);
DECLARE @Level14 nvarchar(MAX);
DECLARE @Level15 nvarchar(MAX);
DECLARE @titles_tmp nvarchar(MAX);
DECLARE @cont_spilit_tittle int;
DECLARE @parent_titles_tmp nvarchar(MAX);
DECLARE @cont_tmp int;
DECLARE @cont int;
SELECT @cont = COUNT(*) FROM dbo.WorkSpace ;
SET @cont_tmp = 0;
WHILE (@cont_tmp < @cont)
BEGIN
SET @W_ID = (SELECT dbo.WorkSpace.W_ID FROM dbo.WorkSpace
ORDER BY W_ID ASC OFFSET @cont_tmp ROWS FETCH NEXT 1 ROWS ONLY)
SET @parent_id = (SELECT dbo.WorkSpace.parent_id FROM dbo.WorkSpace
ORDER BY W_ID ASC OFFSET @cont_tmp ROWS FETCH NEXT 1 ROWS ONLY)
SET @titles_tmp = (SELECT dbo.WorkSpace.title FROM dbo.WorkSpace
ORDER BY W_ID ASC OFFSET @cont_tmp ROWS FETCH NEXT 1 ROWS ONLY)
SET @parent_titles_tmp = (SELECT dbo.WorkSpace.parent_titles FROM dbo.WorkSpace
ORDER BY W_ID ASC OFFSET @cont_tmp ROWS FETCH NEXT 1 ROWS ONLY)
IF OBJECT_ID('tempdb..#MyTempTable') IS NOT NULL
DROP TABLE #MyTempTable
SELECT IDENTITY(INT, 1, 1) AS 'RowID', *
INTO #MyTempTable
FROM StringSplitXML(@parent_titles_tmp, '+')
INSERT INTO #MyTempTable
VALUES (@titles_tmp)
SET @cont_spilit_tittle = (SELECT COUNT(*) FROM #MyTempTable)
IF(@cont_spilit_tittle < 0)
SET @cont_spilit_tittle = 1
WHILE (@cont_spilit_tittle < 15)
BEGIN
INSERT INTO #MyTempTable VALUES ('')
SET @cont_spilit_tittle = CAST(@cont_spilit_tittle AS INT) + 1
END
SET @Level1 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 0 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level2 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 1 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level3 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 2 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level4 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 3 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level5 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 4 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level6 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 5 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level7 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 6 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level8 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 7 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level9 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 8 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level10 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 9 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level11 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 10 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level12 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 11 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level13 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 12 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level14 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 13 ROWS FETCH NEXT 1 ROWS ONLY)
SET @Level15 = (SELECT Value FROM #MyTempTable
ORDER BY RowID ASC OFFSET 14 ROWS FETCH NEXT 1 ROWS ONLY)
INSERT INTO [].[dbo].[NEW_WorkSpace]
([W_ID], [parent_id],
[Level1], [Level2], [Level3], [Level4], [Level5],
[Level6], [Level7], [Level8], [Level9], [Level10],
[Level11], [Level12], [Level13], [Level14], [Level15])
VALUES (@W_ID, @parent_id,
@Level1, @Level2, @Level3, @Level4, @Level5,
@Level6, @Level7, @Level8, @Level9, @Level10,
@Level11, @Level12, @Level13, @Level14, @Level15)
SET @cont_tmp = CAST(@cont_tmp AS INT) + 1
END
RETURN
END
感谢您的帮助。
【问题讨论】:
请向我们展示您的尝试 @squirrel 代码已添加。 正如您提到的 pivote,但在查看您的代码和输出时没有 pivote 概念?? @nits-patel 我只是猜测它可能是枢轴的解决方案,但我没有自己写 是的,因为WHILE
是在 SQL 中做任何事情的最糟糕的方法之一;它不是一种编程语言。查询语言中的WHILE
本来就很慢。
【参考方案1】:
我会在view_parent_id
上使用字符串操作来解决这个问题。然后聚合最终结果:
with cte as (
select w_id, parent_id, view_parent_id,
0 as lev, convert(varchar(max), concat(view_parent_id, '+', w_id, '+')) as parents
from t
union all
select w_id, parent_id, view_parent_id,
1 + lev,
convert(int, left(parents, charindex('+', parents) - 1)),
stuff(parents, 1, charindex('+', parents), '')
from cte
where parents <> ''
)
select w_id, parent_id, view_parent_id,
max(case when lev = 1 then parent_title end) as title_1,
max(case when lev = 2 then parent_title end) as title_2,
max(case when lev = 3 then parent_title end) as title_3,
max(case when lev = 4 then parent_title end) as title_4,
max(case when lev = 5 then parent_title end) as title_5
from (select cte.*, t.title as parent_title, count(*) over (partition by cte.w_id) as cnt
from cte join
t
on t.w_id = cte.parent
where lev > 0
) cte
group by w_id, parent_id, view_parent_id;
Here 是一个 dbfiddle。
至于处理。查询中最昂贵的部分可能是递归 CTE 之后的聚合。递归部分没有做任何连接,所以它应该相当快(字符串操作可能很慢)。
获取标签的连接使用了正确的类型,因此可以使用w_id
上的索引。
【讨论】:
您好 Linoff 先生,我应该一如既往地说您的回答对我有所帮助,但我认为您忘记了我的问题的一部分。实际上答案是正确的而且太快了,但是您错过了每条记录的最后一级。 @JavadAbedi 。 . .这很容易通过将当前的w_id
添加到父母列表中来解决。但是,该列似乎应该只有包含当前节点的完整路径。无论如何,我调整了答案,您可以在 dbfiddle. 中看到它
是的,没错,我将 w_id 添加到父列表中。这个答案解决了我的问题。非常感谢 Linoff 先生。【参考方案2】:
好吧,这个丑陋的假设您最多有 15 个级别,而您的预期结果是错误的;正如我所拥有的W_ID
4
对于Level_3
的值为'DSO'
,但您的预期结果为CX
,即使该行没有以任何方式链接到CX
。
USE Sandbox;
GO
CREATE TABLE dbo.YourTable (W_ID int NOT NULL,
Title varchar(3) NOT NULL,
Parent_id int NULL,
View_Parent_id varchar(100) NULL);
GO
INSERT INTO dbo.YourTable
VALUES (1,'AAA',NULL,NULL),
(2,'BV',1,'1'),
(3,'CX',2,'1+2'),
(4,'DSO',2,'1+2'),
(5,'ER',3,'1+2+3'),
(6,'ER',5,'1+2+3+5');
GO
SELECT *
FROM dbo.YourTable;
GO
WITH rCTE AS(
SELECT (YT.W_ID + 99) AS ID,
YT.W_ID ,
YT.Parent_id,
Title AS level_1,
CONVERT(varchar(3),NULL) AS Level_2,
CONVERT(varchar(3),NULL) AS Level_3,
CONVERT(varchar(3),NULL) AS Level_4,
CONVERT(varchar(3),NULL) AS Level_5,
CONVERT(varchar(3),NULL) AS Level_6,
CONVERT(varchar(3),NULL) AS Level_7,
CONVERT(varchar(3),NULL) AS Level_8,
CONVERT(varchar(3),NULL) AS Level_9,
CONVERT(varchar(3),NULL) AS Level_10,
CONVERT(varchar(3),NULL) AS Level_11,
CONVERT(varchar(3),NULL) AS Level_12,
CONVERT(varchar(3),NULL) AS Level_13,
CONVERT(varchar(3),NULL) AS Level_14,
CONVERT(varchar(3),NULL) AS Level_15,
1 AS [Level]
FROM dbo.YourTable YT
WHERE YT.Parent_id IS NULL
UNION ALL
SELECT (YT.W_ID + 99) AS ID,
YT.W_ID,
YT.Parent_id,
r.Level_1,
CASE r.[Level] WHEN 1 THEN YT.Title ELSE r.Level_2 END AS Level_2,
CASE r.[Level] WHEN 2 THEN YT.Title ELSE r.Level_3 END AS Level_3,
CASE r.[Level] WHEN 3 THEN YT.Title ELSE r.Level_4 END AS Level_4,
CASE r.[Level] WHEN 4 THEN YT.Title ELSE r.Level_5 END AS Level_5,
CASE r.[Level] WHEN 5 THEN YT.Title ELSE r.Level_6 END AS Level_6,
CASE r.[Level] WHEN 6 THEN YT.Title ELSE r.Level_7 END AS Level_7,
CASE r.[Level] WHEN 7 THEN YT.Title ELSE r.Level_8 END AS Level_8,
CASE r.[Level] WHEN 8 THEN YT.Title ELSE r.Level_9 END AS Level_9,
CASE r.[Level] WHEN 9 THEN YT.Title ELSE r.Level_10 END AS Level_10,
CASE r.[Level] WHEN 10 THEN YT.Title ELSE r.Level_11 END AS Level_11,
CASE r.[Level] WHEN 11 THEN YT.Title ELSE r.Level_12 END AS Level_12,
CASE r.[Level] WHEN 12 THEN YT.Title ELSE r.Level_13 END AS Level_13,
CASE r.[Level] WHEN 13 THEN YT.Title ELSE r.Level_14 END AS Level_14,
CASE r.[Level] WHEN 14 THEN YT.Title ELSE r.Level_15 END AS Level_15,
r.[Level] + 1 AS [Level]
FROM dbo.YourTable YT
JOIN rCTe r ON YT.Parent_id = r.W_ID)
SELECT r.ID,
r.W_ID,
r.Parent_id,
r.Level_1,
r.Level_2,
r.Level_3,
r.Level_4,
r.Level_5,
r.Level_6,
r.Level_7,
r.Level_8,
r.Level_9,
r.Level_10,
r.Level_11,
r.Level_12,
r.Level_13,
r.Level_14,
r.Level_15
FROM rCTE r;
GO
DROP TABLE dbo.YourTable;
db<>fiddle
【讨论】:
感谢您的帮助,我的问题已解决,但速度很慢。请投票给我的问题。 分层数据对于大型数据集总是很慢。如果性能很重要,请使用hierachyid
, @Javadadebi 。我不确定你对你的问题投票是什么意思,你是要我投票吗?如果这确实解决了问题,您应该将其标记为解决方案,以便未来的读者知道答案是有帮助的。
是的,你说得对。我不知道为什么有些人拒绝投票给某个问题,但你的答案太接近了,但我应该说它太慢了,例如 100 条记录需要 20 秒,想象一下 150 万条记录。感谢您的帮助。
那么你应该看看你的索引。如果 100 行需要 100 秒,那么您有一些索引严重不良的表,或者硬件严重不足的服务器。
至于否决票,我怀疑是在您添加尝试之前。不过,只有投反对票的人才能告诉你。以上是关于计算某些字段后如何从另一个表创建新表的主要内容,如果未能解决你的问题,请参考以下文章
如何将自动填充的表单记录(从另一条记录创建)添加到访问表中?