SQL Server - 动态数据透视表 - SQL 注入
很抱歉问了这么长的问题,但这包含了我用来测试场景的所有 SQL,希望能够清楚地说明我在做什么。
我正在构建一些动态 SQL 以在 SQL Server 2005 中生成 PIVOT 表。
下面是执行此操作的代码。通过各种选择显示原始数据、使用 GROUP BY 的值以及我想要的 PIVOT 中的值。
BEGIN TRAN
--Create the table
CREATE TABLE #PivotTest
(
ColumnA nvarchar(500),
ColumnB nvarchar(500),
ColumnC int
)
--Populate the data
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'X', 1)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Y', 2)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Z', 3)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'X', 4)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Y', 5)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Z', 6)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'X', 7)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Y', 8)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Z', 9)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'X', 10)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'Y', 11)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'Z', 12)
--The data
SELECT * FROM #PivotTest
--Group BY
SELECT
ColumnA,
ColumnB,
SUM(ColumnC)
FROM
#PivotTest
GROUP BY
ColumnA,
ColumnB
--Manual PIVOT
SELECT
*
FROM
(
SELECT
ColumnA,
ColumnB,
ColumnC
FROM
#PivotTest
) DATA
PIVOT
(
SUM(DATA.ColumnC)
FOR
ColumnB
IN
(
[X],[Y],[Z]
)
) PVT
--Dynamic PIVOT
DECLARE @columns nvarchar(max)
SELECT
@columns =
STUFF
(
(
SELECT DISTINCT
', [' + ColumnB + ']'
FROM
#PivotTest
FOR XML PATH('')
), 1, 1, ''
)
EXEC
('
SELECT
*
FROM
(
SELECT
ColumnA,
ColumnB,
ColumnC
FROM
#PivotTest
) DATA
PIVOT
(
SUM(DATA.ColumnC)
FOR
ColumnB
IN
(
' + @columns + '
)
) PVT
')
--The data again
SELECT * FROM #PivotTest
ROLLBACK
每当我生成任何动态 SQL 时,我总是会意识到 SQL 注入攻击。因此,我添加了以下行和其他 INSERT 语句。
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'FOO])) PVT; DROP TABLE #PivotTest;SELECT ((GETDATE()--', 1)
当我现在运行 SQL 时,你瞧,EXEC 部分删除了 #PivotTest 表,从而使最后一个 SELECT 失败。
所以我的问题是,有谁知道一种在不冒 SQL 注入攻击风险的情况下执行动态 PIVOT 的方法吗?
Sorry for the long question but this contains all the SQL I've used to test the scenario to hopefully make it clear as to what I'm doing.
I'm build up some dynamic SQL to produce a PIVOT table in SQL Server 2005.
Below is code to do this. With various selects showing the raw data the values using GROUP BY and the values in a PIVOT as I want them.
BEGIN TRAN
--Create the table
CREATE TABLE #PivotTest
(
ColumnA nvarchar(500),
ColumnB nvarchar(500),
ColumnC int
)
--Populate the data
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'X', 1)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Y', 2)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Z', 3)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'X', 4)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'Y', 5)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Z', 6)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'X', 7)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Y', 8)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('B', 'Z', 9)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'X', 10)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'Y', 11)
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('C', 'Z', 12)
--The data
SELECT * FROM #PivotTest
--Group BY
SELECT
ColumnA,
ColumnB,
SUM(ColumnC)
FROM
#PivotTest
GROUP BY
ColumnA,
ColumnB
--Manual PIVOT
SELECT
*
FROM
(
SELECT
ColumnA,
ColumnB,
ColumnC
FROM
#PivotTest
) DATA
PIVOT
(
SUM(DATA.ColumnC)
FOR
ColumnB
IN
(
[X],[Y],[Z]
)
) PVT
--Dynamic PIVOT
DECLARE @columns nvarchar(max)
SELECT
@columns =
STUFF
(
(
SELECT DISTINCT
', [' + ColumnB + ']'
FROM
#PivotTest
FOR XML PATH('')
), 1, 1, ''
)
EXEC
('
SELECT
*
FROM
(
SELECT
ColumnA,
ColumnB,
ColumnC
FROM
#PivotTest
) DATA
PIVOT
(
SUM(DATA.ColumnC)
FOR
ColumnB
IN
(
' + @columns + '
)
) PVT
')
--The data again
SELECT * FROM #PivotTest
ROLLBACK
Anytime that I produce any dynamic SQL I'm always aware of SQL Injection attacks. Therefore I've added the following line with the other INSERT statements.
INSERT INTO #PivotTest (ColumnA, ColumnB, ColumnC) VALUES('A', 'FOO])) PVT; DROP TABLE #PivotTest;SELECT ((GETDATE()--', 1)
When I now run the SQL, low and behold, the EXEC part drops the #PivotTest table thus making the last SELECT fail.
So my question is, does anyone know of a way to perform a dynamic PIVOT without risking SQL Injection attacks?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(3)
我们已经做了很多与您的例子类似的工作。我们并不担心 SQL 注入,部分原因是我们对所传输的数据拥有完全的控制权——恶意代码不可能通过 ETL 进入我们的数据仓库。
一些想法和建议:
哈。需要编写所有这些才能记住函数 QUOTENAME()。快速测试似乎表明,像这样将其添加到代码中会起作用(您会得到一个错误,而不是删除的临时表):
这应该适用于枢轴(和逆枢轴)情况,因为您几乎总是必须[括号]你的价值观。
We've done a lot of work similar to your example. We haven't worried about SQL injenction, in part because we have complete and total control over the data being pivoted--there's just no way malicious code could get through ETL into our data warehouse.
Some thoughts and advice:
Hah. It took writing all that to remember function QUOTENAME(). A quick test would seem to indicate that adding it to your code like so would work (You'll get an error, not a dropped temp table):
This should work for pivot (and unpivot) situations, since you almost always have to [bracket] your values.
一些重构...
用法示例:
A bit of refactoring...
Usage example: