我在'result'字段的'dataset'表中有4行，它由＆符号（＆amp;）分隔。我想根据字段名称将“结果”字段划分为多个字段。我如何在SSIS或SQL中执行此操作？字段名称是高度的。如果没有.net脚本，那将是可行的。如果你注意到，字段名称是混乱的。
结果（fieldname）
deptid = 1＆amp; 名字 = din＆amp; 姓氏 = kal＆amp;的中间名 =琴
名字 = raj＆amp; 姓氏 = puli＆amp; 中间名 = kumar＆amp;的 DEPTID = 2
名字 = pavan＆amp; 姓氏 = gud＆amp;的中间名 = kumarrao
deptid = 7＆amp; 名字 = sha＆amp;的名字 =绞纱
输出
**deptid** **firstname** **lastname** **middlename**
1 din kal kum
2 raj puli kumar
pavan gud kumarrao
7 sha hank
答案 0 :(得分：1)
CROSS APPLY该表格具有某种形式的split string功能。使用这些结果创建动态SQL命令并执行。将这些结果保留在新表格中DROP旧表格。存储这样的数据是not okay。
随机SplitString：
IF NOT EXISTS ( SELECT 1
FROM sys.objects
WHERE name = 'SplitString'
AND type = 'TF' )
BEGIN
--DROP FUNCTION dbo.SplitString;
EXEC( 'CREATE FUNCTION dbo.SplitString() RETURNS @t TABLE ( x BIT ) AS BEGIN RETURN; END' );
END;
GO
ALTER FUNCTION dbo.SplitString
(
@String VARCHAR( MAX ),
@Delimiter VARCHAR( 16 )
) RETURNS @Values TABLE
(
SplitString VARCHAR( MAX )
) AS BEGIN
/*
dbo.SplitString
@Param:
String, the string to be split,
Delimiter, the delimiter to use.
@Returns:
Table, each tuple consisting of a tokenized value of the original string,
split by the provided delimiter. Empty expressions result in NULL values.
@Example:
SELECT *
FROM dbo.SplitString( 'This is a test', ' ' );
SELECT *
FROM dbo.SplitString( NULL, ' ' );
*/
DECLARE @i INTEGER,
@Split VARCHAR( MAX ),
@RowCount INTEGER;
SET @RowCount = 0;
SET @i = -1;
WHILE ( LEN( ISNULL( @String, '' ) ) > 0 )
BEGIN
SET @String = LTRIM( RTRIM( @String ) );
SET @i = CHARINDEX( @Delimiter, @String );
IF ( @i = 0 )
BEGIN
INSERT INTO @Values ( SplitString )
VALUES ( LTRIM( RTRIM( @String ) ) );
SET @String = '';
END ELSE BEGIN
INSERT INTO @Values ( SplitString )
VALUES ( LTRIM( RTRIM( LEFT( @String, @i - LEN( @Delimiter ) ) ) ) );
SET @String = RIGHT( @String, LEN( @String ) - @i );
END;
SET @RowCount = @RowCount + 1;
END;
IF ( @RowCount = 0 )
BEGIN
INSERT INTO @Values ( SplitString )
VALUES ( NULL );
END;
RETURN;
END;
GO
测试用例：
IF NOT EXISTS ( SELECT 1
FROM sys.objects
WHERE name = 'DropThisNotAtomicNonsense'
AND type = 'U' )
BEGIN
--DROP TABLE dbo.DropThisNotAtomicNonsense;
CREATE TABLE dbo.DropThisNotAtomicNonsense
(
DropThisNotAtomicNonsense_PK TINYINT IDENTITY( 1, 1 ) NOT NULL,
PRIMARY KEY ( DropThisNotAtomicNonsense_PK ),
Result VARCHAR( MAX )
);
INSERT INTO dbo.DropThisNotAtomicNonsense ( Result )
VALUES ( 'deptid=1 & firstname=din & lastname=kal & middlename=kum' ),
( 'firstname=raj & lastname=puli & middlename=kumar & deptid=2' ),
( 'firstname=pavan & lastname=gud & middlename=kumarrao' ),
( 'deptid=7 & firstname=sha & lastname=hank' );
END;
GO
解决的问题：
SET NOCOUNT ON;
DECLARE @Cols VARCHAR( MAX ),
@SQL NVARCHAR( MAX );
CREATE TABLE #t_KeyValue
(
OriginalEntity TINYINT,
FieldName VARCHAR( MAX ),
FieldValue VARCHAR( MAX )
);
INSERT INTO #t_KeyValue ( OriginalEntity, FieldName, FieldValue )
SELECT OriginalEntity = ugh.DropThisNotAtomicNonsense_PK,
FieldName = LEFT( x.SplitString, CHARINDEX( '=', x.SplitString ) - 1 ),
FieldValue = RIGHT( x.SplitString, CHARINDEX( '=', REVERSE( x.SplitString ) ) - 1 )
FROM dbo.DropThisNotAtomicNonsense ugh
CROSS APPLY dbo.SplitString( ugh.Result, '&' ) x;
CREATE TABLE #t_ColumnList
(
FieldName VARCHAR( MAX ),
Processed TINYINT NOT NULL DEFAULT 0
);
INSERT INTO #t_ColumnList ( FieldName )
SELECT DISTINCT FieldName
FROM #t_KeyValue;
IF NOT EXISTS ( SELECT 1
FROM sys.objects
WHERE name = 'NormalTable'
AND type = 'U' )
BEGIN
--DROP TABLE dbo.NormalTable;
CREATE TABLE dbo.NormalTable
(
NormalTable_PK TINYINT IDENTITY( 1, 1 ) NOT NULL,
PRIMARY KEY ( NormalTable_PK ),
OriginalEntity TINYINT
)
WHILE EXISTS ( SELECT 1
FROM #t_ColumnList
WHERE Processed = 0 )
BEGIN
SELECT @SQL = FieldName
FROM #t_ColumnList
WHERE Processed = 0;
UPDATE #t_ColumnList
SET Processed = 1
WHERE FieldName = @SQL;
SET @SQL = N'
ALTER TABLE dbo.NormalTable
ADD [' + REPLACE( @SQL, '''', '' ) + '] VARCHAR( 128 );'
EXECUTE dbo.sp_executesql @command = @SQL;
END;
SELECT @Cols = ISNULL( @Cols, '' ) + '[' + FieldName + '], '
FROM ( SELECT DISTINCT FieldName
FROM #t_KeyValue ) l
SET @Cols = LEFT( @Cols, LEN( @Cols ) - 1 );
SET @SQL = N'
INSERT INTO dbo.NormalTable ( OriginalEntity, ' + @Cols + ' )
SELECT OriginalEntity, ' + @Cols + '
FROM ( SELECT OriginalEntity, FieldName, FieldValue
FROM #t_KeyValue ) s
PIVOT ( MAX( FieldValue ) FOR
FieldName IN ( ' + @Cols + ' ) ) p;';
RAISERROR ( @SQL, 0, 1 ) WITH NOWAIT;
EXECUTE sp_executesql @command = @SQL;
END;
GO
DROP TABLE #t_ColumnList;
DROP TABLE #t_KeyValue;
GO
SELECT *
FROM dbo.NormalTable;
GO
DROP TABLE dbo.DropThisNotAtomicNonsense;
GO
SET NOCOUNT OFF;
现在，使用漂亮的新表，您可以使用ALTER ... COLUMN语句切换到这些列上的某些proper data types。
答案 1 :(得分：0)
with x as (
select cast('<e col="' + replace(replace(result, '=', '">'), '&', '</e><e col="') + '</e>' as xml) as x
from dataset
)
select x.value('(/e[contains(@col, "deptid")])[1]', 'int') as deptid
,x.value('(/e[contains(@col, "firstname")])[1]', 'nvarchar(255)') as firstname
,x.value('(/e[contains(@col, "lastname")])[1]', 'nvarchar(255)') as lastname
,x.value('(/e[contains(@col, "middlename")])[1]', 'nvarchar(255)') as middlename
from x
应该提供你想要的东西。它使用中间XML对象和xml处理来实现结果。中间列看起来像
<e col="deptid">1 </e><e col=" firstname">din </e><e col=" lastname">kal </e><e col=" middlename">kum</e>
此解决方案并非绝对万无一失，但如果您确定格式稳定且不包含完全奇怪的格式化条目，则应该可以正常工作。注意：我需要contains，因为SQL Server似乎在XPath中没有trim函数。如果数据中包含&或=用于其他目的，除了将字段名称与内容或分隔字段分开之外，这将失败，即i。即如果这些字符包含在名称中。