代码之家 › 专栏 › 技术社区 › JohnIdol

SQL2005中动态透视所需的帮助

dynamic-pivot pivot-table tsql sql-server-2005

JohnIdol · 技术社区 · 15 年前

我有一个名称-值对表,在其中存储标记:

TAGID | NAME | VALUE

我有一张表,上面写着这些标签适用于

THINGID | TAGID

如果列/字段都是给定thingid的可能标记名(标记表中的名称字段),并且值是相应的标记值,则需要查询来生成结果集。

THINGID | TAGNAME1 | TAGNAME2 | ... |etc.

我可以找到固定列的例子,但没有这样的例子。

1 回复 | 直到 15 年前

Stanislav Kniazev 15 年前

与稍有不同的模式(thing_id,tag_name,tag_value)相似的东西-不带tags表:

CREATE PROCEDURE dbo.PivotData(@table VARCHAR(128), @basefield
VARCHAR(128), @namefield VARCHAR(128), @valuefield VARCHAR(128))
AS
DECLARE
 @sql NVARCHAR(MAX)
SET @sql =
 'DECLARE
 @colName VARCHAR(128),
 @sqlBegin NVARCHAR(MAX),
 @sqlMiddle NVARCHAR(MAX),
 @sqlEnd NVARCHAR(MAX),
 @counter INT
 SET @counter = 1
 SET @sqlBegin = N''SELECT DISTINCT t0.'' + ''' + QUOTENAME(@basefield) + '''
 SET @sqlMiddle = N'' FROM '' + ''' + QUOTENAME(@table) + ''' + '' AS t0 ''
 DECLARE cols CURSOR FOR
 SELECT DISTINCT TOP 100 PERCENT ' + QUOTENAME(@namefield) + '
 FROM ' + QUOTENAME(@table) + '
 WHERE ' + QUOTENAME(@basefield) + ' IS NOT NULL
 ORDER BY ' + QUOTENAME(@namefield) + '
 OPEN cols
 FETCH NEXT FROM cols INTO @colName
 WHILE @@FETCH_STATUS = 0 BEGIN
 SET @sqlBegin = @sqlBegin + '', t'' + CAST(@counter AS VARCHAR) +
''.'' + ''' + QUOTENAME(@valuefield) + ''' + '' AS '' +
QUOTENAME(@colName) + ''''
 SET @sqlMiddle = @sqlMiddle + '' LEFT OUTER JOIN '' + ''' +
  QUOTENAME(@table) + ''' + '' AS t'' + CAST(@counter AS VARCHAR) +
'' ON t0.'' + ''' +
  QUOTENAME(@basefield) + ''' + '' = t'' + CAST(@counter AS VARCHAR)
+ ''.'' + ''' +
  QUOTENAME(@basefield) + ''' + '' AND t'' + CAST(@counter AS
VARCHAR) + ''.'' + ''' + QUOTENAME(@namefield) + ''' + ''='' +
QUOTENAME(@colName, '''''''') + ''''
 SET @counter = @counter + 1
 FETCH NEXT FROM cols INTO @colName
 END
 CLOSE cols
 DEALLOCATE cols
 SET @sqlEnd = '' WHERE t0.'' + ''' + QUOTENAME(@basefield) + ''' + ''
IS NOT NULL''
 DECLARE @sql NVARCHAR(MAX)
 SET @sql = @sqlBegin + @sqlMiddle + @sqlEnd
 EXEC sp_executesql @sql'
EXEC sp_executesql @sql
RETURN 0
GO
CREATE TABLE test_data (
id int identity primary key,
person_id int,
person_data_field VARCHAR (128),
person_data_value VARCHAR (128)
)
GO
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (1, 'Name', 'John')
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (1, 'Surname', 'Smith')
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (1, 'Email', 'John@Smith.com')
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (2, 'Name', 'Sarah')
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (2, 'Surname', 'Lee')
INSERT INTO dbo.test_data (person_id, person_data_field, person_data_value)
VALUES (2, 'Phone', '012345678')
GO
EXEC [dbo].[PivotData] @table='test_data',
@basefield='person_id', @namefield='person_data_field', @valuefield='person_data_value'
GO

我发现这样做的速度太慢了,不再进一步调整它,但它可以满足您的要求。