SQLSERVER2000 用过程来模拟MDX分析报表的效果(1)

这个想法的产生比较特殊,当时我在某学会工作,我对数据仓库的学习也初见成效,也做了一两个基于数据仓库的产品,总觉得数据仓库部署比较麻烦,而我的客户仅仅是需要一张分析型的报表.他只要满足多维的旋转,聚合,产生一个层次性的报表就可以满足要求.
     基于这种情况,我想:能不能在数据库下模拟一张这样的报表来满足用户的需要?
     因为写过仓库下的二个基于WEB的展示组件,显示一个MDX返回的CellSet,无非就是分解各轴中的Members的Caption,再取各单元格的属性作为显示格式等.思路并不复杂,不过,因为要设置合并,过程还是比较复杂的.
     那如果,我在数据库中用存储过程返回一个类CellSet的DataTable,那么,不也就可以实现了模拟的效果了吗?
     好了,闲话不说了.看正题吧.

第一部分:相关的存储过程
这个过程的目的就是生成一个类MDX的CellSet的结果.
要解释这个过程,先说CellSet的结构.我觉得因为你在了解了CellSet的结构后,理解这个过程会更简单,而且,我这个过程的思路本身就是从CellSet中来.

1.Cellset 对象
表示多维查询的结果。它是从立方或其它单元集中选出的单元的集合

Connection                   '//AdomdClient.AdomdConnection连接对象
 |
 |-CellSet                   '//查询结果集.
   |
   |-Cell                    '//结果集中的单元
   |-Axes                    '//结果集的轴集合,
   |-ProPerties              '//提供者的 Cellset 信息,可用 Property 检索.

说明

使用直接、类似数组的访问来实现对 Cellset 中数据的检索。可以“深入”到特定的成员以获得该成员的数据。例如，下列代码将返回名为 cst 的单元集第一坐标轴第一位置中第一个成员的标题：
cst.Axes(0).Positions(0).Members(0).Caption

在单元集中没有当前单元的概念，而是使用 Item 方法从单元集中检索指定的 Cell 对象。Item 方法的参数确定检索哪个单元。可以指定单元的唯一序号值。也可通过使用单元集

每个轴上的位置号来检索单元。

使用 Cellset 对象的集合、方法和属性，可以：
通过设置它的 ActiveConnection 属性，使打开的连接与 Cellset 对象相关联。
使用 Open 方法执行并检索多维查询的结果。
使用 Item 方法从 Cellset 中检索 Cell。
使用 Axes 集合返回定义 Cellset 的 Axis 对象。
使用 Filteraxis 属性检索有关用于过滤 Cellset 中数据的维的信息。
使用 Source 属性返回或指定用于定义 Cellset 的查询。
使用 State 属性返回 Cellset 的当前状态（打开、关闭、正在执行或正在连接）。
使用 Close 方法关闭打开的 Cellset。
使用标准的 ADO Properties 集合检索特定提供者的 Cellset 信息。

下面的这个函数就是将一个CellSet转为一个DataTable,从源程序来,有删节.

    ''' <summary>
    ''' 将CellSet转为DataTable.
    ''' </summary>
    ''' <param name="CellSet">CellSet对象.</param>
    ''' <returns>DataTable</returns>
    ''' <remarks></remarks>
    Private Function AdoMdCellSetToDataTable(ByVal CellSet As Microsoft.analysisservices.AdomdClient.CellSet) As DataTable

        Dim Dt As New DataTable
        Dim Dc As New DataColumn
        Dim Dr As DaTarow = nothing
        Dim I As Int32
        Dim ColName As String = ""
        Dim sC As Int32
        Dim sR As Int32 = 0
        Dim RowID As Int32 = 0
        Dim SDr As DaTarow = nothing

        sC = 1
        For Each M As Microsoft.analysisservices.AdomdClient.Member In CellSet.Axes(1).Positions(0).Members
            sC += 1
            Dc = New DataColumn()
            Dc.ColumnName = Microsoft.VisualBasic.Space(sC) '//以空填充空白部分.
            Dt.Columns.Add(Dc)
        Next

        '//生成数据列对象
        sR = 0
        Dim R,C As Int32
        Dim ColID As Int16 = 0
        Dim AxexCount_0 As Int32 = CellSet.Axes(0).Positions.Count
        Dim AxexCount_1 As Int32 = CellSet.Axes(1).Positions.Count
        Dim CName As String

        R = 0 : C = 0
        For Each P As Microsoft.analysisservices.AdomdClient.Position In CellSet.Axes(0).Positions '//分解0轴,Mdx中的 ON Columns 轴.
            CName = ""
            R = 0 : ColID += 1
            For Each M As Microsoft.analysisservices.AdomdClient.Member In P.Members
                If R = 0 Then
                    CName = CName & "[" & Trim$(M.Caption) & "]"
                Else
                    CName = CName & ".[" & Trim$(M.Caption) & "]"
                End If

                R += 1
            Next
            Dc = New DataColumn()
            Dc.ColumnName = CName
            Dt.Columns.Add(Dc)
            C += 1
        Next

        '//添加行数据
        Dim CellVal As String
        Dim Pos As Int32 = 0
        Dim Ts As String
        Dim Cid As Int32

        For Each py As Microsoft.analysisservices.AdomdClient.Position In CellSet.Axes(1).Positions  '//分解1轴,Mdx中的 ON Rows 轴
            Dr = Dt.NewRow()
            RowID += 1
            I = 0
            '//行方向的名称列.
            For Each m As Microsoft.analysisservices.AdomdClient.Member In py.Members
                Ts = m.Caption
                Cid = I
                Dr(Cid) = Ts
                I += 1
            Next

            '//行数据
            For X As Int32 = 1 To AxexCount_0
                CellVal = "" & CellSet(Pos).FormattedValue
                If CellVal = "1.#INF" OrElse CellVal = "-1.#IND" OrElse CellVal = "1#I.NF0" OrElse CellVal = "1#I.NF0%" Then CellVal = "" 
                Dr(Cid+X) = CellVal
                Pos += 1
            Next
            Dt.Rows.Add(Dr)
        Next
        Return Dt

    End Function

 

效果:

 

 

观察上图,发现了什么,原来DataTable与CellSet是很象的.实质上,他们之间是可以通过前台的应用程序来转化,你如果将上图的DaTable 按 "].[" 来分解,按层合并相同项,DataTable也可以变成一个CellSet,有了这个思路,就可以写一个存储过程,将数据库中的数据按你的要求组织成要求的格式.将列的名称用自定义的分隔符分隔,就可以达到相同的目的了.
下面的sql过程就实现了这个目的.自定义的分隔符是 "$@".

效果:

 

/***********************************************************
--调用例子
EXECUTE Prc_CoRSSReport_V2 N'test_data',
    N'海关名称,洲名称,地区名称',
    N'商品名称,年,月',
    N'金额',' WHERE 海关编号 between ''01'' and ''02''   and 时间编号 between 20070401 and 20070401  and 类编号=''10000002'''
***********************************************************/

CREATE PROC Prc_CoRSSReport_V2 (
------------------------------------------------------
--- 模拟 MDX V2.0 版.
---          MSTOP 2008/3/21
------------------------------------------------------
    @nvr_TableName nvarchar(128),
    @nvr_OnRows  nvarchar(1024),  --行列表
    @nvr_OnColumns nvarchar(1024),  --列列表
    @nvr_OnValue nvarchar(64),  --值字段
    @nvr_Where  nvarchar(1024)='' --条件
) AS BEGIN

    DECLARE @NVR_CMD NVARCHAR(1024)
    DECLARE @NVR_Columns NVARCHAR(1024)
    DECLARE @NVR_TmpTableName NVARCHAR(128)
    DECLARE @NVR_NewColName varchar(16)
    DECLARE @nvr_sOnRows NVARCHAR(1024)
    DECLARE @nvr_OnRows_B NVARCHAR(1024)
    DECLARE @nvr_OnColumns_B nvarchar(1024)
    DECLARE @NVR_sqlCOM_00 NVARCHAR(4000)

    SET @nvr_OnRows_B=N'[' + REPLACE(@nvr_OnRows,N',',N'],[') + N']'
    SET @nvr_OnColumns_B=N'[' + REPLACE(@nvr_OnColumns,[') + N']'

    SET @nvr_sOnRows=@nvr_OnRows
    SET @NVR_NewColName=N'Val_' + LEFT(REPLACE(NEWID(),'-',''),6)
    SET @NVR_TmpTableName=N'@TAB_TABLE'
    --提取数据到表变量中.
    SET @NVR_sqlCOM_00=N'DECLARE ' + @NVR_TmpTableName + N' TABLE ([' + REPLACE(@nvr_OnRows,N'] VARCHAR(128),[') + N'] VARCHAR(128),[' + @NVR_NewColName

+ N'] VARCHAR(128),[' + @nvr_OnValue + N'] FLOAT PRIMARY KEY  (' + @nvr_OnRows_B + N',[' + @NVR_NewColName + N']))'
    --合并列的名称.分隔符为 $@,也可以指定其它的分隔符.
    SET @NVR_Columns=N'RTRIM([' + REPLACE(@nvr_OnColumns,N']) + ''$@''+RTRIM([') + N'])+''$@'''
    SET @NVR_sqlCOM_00=@NVR_sqlCOM_00 + CHAR(13) + N'INSERT INTO ' + @NVR_TmpTableName + N' SELECT ' +@nvr_OnRows_B + N',(' + @NVR_Columns + N') AS [' +

@NVR_NewColName + N'],SUM([' + @nvr_OnValue + N']) AS '
            + @nvr_OnValue + N' FROM ' + @nvr_TableName
            + @nvr_Where
            + N' GROUP BY ' + @nvr_OnColumns_B   + N',' +  @nvr_OnRows_B
            + N' ORDER BY ' + @nvr_OnRows_B  + CHAR(13)

    --因为动态生成的sql语句很长,所以定义多个sql变量.
    DECLARE @NVR_XCOLNAME AS NVARCHAR(128)
    DECLARE @NVR_sql0 AS NVARCHAR(4000)
    DECLARE @NVR_sql1 AS NVARCHAR(4000)
    DECLARE @NVR_sql2 AS NVARCHAR(4000)
    DECLARE @NVR_sql3 AS NVARCHAR(4000)
    DECLARE @NVR_sql4 AS NVARCHAR(4000)
    DECLARE @NVR_sql5 AS NVARCHAR(4000)
    DECLARE @NVR_sql6 AS NVARCHAR(4000)
    DECLARE @NVR_sql7 AS NVARCHAR(4000)
    DECLARE @NVR_sql8 AS NVARCHAR(4000)
    DECLARE @NVR_sql9 AS NVARCHAR(4000)
    DECLARE @NVR_sql10 AS NVARCHAR(4000)
    DECLARE @NVR_sql11 AS NVARCHAR(4000)
    DECLARE @NVR_sql12 AS NVARCHAR(4000)
    DECLARE @NVR_sql13 AS NVARCHAR(4000)
    DECLARE @NVR_sql14 AS NVARCHAR(4000)
    DECLARE @NVR_sql15 AS NVARCHAR(4000)
    DECLARE @NVR_sql16 AS NVARCHAR(4000)
    DECLARE @NVR_sql17 AS NVARCHAR(4000)
    DECLARE @NVR_sql18 AS NVARCHAR(4000)
    DECLARE @NVR_sql19 AS NVARCHAR(4000)
    DECLARE @NVR_sql20 AS NVARCHAR(4000)
    DECLARE @INT_ID AS INT

    SET @NVR_XCOLNAME=''
    SELECT

@NVR_sql0='',@INT_ID=0,@NVR_sql0='',@NVR_sql1='',@NVR_sql2='',@NVR_sql3='',@NVR_sql4='',@NVR_sql5='',@NVR_sql6='',@NVR_sql7='',@NVR_sql8='',@NVR_sql9='',
       

@NVR_sql10='',@NVR_sql11='',@NVR_sql12='',@NVR_sql13='',@NVR_sql14='',@NVR_sql15='',@NVR_sql16='',@NVR_sql17='',@NVR_sql18='',@NVR_sql19='',@NVR_sql20=''
     
    DECLARE @INT_I INT
    DECLARE @NVR_FLDNAME NVARCHAR(4000)
   
    SET @NVR_sql0 = N' SELECT ' +  @nvr_OnRows_B
    SET @NVR_CMD=N'DECLARE CORSS_CURSOR CURSOR FOR SELECT disTINCT [' + @NVR_NewColName + N'] FROM (SELECT ' + @NVR_Columns + N' AS [' + @NVR_NewColName +

N'] FROM ' + @nvr_TableName + N' ' + @NVR_WHERE  + N' ) AS TIISHFURL ORDER BY [' + @NVR_NewColName + N'] FOR READ ONLY ' --生成游标
    EXECUTE (@NVR_CMD)
    OPEN CORSS_CURSOR
   
    WHILE (1=1)  BEGIN
        FETCH NEXT FROM CORSS_CURSOR INTO @NVR_XCOLNAME  --遍历游标，将列头信息放入变量@NVR_XCOLNAME
        IF (@@FETCH_STATUS <>0) BREAK
       
        SET @INT_ID=@INT_ID+1
        SET @NVR_XCOLNAME=REPLACE(@NVR_XCOLNAME,CHAR(39),CHAR(39)+CHAR(39))
       
        ------------------------------------------------------------------------------
        IF @INT_ID <=20
            SET @NVR_sql1 = @NVR_sql1 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF @INT_ID <=40
            SET @NVR_sql2 = @NVR_sql2 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=60
            SET @NVR_sql3 = @NVR_sql3 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=80
            SET @NVR_sql4 = @NVR_sql4 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=100
            SET @NVR_sql5 = @NVR_sql5 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=120
            SET @NVR_sql6 = @NVR_sql6 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=140
            SET @NVR_sql7 = @NVR_sql7 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=160
            SET @NVR_sql8 = @NVR_sql8 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=180
            SET @NVR_sql9 = @NVR_sql9 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END) AS

[' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=200
            SET @NVR_sql10 = @NVR_sql10 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=220
            SET @NVR_sql11 = @NVR_sql11 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=240
            SET @NVR_sql12 = @NVR_sql12 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=260
            SET @NVR_sql13 = @NVR_sql13 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=280
            SET @NVR_sql14 = @NVR_sql14 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=300
            SET @NVR_sql15 = @NVR_sql15 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=320
            SET @NVR_sql16 = @NVR_sql16 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=340
            SET @NVR_sql17 = @NVR_sql17 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=360
            SET @NVR_sql18 = @NVR_sql18 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=380
            SET @NVR_sql19 = @NVR_sql19 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
        ELSE IF  @INT_ID <=400
            SET @NVR_sql20 = @NVR_sql20 + N',SUM(CASE [' + @NVR_NewColName + N'] WHEN ''' + @NVR_XCOLNAME + N''' THEN [' + @nvr_OnValue + N'] ELSE NULL END)

AS [' + @NVR_XCOLNAME + N']' --构造查询
    END -- WHILE (1 = 1)
    Close CORSS_CURSOR
    DEALLOCATE CORSS_CURSOR

    EXECUTE ( @NVR_sqlCOM_00 + @NVR_sql0 + @NVR_sql1 + @NVR_sql2  + @NVR_sql3 + @NVR_sql4 + @NVR_sql5 + @NVR_sql6 + @NVR_sql7 + @NVR_sql8 + @NVR_sql9
     + @NVR_sql10 + @NVR_sql11 + @NVR_sql12  + @NVR_sql13 + @NVR_sql14 + @NVR_sql15 + @NVR_sql16 + @NVR_sql17 + @NVR_sql18 + @NVR_sql19 + @NVR_sql20
     + N' FROM ' + @NVR_TmpTableName + N' GROUP BY ' + @nvr_OnRows ) -- 执行
    RETURN 0 --释放游标，返回0表示成功

End

GO

 

来一张最终效果,否则你们没兴趣看了. 哈哈...

 

 

sqlSERVER2000 用过程来模拟MDX分析报表的效果(2)

 

本文来自CSDN博客，转载请标明出处：http://blog.csdn.net/mstop/archive/2008/10/11/3058767.aspx