TSQL - 扩展由交叉应用和数据透视组成的查询

Question

TSQL - 扩展由交叉应用和数据透视组成的查询

Ton*_*ton 5 sql t-sql sql-server pivot cross-apply

这个问题是基于我以前的问题.我需要扩展查询,以便我可以合并另外两个表(在另一个服务器实例上运行).

在这个小提琴我添加了这两个表:

CREATE TABLE LookUp
 ([docID] varchar(10), [docType] varchar(100), [PartNumber] varchar(100), [internalID] varchar(100));
INSERT INTO LookUp
 ([docID],[docType],[PartNumber], [internalID])
VALUES
  ('D0305415', 'docTypeSub', 'X0455', null),
  ('D0157632', 'docTypeMain', null, 'XY05570-XY05571'),
  ('D0181511', 'docTypeMain',null, 'XY05572-XY05573'),
  ('D0157633', 'docTypeMain', null, 'XY06380-XY06381'),
  ('D0156037', 'docTypeSub', 'X0326', null),
  ('D0151874', 'docTypeMain', null, 'XY05345');

CREATE TABLE Links
  ([docIDTop] varchar(10), [docIDBottom] varchar(10));
INSERT INTO Links
  ([docIDTop],[docIDBottom])
VALUES
  ('D0157632', 'D0305415'),
  ('D0181511', 'D0305415'),
  ('D0157633', 'D0305415'),
  ('D0151874', 'D0156037');

Run Code Online (Sandbox Code Playgroud)

关于输出,我需要根据PartNumber列在逗号分隔列中显示新的internalID列.

这是输出正确数据的查询:

 select c.docType AS c_docTypeSub, c.docID AS C_docID, c.PartNumber AS C_PartNumber , 
 b.docIDTop AS B_docIdTop, b.docIDBottom AS B_docIdBottom,  a.* 
 FROM LookUp a, Links b, LookUp c
 WHERE a.docType = 'docTypeMain' 
 and a.docID = b.docIDTop and b.docIDBottom = c.docID 
 and c.docType = 'docTypeSub'
 ;

Run Code Online (Sandbox Code Playgroud)

我的问题是将这些部分放在一起,以便我可以在下面的旧查询中显示InternalID:

----------------
-- OLD Query -- 
----------------
WITH CTE_no_nums
AS
(
SELECT  docID,
        CASE
            WHEN PATINDEX('%[0-9]%',column1) > 0
                THEN SUBSTRING(column1,0,PATINDEX('%[0-9]%',column1))
            ELSE column1
        END AS cols,
        COALESCE(column2,column3) AS vals
FROM miscValues
WHERE       column2 IS NOT NULL
        OR  column3 IS NOT NULL
),
CTE_Pivot
    AS
    (
    SELECT docID,partNumber,prio,[length],material
    FROM CTE_no_nums
    PIVOT
    (
        MAX(vals) FOR cols IN (partNumber,prio,[length],material)
    ) pvt
)

SELECT  A.docId + ' # ' + B.vals AS [DocID # Plant],
        A.docID,
        A.partNumber,
        A.prio,
        B.vals AS Plant,
        A.partNumber + '#' + A.material + '#' + A.[length] AS Identification,
        A.[length],
        SUBSTRING(CA.colors,0,LEN(CA.colors)) colors --substring removes last comma
FROM CTE_Pivot A
INNER JOIN CTE_no_nums B
    ON      A.docID = B.docID
        AND B.cols = 'Plant'
CROSS APPLY (   SELECT vals + ',' 
                FROM CTE_no_nums C 
                WHERE   cols = 'Color' 
                    AND C.docID = A.docID 
                FOR XML PATH('') 
            ) CA(colors)
            ;

Run Code Online (Sandbox Code Playgroud)

希望你能告诉我这是如何实现的.如果有什么不清楚的话随便问.不,我不负责数据结构:-)

谢谢.

Answer 1

Shn*_*ugo 2

胡安·鲁伊斯·德卡斯蒂利亚扩展了我给出的答案，并让我看到了你的“链接”表。

这是我的最后建议，通过另一个 CTE 解决您的问题：

CREATE TABLE MiscValues
    ([docID] varchar(10) ,[rowNumber] int,  [Column1] varchar(100), [Column2] varchar(100)
     , [Column3] varchar(100))
;
INSERT INTO MiscValues
    ([docID],[rowNumber],[Column1], [Column2], [Column3])
VALUES
    ('D0001',1, 'PartNumber', 'X0455', NULL),
    ('D0001',2, 'Prio', '1', NULL),
    ('D0001',3, 'Plant1', NULL, NULL),
    ('D0001',4, 'Plant2', 'PlantB', NULL),
    ('D0001',5, 'Plant3', 'PlantC', NULL),
    ('D0001',6, 'Plant4',  NULL, NULL),
    ('D0001',7, 'Color1', 'white', NULL),
    ('D0001',8, 'Color2', 'black', NULL),
    ('D0001',9, 'Color3', 'blue', NULL),
    ('D0001',10, 'Material', 'MA123', NULL),
    ('D0001',11, 'Length',  NULL, '10.87'),

    ('D0002',1, 'PartNumber', 'X0326', NULL),
    ('D0002',2, 'Prio', '2', NULL),
    ('D0002',3, 'Plant1', 'PlantA', NULL),
    ('D0002',4, 'Plant2', NULL, NULL),
    ('D0002',5, 'Plant3', 'PlantC', NULL),
    ('D0002',6, 'Plant4', 'PlantD', NULL),
    ('D0002',7, 'Color1', NULL, NULL),
    ('D0002',8, 'Color2', 'black', NULL),
    ('D0002',9, 'Color3', NULL, NULL),
    ('D0002',10, 'Color4', 'yellow', NULL),
    ('D0002',11, 'Material', 'MA456', NULL),
    ('D0002',12, 'Length', NULL, '16.43')
;

CREATE TABLE LookUp([docID] varchar(10), [docType] varchar(100), [PartNumber] varchar(100), [internalID] varchar(100));
INSERT INTO LookUp([docID],[docType],[PartNumber], [internalID])
VALUES
  ('D0305415', 'docTypeSub', 'X0455', null),
  ('D0157632', 'docTypeMain', null, 'XY05570-XY05571'),
  ('D0181511', 'docTypeMain',null, 'XY05572-XY05573'),
  ('D0157633', 'docTypeMain', null, 'XY06380-XY06381'),
  ('D0156037', 'docTypeSub', 'X0326', null),
  ('D0151874', 'docTypeMain', null, 'XY05345');

CREATE TABLE Links ([docIDTop] varchar(10), [docIDBottom] varchar(10));
INSERT INTO Links ([docIDTop],[docIDBottom])
VALUES
  ('D0157632', 'D0305415'),
  ('D0181511', 'D0305415'),
  ('D0157633', 'D0305415'),
  ('D0151874', 'D0156037');

WITH CTE_no_nums
AS
(
SELECT  docID,
        CASE
            WHEN PATINDEX('%[0-9]%',column1) > 0
                THEN SUBSTRING(column1,0,PATINDEX('%[0-9]%',column1))
            ELSE column1
        END AS cols,
        COALESCE(column2,column3) AS vals
FROM miscValues
WHERE       column2 IS NOT NULL
        OR  column3 IS NOT NULL
),
CTE_Pivot
    AS
    (
    SELECT docID,partNumber,prio,[length],material
    FROM CTE_no_nums
    PIVOT
    (
        MAX(vals) FOR cols IN (partNumber,prio,[length],material)
    ) pvt
),
CTE_InternalIDs AS
(
      SELECT *
      ,STUFF
      ( 
        (SELECT ', ' + internalID
         FROM LookUp AS L2
         INNER JOIN Links L ON L2.docID=L.docIDTop
         WHERE L2.internalID IS NOT NULL 
           AND L.docIDBottom=L1.docID
         FOR XML PATH('')
      ),1,2,'') AS ConcatenatedInternalIDs
FROM LookUp AS L1
WHERE L1.internalID IS NULL
)
SELECT  A.docId + ' # ' + B.vals AS [DocID # Plant],
        A.docID,
        A.partNumber,
        A.prio,
        B.vals AS Plant,
        A.partNumber + '#' + A.material + '#' + A.[length] AS Identification,
        A.[length],
        SUBSTRING(CA.colors,0,LEN(CA.colors)) colors, --substring removes last comma
        IIDs.ConcatenatedInternalIDs
FROM CTE_Pivot A
INNER JOIN CTE_no_nums B
    ON      A.docID = B.docID
        AND B.cols = 'Plant'
INNER JOIN CTE_InternalIDs AS IIDs ON A.partNumber = IIDs.PartNumber
CROSS APPLY (   SELECT vals + ',' 
                FROM CTE_no_nums C 
                WHERE   cols = 'Color' 
                    AND C.docID = A.docID 
                FOR XML PATH('') 
            ) CA(colors)
            ;
--Clean up...
/*
DROP TABLE Links;
DROP TABLE LookUp;
DROP TABLE miscValues;
*/

Run Code Online (Sandbox Code Playgroud)

上面的结果是这样的：

D0001 # PlantB  D0001   X0455   1   PlantB  X0455#MA123#10.87   10.87   white,black,blue    XY05570-XY05571, XY05572-XY05573, XY06380-XY06381
D0001 # PlantC  D0001   X0455   1   PlantC  X0455#MA123#10.87   10.87   white,black,blue    XY05570-XY05571, XY05572-XY05573, XY06380-XY06381
D0002 # PlantA  D0002   X0326   2   PlantA  X0326#MA456#16.43   16.43   black,yellow    XY05345
D0002 # PlantC  D0002   X0326   2   PlantC  X0326#MA456#16.43   16.43   black,yellow    XY05345
D0002 # PlantD  D0002   X0326   2   PlantD  X0326#MA456#16.43   16.43   black,yellow    XY05345

Run Code Online (Sandbox Code Playgroud)

编辑：从这里开始，您会找到我的第一个答案（用于理解胡安·鲁伊斯的答案）：

我不确定，如果我理解正确的话...您想要将一个串联列表添加到您的查询中，其中所有内部ID条目都适合LookUp的PartNumber。

您遇到的问题：没有隐式排序顺序......

您插入的

    VALUES
  ('D0305415', 'docTypeSub', 'X0455', null),
  ('D0157632', 'docTypeMain', null, 'XY05570-XY05571'),
  ('D0181511', 'docTypeMain',null, 'XY05572-XY05573'),
  ('D0157633', 'docTypeMain', null, 'XY06380-XY06381'),
  ('D0156037', 'docTypeSub', 'X0326', null),
  ('D0151874', 'docTypeMain', null, 'XY05345');

Run Code Online (Sandbox Code Playgroud)

似乎将值“XY05570-XY05571”、“XY05572-XY05573”和“XY06380-XY06381”“绑定”到零件号“X0455”，将值“XY05345”“绑定”到零件号“X0326”。但这是错误的！！！

你可以这样做

VALUES
  ('D0305415', 'docTypeSub', 'X0455', null),
  ('D0157632', 'docTypeMain', 'X0455', 'XY05570-XY05571'),
  ('D0181511', 'docTypeMain','X0455', 'XY05572-XY05573'),
  ('D0157633', 'docTypeMain', 'X0455', 'XY06380-XY06381'),
  ('D0156037', 'docTypeSub', 'X0326', null),
  ('D0151874', 'docTypeMain', 'X0326', 'XY05345');

Run Code Online (Sandbox Code Playgroud)

或者，您可以添加一个 IDENTITY 列，并修改那些内部 ID 为 NULL 的条目之间的所有条目。

使用第一个（填充每行的 PartNumber 列），您可以获得如下的串联列表：

    select c.docType AS c_docTypeSub, c.docID AS C_docID, c.PartNumber AS C_PartNumber , 
 b.docIDTop AS B_docIdTop, b.docIDBottom AS B_docIdBottom,  a.*, 
 STUFF((SELECT ', ' + x.internalID 
         FROM LookUp AS x 
         WHERE x.PartNumber=c.PartNumber 
         FOR XML PATH(''),TYPE).value('.','nvarchar(max)'),1,2,'') AS ConcatInternalID
 FROM LookUp a, Links b, LookUp c
 WHERE a.docType = 'docTypeMain' 
 and a.docID = b.docIDTop and b.docIDBottom = c.docID 
 and c.docType = 'docTypeSub'
 ;

Run Code Online (Sandbox Code Playgroud)

当然，您也可以将其添加到“旧查询”中：

只需将其添加到最终的 SELECT 中即可

[...CTEs before...]
SELECT  A.docId + ' # ' + B.vals AS [DocID # Plant],
        A.docID,
        A.partNumber,
        A.prio,
        B.vals AS Plant,
        A.partNumber + '#' + A.material + '#' + A.[length] AS Identification,
        A.[length],
        SUBSTRING(CA.colors,0,LEN(CA.colors)) colors, --substring removes last comma
        STUFF((SELECT ', ' + x.internalID 
         FROM LookUp AS x 
         WHERE x.PartNumber=A.PartNumber 
         FOR XML PATH(''),TYPE).value('.','nvarchar(max)'),1,2,'') AS ConcatInternalID
[...FROM...]

Run Code Online (Sandbox Code Playgroud)

希望我能很好地理解这一点，这可以帮助你......

归档时间：	10 年，1 月前
查看次数：	241 次
最近记录：	10 年，1 月前