Question

我有一个表（Table1），它具有从多个插入共享的ID：

ID  | RefID | Field_Name | Field_Value | Type
 1  |    1  | NumbAmt    | 1111        | INT
 2  |    1  | LocAdd     | 123 Street  | String
 3  |    1  | LocDesc    | Something   | String
 4  |    1  | LocHidden  | Useless     | Hidden

我无法使用ID，因为它是由插件制作的，RefID是用于将此数据缩小到所有那些 RefID = 1 AND Type！=＆＃39的主要内容;隐藏＆＃39;

每当我做一个案例陈述查询时：

SELECT 
  CASE WHEN Field_Name = 'NumbAmt' THEN Field_Value END Amt,
  CASE WHEN Field_Name = 'LocAdd' THEN Field_Value END Address,
  CASE WHEN Field_Name = 'LocDesc' THEN Field_Value END Description
FROM Table1
WHERE RefID = 1
  AND Type IN ('INT','String')

返回结果如：

    Amt     |   Address    | Description
1111        |    NULL      | NULL
NULL        | 123 Street   | NULL
NULL        |    NULL      | Something

我的问题是，我如何收集所有数据，但是如果没有显示所有的NULL，它会分成不同的列？（我的假设让我相信临时表）

或者显示如下：

 Amt     |   Address  | Description
1111     | 123 Street | Something

Answer 1

您有几种选择：

将表格加入自身
使用PIVOT
子查询SELECT列表中的所有字段
对每个字段使用OUTER APPLY
使用CTE
考虑重新考虑您的计划！

<强> JOINS

 SELECT
  T1_RefID.RefID, 
  T1_NumbAmt.FieldValue AS NumbAmt,
  T1_LocAdd.FieldValue AS LocAdd
FROM
  (SELECT DISTINCT RefID FROM Table1) T1_RefID
  LEFT JOIN Table1 T1_NumbAmt
    ON T1_RefID.RefID = T1_NumbAmt.RefID AND T1_NumbAmt.FieldName = 'NumbAmt' AND T1_NumbAmt.Type != 'Hidden'
  LEFT JOIN Table1 T1_LocAdd
    ON T1_RefID.RefID = T1_LocAdd.RefID AND T1_LocAdd.FieldName = 'LocAdd' AND T1_LocAdd.Type != 'Hidden'
  /* And so on*/

<强> PIVOT

SELECT
  *
FROM (
  SELECT
    RefID, FieldName, FieldValue
  FROM
    Table1
  WHERE
    Type != 'Hidden'
) AS src
PIVOT (
  MAX(FieldValue)
  FOR FieldName IN (NumbAmt, LocAdd, LocDesc)
) AS PVT

SQL Fiddle

Answer 2

你当然可以写下这句话：

WITH Record(Field_Value, Field_Name) AS (
  SELECT Field_Value, Field_Name FROM Table1 WHERE RefID = 1
)
SELECT 
  (SELECT Field_Value FROM Record WHERE Field_Name = 'NumbAmt') Amt,
  (SELECT Field_Value FROM Record WHERE Field_Name = 'LocAdd')  Address,
  (SELECT Field_Value FROM Record WHERE Field_Name = 'LocDesc') Description
FROM
  Record

但我强烈建议避免使用entity-attribute-value anti-pattern，并重新设计您的架构。

Answer 3

您可以为此创建一个视图并使用它

SELECT T.RefID, TAmt.Field_Value as Amt, TAddress.Field_Value as Address, TType.Field_Value as Type FROM
(
    SELECT
        DISTINCT RefID
    FROM
        Table1
    WHERE
        Type IN ('INT', 'String')
) T
LEFT JOIN
    Table1 AS TAmt
ON
    T.RefID = TAmt.RefID
    AND TAmt.Field_Name = 'NumbAmt'
LEFT JOIN
    Table1 AS TAddress
ON
    T.RefID = TAddress.RefID
    AND TAddress.Field_Name = 'LocAdd'
LEFT JOIN
    Table1 AS TType
ON
    T.RefID = TType.RefID
    AND TType.Field_Name = 'LocDesc'

Answer 4

在关键字段上使用group by，并使用一些聚合函数来提升空值的正确值（我相信max会做，但不能测试我现在坐的位置）

SELECT 
  RefID,
  MAX( CASE WHEN Field_Name = 'NumbAmt' THEN Field_Value ELSE NULL END ) Amt,
  MAX( CASE WHEN Field_Name = 'LocAdd' THEN Field_Value ELSE NULL END ) Address,
  MAX( CASE WHEN Field_Name = 'LocDesc' THEN Field_Value ELSE NULL END ) Description
FROM Table1
GROUP BY RefID

SQL：选择NOT NULL数据与共享和更改的列？

4 个答案: