简体   繁体   English

使用 SQL Server 中的 CTAS 将数据从具有不同列名的源映射到现有 SQL 表

[英]Map data with CTAS in SQL Server from source with different column names to existing SQL table

Scenario:设想:

Is source table SourceData (Name, Number, Date)是源表SourceData (Name, Number, Date)

Is existing table ProdData (ProdName, ProdNumber, CreatedDate)是否存在表ProdData (ProdName、ProdNumber、CreatedDate)

Requirement:要求:

Dont import from source if already exists in prod data!!!如果生产数据中已经存在,请不要从源导入!!!

Import rows from source to prod data, keep existing rows, append new ones, map columns like:将行从源导入到生产数据,保留现有行,附加新行,映射列,如:

  • Name -> ProdName名称 -> 产品名称
  • Number -> ProdNumber编号 -> 产品编号
  • Date -> CreatedDate ( IF Date NULL add SystemDate time )日期 -> CreatedDate ( IF Date NULL add SystemDate time )

Output data example:输出数据示例:

SourceData源数据

Name | Number | Date
-----+--------+------ 
A    | 1      | 2012
B    | 2      | NULL

ProdData生产数据

ProdName | ProdNumber | CreatedDate
---------+------------+------------
Existing |    123     | 2018
A        |    1       | 2012    
B        |    2       | 2020

Would something like this fit your needs?这样的东西会满足您的需求吗?

INSERT ProdData (ProdName, ProdNumber, CreatedDate)
SELECT 
    SourceData.name as "ProdName",
    SourceData.Number as "ProdNumber",
    NVL(SourceData.Date,to_date(sysdate,'YYYY') as "CreatedDate"
FROM 
    SourceData
WHERE 
    NOT EXISTS (SELECT
                    1 
                FROM 
                    ProdData PD2 
                WHERE 
                    PD2.ProdName = SourceData.name
                    and PD2.ProdNumber = SourceData.Number
                    and NVL(PD2.CreatedDate,to_date(sysdate,'YYYY') = nvl(SourceData.Date,to_date(sysdate,'YYYY');

Another alternate option may be something like:另一个替代选项可能类似于:

CREATE TABLE ProdData_new
WITH
    (
      DISTRIBUTION = HASH(ProdName)
    , CLUSTERED INDEX (ProdNumber)
    )
AS
SELECT 
      pdold.ProdName AS ProdName
    , pdold.ProdNumber AS ProdNumber
    , NVL(pdold.CreatedDate,to_date(sysdate,'YYYY')) AS CreatedDate
FROM      
    ProdData AS pdold

UNION 

SELECT 
      sdold.name AS ProdName
    , sdold.Number AS ProdNumber
    , NVL(sdold.Date,to_date(sysdate,'YYYY') AS CreatedDate
FROM 
    SourceData AS sdold
;

--Optional Renaming of old tables
--RENAME OBJECT ProdData TO ProdData_old;
--RENAME OBJECT SourceData TO SourceData_old;
--RENAME OBJECT ProdData_new TO ProdData;
--DROP TABLE ProdData_old;
--DROP TABLE SourceData_old;

Or possibly something more like:或者可能更像:

CREATE TABLE ProdData_new
    ( ProdName NVARCHAR NOT NULL 
    , ProdNumber INT NOT NULL
    , CreatedDate INT NOT NULL)
WITH
    ( DISTRIBUTION = HASH(ProdName)
    , CLUSTERED INDEX (ProdNumber) )
AS
SELECT 
      pdold.ProdName AS ProdName
    , pdold.ProdNumber AS ProdNumber
    , NVL(pdold.CreatedDate,to_date(sysdate,'YYYY')) AS CreatedDate
FROM      
    ProdData AS pdold

UNION 

SELECT 
      sdold.name AS ProdName
    , sdold.Number AS ProdNumber
    , NVL(sdold.Date,to_date(sysdate,'YYYY') AS CreatedDate
FROM 
    SourceData AS sdold
;

--Optional Renaming of old tables
--RENAME OBJECT ProdData TO ProdData_old;
--RENAME OBJECT SourceData TO SourceData_old;
--RENAME OBJECT ProdData_new TO ProdData;
--DROP TABLE ProdData_old;
--DROP TABLE SourceData_old;

This is all somewhat guess work without knowing what you have tried, your specific syntax and your DB_Schema_Table setup.在不知道您尝试过什么、您的特定语法和您的 DB_Schema_Table 设置的情况下,这都是一些猜测工作。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM