编写带有多个CTE子查询SQL/R的单个查询

我想从数据库中提取一些数据，我正在使用RStudio进行查询。我想写的是：

第一个CTE声明获取我所有必要的信息

第二个CTE语句将为两个行号添加两个新列，这两个行号由不同的组划分。将为超前和滞后值添加两个附加列

第三个CTE将生成另外两列，其中两列使用嵌套的case_when语句来提供NewOpen和NewClosed日期

到目前为止，我所拥有的：

q5<- sqlQuery(ch,paste(" 
;with CTE AS
(
    select 
    oz.id as AccountID
    ,ac.PROD_TYPE_CDE as ProductTypeCode
    ,CASE WHEN ac.OPEN_DTE='0001-01-01' then null else ac.OPEN_DTE END as OpenDate
    ,CASE WHEN ac.CLOS_DTE = '0001-01-01' then null else ac.CLOS_DTE END as ClosedDate
    ,df.proc_dte as FullDate

   FROM
   dbs.tb_dbs_acct_fact df
   inner join
   dbs.tb_acct_details ac on df.dw_serv_id = ac.dw_serv_id
   left outer join
   dbs.tb_oz_id oz on df.proc_dte = oz.proc_dte 
),

cte1 as 
(
   select *
  ,row_nbr = row_number() over( partition by AccountID order by AccountID, FullDate asc )
  ,row_nbr2 = row_number() over( partition by AccountID,ProductTypeCode order by AccountID, FullDate asc )
  ,lag(ProductTypeCode) over(partition by AccountID order by FullDate asc ) as Lagging
  ,LEAD(ProductTypeCode) over(partition by AccountID order order by FullDate asc  ) as Leading
  FROM CTE   
),

cte2 as (select *
  ,case when cte1.row_nbr = 1 & cte1.Lagging=cte1.ProductTypeCode then cte1.OpenDate else
   case when cte1.Lagging<>cte1.ProductTypeCode then cte1.FullDate else NULL END END as NewOpen
  ,case when cte1.ClosedDate IS NOT NULL then cte1.ClosedDate else 
   case when cte1.Leading <> cte1.ProductTypeCode then cte1.FullDate else NULL END END as NewClosed
  FROM cte1
);"))

---

q5如前所述，
WITH
是定义最终查询中使用的CTE的语句。您的查询仅包含CTE定义，但从未在最终语句中实际使用任何定义。此外，由于窗口函数可以在任何级别上运行，因此可以将前两个CTE组合在一起。最后一个CTE可能可以作为您的最终
SELECT
语句

sql <- "WITH CTE AS
         (SELECT 
              oz.id AS AccountID
             , ac.PROD_TYPE_CDE as ProductTypeCode
             , CASE 
                    WHEN ac.OPEN_DTE='0001-01-01' 
                    THEN NULL 
                    ELSE ac.OPEN_DTE 
               END AS OpenDate
             , CASE 
                    WHEN ac.CLOS_DTE = '0001-01-01' 
                    THEN NULL 
                    ELSE ac.CLOS_DTE
               END AS ClosedDate
             , df.proc_dte AS FullDate

             , ROW_NUMBER() OVER (PARTITION BY oz.id 
                                  ORDER BY oz.id, df.proc_dte) AS row_nbr 
             , ROW_NUMBER() OVER (PARTITION BY oz.id, ac.PROD_TYPE_CDE 
                                  ORDER BY oz.id, df.proc_dte) AS row_nbr2
             , LAG(ac.PROD_TYPE_CDE) OVER (PARTITION BY oz.id 
                                           ORDER BY df.proc_dte) AS Lagging
             , LEAD(ac.PROD_TYPE_CDE) OVER (PARTITION BY oz.id
                                            ORDER BY df.proc_dte) AS Leading
          FROM
              dbs.tb_dbs_acct_fact df
          INNER JOIN
              dbs.tb_acct_details ac ON df.dw_serv_id = ac.dw_serv_id
          LEFT OUTER JOIN
              dbs.tb_oz_id oz ON df.proc_dte = oz.proc_dte 
         )

      SELECT *
          , CASE 
                WHEN row_nbr = 1 & Lagging = ProductTypeCode 
                THEN OpenDate 
                ELSE
                    CASE
                        WHEN Lagging <> ProductTypeCode 
                        THEN FullDate 
                        ELSE NULL 
                    END 
            END AS NewOpen
          , CASE 
                WHEN ClosedDate IS NOT NULL 
                THEN ClosedDate 
                ELSE 
                    CASE 
                        WHEN Leading <> ProductTypeCode 
                        THEN FullDate 
                        ELSE NULL 
                    END 
            END AS NewClosed
      FROM CTE;"

q5 <- sqlQuery(ch, sql)

sql如前所述，
WITH
是定义要在最终查询中使用的CTE的语句。您的查询仅包含CTE定义，但从未在最终语句中实际使用任何定义。此外，由于窗口函数可以在任何级别上运行，因此可以将前两个CTE组合在一起。最后一个CTE可能可以作为您的最终
SELECT
语句

sql <- "WITH CTE AS
         (SELECT 
              oz.id AS AccountID
             , ac.PROD_TYPE_CDE as ProductTypeCode
             , CASE 
                    WHEN ac.OPEN_DTE='0001-01-01' 
                    THEN NULL 
                    ELSE ac.OPEN_DTE 
               END AS OpenDate
             , CASE 
                    WHEN ac.CLOS_DTE = '0001-01-01' 
                    THEN NULL 
                    ELSE ac.CLOS_DTE
               END AS ClosedDate
             , df.proc_dte AS FullDate

             , ROW_NUMBER() OVER (PARTITION BY oz.id 
                                  ORDER BY oz.id, df.proc_dte) AS row_nbr 
             , ROW_NUMBER() OVER (PARTITION BY oz.id, ac.PROD_TYPE_CDE 
                                  ORDER BY oz.id, df.proc_dte) AS row_nbr2
             , LAG(ac.PROD_TYPE_CDE) OVER (PARTITION BY oz.id 
                                           ORDER BY df.proc_dte) AS Lagging
             , LEAD(ac.PROD_TYPE_CDE) OVER (PARTITION BY oz.id
                                            ORDER BY df.proc_dte) AS Leading
          FROM
              dbs.tb_dbs_acct_fact df
          INNER JOIN
              dbs.tb_acct_details ac ON df.dw_serv_id = ac.dw_serv_id
          LEFT OUTER JOIN
              dbs.tb_oz_id oz ON df.proc_dte = oz.proc_dte 
         )

      SELECT *
          , CASE 
                WHEN row_nbr = 1 & Lagging = ProductTypeCode 
                THEN OpenDate 
                ELSE
                    CASE
                        WHEN Lagging <> ProductTypeCode 
                        THEN FullDate 
                        ELSE NULL 
                    END 
            END AS NewOpen
          , CASE 
                WHEN ClosedDate IS NOT NULL 
                THEN ClosedDate 
                ELSE 
                    CASE 
                        WHEN Leading <> ProductTypeCode 
                        THEN FullDate 
                        ELSE NULL 
                    END 
            END AS NewClosed
      FROM CTE;"

q5 <- sqlQuery(ch, sql)

sql您从不从CTEs中选择
；声明不完整。谢谢你的评论。当你说我从不
从CTEs中选择
时，你能再详细说明一下吗？我想我没有跟踪你，我想我现在跟踪你了。你的意思是在我的代码末尾，我需要包括
SELECT*FROM CTE，cte1
@a_js12是的，你需要包括最后的SELECT。使用CTE的通常模式是
，cte1为（…），CTE2为（…），选择…
。您没有最终选择。您从未从CTEs中选择
；声明不完整。谢谢你的评论。当你说我从不
从CTEs中选择
时，你能再详细说明一下吗？我想我没有跟踪你，我想我现在跟踪你了。你的意思是在我的代码末尾，我需要包括
SELECT*FROM CTE，cte1
@a_js12是的，你需要包括最后的SELECT。使用CTE的通常模式是
，cte1为（…），CTE2为（…），选择…
。您没有最终选择。