如何在 Oracle 10 中连接两个表的结果

Posted 2023-05-09

【中文标题】如何在 Oracle 10 中连接两个表的结果

【英文标题】：How to join results from two tables in Oracle 10

【发布时间】：2009-10-13 12:50:15

【问题描述】：

假设我有 2 个结构相同的表：STOCK 和 NEW_STOCK。 这些表的主键由 (ID_DATE, ID_SELLER, ID_INVOICE, ID_DOC) 组成。

现在，我需要为每个（ID_DATE、ID_SELLER、ID_INVOICE、ID_DOC）​​获取与此要求相关的金额（字段 AMOUNT）的值：

如果NEW_STOCK 中存在记录，我从NEW_STOCK 获取AMOUNT，否则，我从STOCK 表获取AMOUNT。

请注意，ID_DATE 和 ID_SELLER 是提供给查询的输入，即仅考虑 STOCK 表的查询将如下所示：

select AMOUNT, ID_DATE, ID_SELLER, ID_INVOICE
    from STOCK
    where ID_DATE = 1
      and ID_SELLER = 'SELL1';

STOCK：

+---------+-----------+------------+--------+--------+
| ID_DATE | ID_SELLER | ID_INVOICE | ID_DOC | AMOUNT |
+---------+-----------+------------+--------+--------+
|       1 |     SELL1 |        IN1 |   DOC1 |    100 |
|       1 |     SELL1 |        IN2 |   DOC2 |     50 |
|       1 |     SELL1 |        IN3 |   DOC3 |     42 |
+---------+-----------+------------+--------+--------+

NEW_STOCK:

+---------+-----------+------------+--------+--------+
| ID_DATE | ID_SELLER | ID_INVOICE | ID_DOC | AMOUNT |
+---------+-----------+------------+--------+--------+
|       1 |     SELL1 |        IN2 |   DOC2 |     12 |
+---------+-----------+------------+--------+--------+

那么，我必须得到以下结果：

+---------+-----------+------------+--------+--------+
| ID_DATE | ID_SELLER | ID_INVOICE | ID_DOC | AMOUNT |
+---------+-----------+------------+--------+--------+
|       1 |     SELL1 |        IN1 |   DOC1 |    100 |
|       1 |     SELL2 |        IN2 |   DOC2 |     12 |
|       1 |     SELL3 |        IN3 |   DOC3 |     42 |
+---------+-----------+------------+--------+--------+

ps：我正在开发 Oracle 10。

【参考方案1】：

使用外连接和 NVL(arg1, arg2) 函数。 如果它不为 NULL，则返回第一个参数，否则返回第二个参数。示例：

select s.AMOUNT, s.ID_DATE, s.ID_SELLER, s.ID_INVOICE,
   NVL(n.AMOUNT, s.AMOUNT) amount       
from STOCK s, NEW_STOCK n
where s.ID_DATE = n.ID_DATE(+) 
  and s.ID_SELLER = n.ID_SELLER(+)
  and s.ID_INVOICE = n.ID_INVOICE(+)
  and s.ID_DOC = n.ID_DOC(+)
  and s.ID_DATE = 1
  and s.ID_SELLER = 'SELL1';

如果您觉得(+) 更易读，您可以使用LEFT OUTER JOIN 语法代替(+)。我从 v7 开始使用 Oracle，我更喜欢 (+)。 这是LEFT OUTER JOIN 语法：

select s.AMOUNT, s.ID_DATE, s.ID_SELLER, s.ID_INVOICE,
   NVL(n.AMOUNT, s.AMOUNT) amount       
from  STOCK s left outer join NEW_STOCK n 
      on s.ID_DATE = n.ID_DATE
         and s.ID_SELLER = n.ID_SELLER
         and s.ID_INVOICE = n.ID_INVOICE
         and s.ID_DOC = n.ID_DOC
where s.ID_DATE = 1
  and s.ID_SELLER = 'SELL1';

【讨论】：

谢谢，这是我尝试过的，只是我在 where 部分添加了“n.ID_DATE=1 and n.ID_SELLER = 'SELL1'”，这不适用于仅存在于 STOCK 表中的记录.顺便问一下，当有多个字段用于连接时，如何编写“左外连接”？

不，“左外连接”没问题。谢谢。

【参考方案2】：

SELECT * FROM (
    SELECT * FROM new_stock
    UNION ALL
    SELECT * FROM stock
    WHERE (ID_DATE,ID_SELLER,ID_INVOICE,ID_DOC) NOT IN 
     (SELECT ID_DATE,ID_SELLER,ID_INVOICE,ID_DOC FROM new_stock)
)
WHERE ID_DATE = 1 
   AND ID_SELLER = 'SELL1';

【参考方案3】：

以下应该适用：

SELECT s.AMOUNT, s.ID_DATE, s.ID_SELLER, s.ID_INVOICE    
FROM STOCK s
LEFT JOIN NEW_STOCK ns 
   ON s.ID_DATE = ns.ID_DATE 
   AND s.ID_SELLER = ns.ID_SELLER 
   AND s.ID_INVOICE = ns.ID_INVOICE
WHERE s.ID_DATE = 1      
  AND s.ID_SELLER = 'SELL1'
  AND ns.ID_DATE IS NULL
UNION
SELECT AMOUNT, ID_DATE, ID_SELLER, ID_INVOICE    
FROM NEW_STOCK    
WHERE ID_DATE = 1
  AND ID_SELLER = 'SELL1';

从使用 NEW_STOCK 表结果的 LEFT JOIN 和 UNION 中排除匹配的行。

【参考方案4】：

SELECT COALESCE(NS.AMOUNT, S.AMOUNT) AMOUNT, 
       S.ID_DATE, 
       S.ID_SELLER, 
       S.ID_INVOICE
  FROM STOCK S 
  LEFT JOIN NEW_STOCK NS ON S.ID_DATE = NS.ID_DATE
                        AND S.ID_SELLER = NS.ID_SELLER
                        AND S.ID_INVOICE = NS.ID_INVOICE
                        AND S.ID_DOC = NS.ID_DOC 
 WHERE S.ID_DATE = 1
   AND S.ID_SELLER = 'SELL1'