Merge使用方法

oracle merge使用方法

oracle merge语法


merge into biz_module_apccode o
using (select * from temp_fa) n
on (o.biz_module_id=142 and o.apccode_id=n.apccode_id)
when matched then
update set o.order_by=n.order_by



/////////////////////////////////////////////////////
Oracle9i引入了MERGE命令,你能够在一个SQL语句中对一个表同时执行inserts和updates操作. MERGE命令从一个或多个数据源中选择行来updating或inserting到一个或多个表.在Oracle 10g中MERGE有如下一些改进：

1、UPDATE或INSERT子句是可选的

2、UPDATE和INSERT子句可以加WHERE子句

3、在ON条件中使用常量过滤谓词来insert所有的行到目标表中,不需要连接源表和目标表

4、UPDATE子句后面可以跟DELETE子句来去除一些不需要的行

首先创建示例表：
[pre]create table PRODUCTS
  (
  PRODUCT_ID INTEGER,
  PRODUCT_NAME VARCHAR2(60),
  CATEGORY VARCHAR2(60)
  );

  insert into PRODUCTS values (1501, 'VIVITAR 35MM', 'ELECTRNCS');
  insert into PRODUCTS values (1502, 'OLYMPUS IS50', 'ELECTRNCS');
  insert into PRODUCTS values (1600, 'PLAY GYM', 'TOYS');
  insert into PRODUCTS values (1601, 'LAMAZE', 'TOYS');
  insert into PRODUCTS values (1666, 'HARRY POTTER', 'DVD');
  commit;

  create table NEWPRODUCTS
  (
  PRODUCT_ID INTEGER,
  PRODUCT_NAME VARCHAR2(60),
  CATEGORY VARCHAR2(60)
  );

  insert into NEWPRODUCTS values (1502, 'OLYMPUS CAMERA', 'ELECTRNCS');
  insert into NEWPRODUCTS values (1601, 'LAMAZE', 'TOYS');
  insert into NEWPRODUCTS values (1666, 'HARRY POTTER', 'TOYS');
  insert into NEWPRODUCTS values (1700, 'WAIT INTERFACE', 'BOOKS');
  commit;[/pre]
1、可省略的UPDATE或INSERT子句

在Oracle 9i,MERGE语句要求你必须同时指定INSERT和UPDATE子句.而在Oracle 10g, 你可以省略UPDATE或INSERT子句中的一个.下面的例子根据表NEWPRODUCTS的PRODUCT_ID字段是否匹配来updates表PRODUCTS的信息:


[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (p.product_id = np.product_id)
  4 WHEN MATCHED THEN
  5 UPDATE
  6 SET p.product_name = np.product_name,
  7 p.category = np.category;

  3 rows merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1502 OLYMPUS CAMERA ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER TOYS
  SQL>
  SQL> ROLLBACK;
  Rollback complete.
  SQL>[/pre]



在上面例子中, MERGE语句影响到是产品id为1502,1601和1666的行. 它们的产品名字和种 类被更新为表newproducts中的值. 下面例子省略UPDATE子句,把表NEWPRODUCTS中新的PRODUCT_ID插入到表PRODUCTS中,对于在两个表中能够匹配上PRODUCT_ID的数据不作任何处理.从这个例子你能看到PRODUCT_ID=1700的行被插入到表PRODUCTS中.

[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (p.product_id = np.product_id)
  4 WHEN NOT MATCHED THEN
  5 INSERT
  6 VALUES (np.product_id, np.product_name,
  7 np.category);

  1 row merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1502 OLYMPUS IS50 ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER DVD
  1700 WAIT INTERFACE BOOKS[/pre]
2、带条件的Updates和Inserts子句

你能够添加WHERE子句到UPDATE或INSERT子句中去, 来跳过update或insert操作对某些行的处理. 下面例子根据表NEWPRODUCTS来更新表PRODUCTS数据, 但必须字段CATEGORY也得同时匹配上:

[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (p.product_id = np.product_id)
  4 WHEN MATCHED THEN
  5 UPDATE
  6 SET p.product_name = np.product_name
  7 WHERE p.category = np.category;

  2 rows merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1502 OLYMPUS CAMERA ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER DVD
  SQL>
  SQL> rollback;[/pre]



在这个例子中, 产品ID为1502,1601和1666匹配ON条件但是1666的category不匹配. 因此MERGE命令只更新两行数据. 下面例子展示了在Updates和Inserts子句都使用WHERE子句:

[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (p.product_id = np.product_id)
  4 WHEN MATCHED THEN
  5 UPDATE
  6 SET p.product_name = np.product_name,
  7 p.category = np.category
  8 WHERE p.category = 'DVD'
  9 WHEN NOT MATCHED THEN
  10 INSERT
  11 VALUES (np.product_id, np.product_name, np.category)
  12 WHERE np.category != 'BOOKS'
  SQL> /

  1 row merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1502 OLYMPUS IS50 ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER TOYS

  SQL>[/pre]


注意由于有WHERE子句INSERT没有插入所有不匹配ON条件的行到表PRODUCTS.
3、无条件的Inserts

你能够不用连接源表和目标表就把源表的数据插入到目标表中.这对于你想插入所有行到目标表时是非常有用的. Oracle 10g现在支持在ON条件中使用常量过滤谓词. 举个常量过滤谓词例子ON(1=0). 下面例子从源表插入行到表PRODUCTS, 不检查这些行是否在表PRODUCTS中存在:








[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (1=0)
  4 WHEN NOT MATCHED THEN
  5 INSERT
  6 VALUES (np.product_id, np.product_name, np.category)
  7 WHERE np.category = 'BOOKS'
  SQL> /

  1 row merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1502 OLYMPUS IS50 ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER DVD
  1700 WAIT INTERFACE BOOKS
  6 rows selected.
  SQL>[/pre]



4、新增加的DELETE子句

Oracle 10g中的MERGE提供了在执行数据操作时清除行的选项.你能够在WHEN MATCHED THEN UPDATE子句中包含DELETE子句.DELETE子句必须有一个WHERE条件来删除匹配某些条件的行.匹配DELETE WHERE条件但不匹配ON条件的行不会被从表中删除.

下面例子验证DELETE子句. 我们从表NEWPRODUCTS中合并行到表PRODUCTS中, 但删除category为ELECTRNCS的行.

[pre]SQL> MERGE INTO products p
  2 USING newproducts np
  3 ON (p.product_id = np.product_id)
  4 WHEN MATCHED THEN
  5 UPDATE
  6 SET p.product_name = np.product_name,
  7 p.category = np.category
  8 DELETE WHERE (p.category = 'ELECTRNCS')
  9 WHEN NOT MATCHED THEN
  10 INSERT
  11 VALUES (np.product_id, np.product_name, np.category)
  SQL> /

  4 rows merged.

  SQL> SELECT * FROM products;

  PRODUCT_ID PRODUCT_NAME CATEGORY
  ---------- -------------------- ----------
  1501 VIVITAR 35MM ELECTRNCS
  1600 PLAY GYM TOYS
  1601 LAMAZE TOYS
  1666 HARRY POTTER TOYS
  1700 WAIT INTERFACE BOOKS
  SQL>[/pre]



产品ID为1502的行从表PRODUCTS中被删除,因为它同时匹配ON条件和DELETE WHERE条件. 产品ID为1501的行匹配DELETE WHERE条件但不匹配ON条件,所以它没有被删除. 产品ID为1700 的行不匹配ON条件, 所以被插入表PRODUCTS.产品ID为1601和1666的行匹配ON条件但不匹配DELETE WHERE条件, 所以被更新为表NEWPRODUCTS中的值.

自己的例子：
MERGE INTO TBL_REALTIME_AD_USER_RPT RAUP
    USING (SELECT T1.USER_ID,
                  T2.MANAGER_ID,
                  T1.ROTATE_AD_ID,
                  TO_CHAR(T1.VISIT_DATE, 'YYYYMMDD') REPORT_DATE,
                  COUNT(*) CV,
                  COUNT(DISTINCT COOKIE_ID) UV
             FROM TBL_REALTIME_CLICK_LOG T1, TBL_USER T2
            WHERE T1.USER_ID = T2.ID
              AND TO_CHAR(T1.VISIT_DATE, 'YYYYMMDD') = V_REPORT_DATE
            GROUP BY T1.USER_ID,
                     T2.MANAGER_ID,
                     T1.ROTATE_AD_ID,
                     TO_CHAR(T1.VISIT_DATE, 'YYYYMMDD')) S
    ON (RAUP.USER_ID = S.USER_ID AND RAUP.AD_ID = S.ROTATE_AD_ID AND RAUP.REPORT_DATE = S.REPORT_DATE)
    WHEN MATCHED THEN
      UPDATE SET RAUP.CLICK = S.CV, RAUP.UV = S.UV
    WHEN NOT MATCHED THEN
      INSERT
        (ID, USER_ID, MANAGER_ID, AD_ID, REPORT_DATE, CLICK, UV)
      VALUES
        (SEQ_REALTIME_AD_USER_RPT_ID.NEXTVAL,
         S.USER_ID,
         S.MANAGER_ID,
         S.ROTATE_AD_ID,
         S.REPORT_DATE,
         S.CV,
         S.UV);