JOIN 表查询给出奇怪的结果

Question

表 1 - DBT_C_INT_ACCR

select INT_ACC.BRCH_CODE, INT_ACC.PROD_CODE, INT_ACC.REF_ID from DBT_C_INT_ACCR INT_ACC where BRCH_CODE = 784 and PROD_CODE = 'GF' and REF_ID = 'GFE1200077';

BRCH_CODE   PROD_CODE   REF_ID  
784             GF      GFE1200077
784             GF      GFE1200077
784             GF      GFE1200077
784             GF      GFE1200077

---

表 2 - dbt_c_rate_refresh

select INT_ACC.BRCH_CODE, INT_ACC.PROD_CODE, INT_ACC.REF_ID from dbt_c_rate_refresh INT_ACC where BRCH_CODE = 784 and PROD_CODE = 'GF' and REF_ID = 'GFE1200077';


BRCH_CODE   PROD_CODE   REF_ID  
784         GF          GFE1200077
784         GF          GFE1200077
784         GF          GFE1200077
784         GF          GFE1200077
784         GF          GFE1200077
784         GF          GFE1200077
784         GF          GFE1200077

---

现在，当尝试使用下面的 Join 时，它会给出 28 行而不是 4 行。有什么解释吗？

select INT_ACC.ref_id, RATE_REFRESH.ref_id from DBT_C_INT_ACCR INT_ACC left OUTER JOIN dbt_c_rate_refresh RATE_REFRESH ON (INT_ACC.BRCH_CODE = RATE_REFRESH.BRCH_CODE and INT_ACC.PROD_CODE = RATE_REFRESH.PROD_CODE and INT_ACC.REF_ID = RATE_REFRESH.REF_ID)
where INT_ACC.BRCH_CODE = '784'  and INT_ACC.PROD_CODE = 'GF' and INT_ACC.REF_ID = 'GFE1200077';

REF_ID      REF_ID_1
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077
GFE1200077  GFE1200077

Answer 1

因为您没有加入独特的事物，所以您实际上是在两个表中具有重复连接条件值的行之间进行笛卡尔连接（也称为交叉连接）。

为了演示，我模仿了您的表格并添加了一个唯一标识符 rn，以便您可以看到执行连接时会发生什么：

with dbt_c_int_accr as (select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 1 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 2 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 3 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 4 rn from dual),
 dbt_c_rate_refresh as (select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 1 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 2 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 3 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 4 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 5 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 6 rn from dual union all
                        select 784 brch_code, 'GF' prod_code, 'GFE1200077' ref_id, 7 rn from dual)
-- end of mimicking your tables with data in. You wouldn't need the above subqueries, as you have the tables themselves.
-- See the SQL below:
select int_acc.ref_id,
       rate_refresh.ref_id,
       int_acc.rn,
       rate_refresh.rn
from   dbt_c_int_accr int_acc
       left outer join dbt_c_rate_refresh rate_refresh on (int_acc.brch_code = rate_refresh.brch_code
                                                           and int_acc.prod_code = rate_refresh.prod_code
                                                           and int_acc.ref_id = rate_refresh.ref_id)
where  int_acc.brch_code = '784'
and    int_acc.prod_code = 'GF'
and    int_acc.ref_id = 'GFE1200077'
order by int_acc.rn,
         rate_refresh.rn;


REF_ID     REF_ID_1           RN       RN_1
---------- ---------- ---------- ----------
GFE1200077 GFE1200077          1          1
GFE1200077 GFE1200077          1          2
GFE1200077 GFE1200077          1          3
GFE1200077 GFE1200077          1          4
GFE1200077 GFE1200077          1          5
GFE1200077 GFE1200077          1          6
GFE1200077 GFE1200077          1          7
GFE1200077 GFE1200077          2          1
GFE1200077 GFE1200077          2          2
GFE1200077 GFE1200077          2          3
GFE1200077 GFE1200077          2          4
GFE1200077 GFE1200077          2          5
GFE1200077 GFE1200077          2          6
GFE1200077 GFE1200077          2          7
GFE1200077 GFE1200077          3          1
GFE1200077 GFE1200077          3          2
GFE1200077 GFE1200077          3          3
GFE1200077 GFE1200077          3          4
GFE1200077 GFE1200077          3          5
GFE1200077 GFE1200077          3          6
GFE1200077 GFE1200077          3          7
GFE1200077 GFE1200077          4          1
GFE1200077 GFE1200077          4          2
GFE1200077 GFE1200077          4          3
GFE1200077 GFE1200077          4          4
GFE1200077 GFE1200077          4          5
GFE1200077 GFE1200077          4          6
GFE1200077 GFE1200077          4          7

希望您能看到dbt_c_int_accr 中带有rn = 1 的行与dbt_c_rate_refresh 中的每一行相匹配。

这是因为 dbt_c_int_accr 中 rn = 1 的 brch_code、prod_code 和 ref_id 列与其他表中所有 7 行的相同列匹配 - 根据您的连接条件。 dbt_c_int_accr 中的其他 3 行也是如此。

你得到 28 行，因为 4 * 7 = 28。

这不是一个奇怪的结果； 正是您在该特定连接条件下所要求的。如果您追求不同的结果，那么您必须相应地修改您的加入。