假设我有这种类型的数据集:
row Num_A Num_B Num_C Reg_1 Reg_2
------ ----- ----- ----- ----- -----
1 9400 9500 9700 AAA CCC
2 9700 9400 9500 BBB DDD
3 9500 9400 9200 EEE AAA
4 9100 9300 9900 AAA DDD
请考虑以下条件:
在第1行,Num_B
"95000"
出现在Num_A
(第3行),而通讯员Num_A
(第1行)"94000"
等于其Num_B
(第3行)(情况为“94000”)
所以新的列Comm_1
将有Num_A
和Num_B
的组合作为其值。同样适用于Comm_2
至于Comm_3
,条件与Comm_1
或Comm_2
相同,并且第1行Reg_1
(或Reg_2
)的'AAA'
(或'CCC'
)等于第3行的Reg_1
或Reg_2
(这种情况)的附加条件。
我想构造一个查询,以便结果如下所示:
row Num_A Num_B Num_C Reg_1 Reg_2 Comm_1 Comm_2 Comm_3
------ ----- ----- ----- ----- ----- ----- ----- -----
1 9400 9500 9700 AAA CCC 94009500 94009700 94009500
2 9700 9400 9500 BBB DDD 97009400 97009500 NULL
3 9500 9400 9200 EEE AAA 95009400 NULL 95009400
4 9100 9300 9900 AAA DDD NULL NULL NULL
我尝试了以下基本语法,但它不起作用:
SELECT
IF(Num_B in (select distinct Num_A from mydata) , concat(Num_A,Num_B), Null) as Comm_1,
IF(Num_C in (select distinct Num_A from mydata) , concat(Num_A,Num_C), Null) as Comm_2
FROM mydata
所以请提供任何提示并指出我阅读pl/sql
中的正确点或我需要学习的程序。
您可以通过外部自联接获得您想要的内容(我认为):
-- CTE for sample data, including row_num to maintain display order
with mydata (row_num, Num_A, Num_B, Num_C, Reg_1, Reg_2) as (
select 1, 94000, 95000, 97000, 'AAA', 'CCC' from dual
union all select 2, 97000, 94000, 95000, 'BBB', 'DDD' from dual
union all select 3, 95000, 94000, 92000, 'EEE', 'AAA' from dual
union all select 4, 91000, 93000, 99000, 'AAA', 'DDD' from dual
)
-- actual query
select m.*,
case when m1.num_a is not null then m.num_a || m1.num_a end as comm_1,
case when m2.num_a is not null then m.num_a || m2.num_a end as comm_2,
case when m3.num_a is not null then m.num_a || m3.num_a end as comm_3
from mydata m
left join mydata m1 on m1.num_a = m.num_b
left join mydata m2 on m2.num_a = m.num_c
left join mydata m3 on m3.num_a = m.num_b and (m3.reg_1 = m.reg_2 or m3.reg_2 = m.reg_1)
order by m.row_num;
ROW_NUM NUM_A NUM_B NUM_C REG REG COMM_1 COMM_2 COMM_3
---------- ---------- ---------- ---------- --- --- ---------- ---------- ----------
1 94000 95000 97000 AAA CCC 9400095000 9400097000 9400095000
2 97000 94000 95000 BBB DDD 9700094000 9700095000
3 95000 94000 92000 EEE AAA 9500094000 9500094000
4 91000 93000 99000 AAA DDD
(我在row_num
中只是为了保持显示顺序相同,否则不会使用;我认为这实际上不是表格中的一列。)
comm_1
值基于左连接到另一行,其num_a
匹配此行的num_b
。
comm_2
值基于左连接到另一行,其num_a
匹配此行的num_c
。
comm_3
值与comm_1
相同,但也必须匹配其reg_1
与另一行的reg_2
,反之亦然。
但是,这会计算第3行的值为9500094000而不是问题中显示的94009500 - 不清楚哪个是正确的。
如果任何外连接有多个匹配,这也将获得重复行;您的样本数据不是这种情况,但需要注意的事项。如果可能的话,我不知道你会怎么想处理它。
您也可以使用单个外部联接以及列表达式中的更多逻辑/聚合来执行此操作:
select m.row_num, m.num_a, m.num_b, m.num_c, m.reg_1, m.reg_2,
max(case when m1.num_a = m.num_b then m.num_a || m1.num_a end) as comm_1,
max(case when m1.num_a = m.num_c then m.num_a || m1.num_a end) as comm_2,
max(case when m1.num_a = m.num_b and (m1.reg_1 = m.reg_2 or m1.reg_2 = m.reg_1)
then m.num_a || m1.num_a end) as comm_3
from mydata m
left join mydata m1 on (m1.num_a = m.num_b) or (m1.num_a = m.num_c)
group by m.row_num, m.num_a, m.num_b, m.num_c, m.reg_1, m.reg_2
order by m.row_num;
ROW_NUM NUM_A NUM_B NUM_C REG REG COMM_1 COMM_2 COMM_3
---------- ---------- ---------- ---------- --- --- ---------- ---------- ----------
1 94000 95000 97000 AAA CCC 9400095000 9400097000 9400095000
2 97000 94000 95000 BBB DDD 9700094000 9700095000
3 95000 94000 92000 EEE AAA 9500094000 9500094000
4 91000 93000 99000 AAA DDD
如果我正确理解逻辑,你可以使用这些子查询:
select row_, Num_A, Num_B, Num_C, Reg_1, Reg_2,
(select min(m.num_a||' '||m.num_b) from mydata where num_a = m.num_b ) comm_1,
(select min(m.num_a||' '||m.num_c) from mydata where num_a = m.num_c ) comm_2,
(select min(m.num_a||' '||m.num_b) from mydata
where num_a = m.num_b and (reg_2 = m.reg_1 or reg_1 = m.reg_2)) comm_3
from mydata m
我使用min
以防万一匹配的行。