表1
|------------------------------------------------------------|
| Para_1 | column_A | column_B | column_C |
--------------------------------------------------------------
| 3007576 | abc | | |
| 3007879 | ab | fg | |
| 3007880 | ad | | x |
| 3007900 | | | |
|------------------------------------------------------------|
表2
|------------------------------------------------------------|
| Para_2 | column_A | column_B | column_C |
--------------------------------------------------------------
| 100 | abcd | fgh | xyz |
| 200 | abc | fg | z |
| 300 | ab | g | xy |
|------------------------------------------------------------|
预期结果:
|------------------------------------------------------------|
| Para_1 | column_A | column_B | column_C | Para_2
--------------------------------------------------------------
| 3007576 | abc | | | 100
| 3007576 | abc | | | 200
| 3007879 | ab | fg | | 100
| 3007879 | ab | fg | | 200
| 3007880 | ad | | x | null
| 3007900 | | | | null
|------------------------------------------------------------|
select table1.*, table2.Para_2, table1.column_A
from table1
left outer join table2
on table2.column_A like ('%'||table1.column_A||'%')
and table2.column_B like ('%'||table1.column_B||'%')
and table2.column_C like ('%'||table1.column_C||'%')
where table1.column_A is not null
and table1.column_B is not null
and table1.column_C is not null
上面的代码似乎还不够..有什么想法吗?
答案 0 :(得分:0)
删除where子句是因为在示例数据的每一行中,至少有一个空列值以及3列A,B,C,因此它过滤了所有内容
<Grade MarkGrade="2" Label="GOOD" TestBins="2">2</Grade>
答案 1 :(得分:0)
这将产生预期的结果:
select table1.*, table2.Para_2
from table1 left outer join table2
on table2.column_A like '%'||table1.column_A||'%'
and table2.column_B like '%'||table1.column_B||'%'
and table2.column_C like '%'||table1.column_C||'%'
and coalesce(table1.column_a, table1.column_b, table1.column_c) is not null;
如您所见,这几乎就是您所拥有的。但是不需要您的where
,只需on
条件的最后一行。如果您不喜欢coalesce
函数,那么也可以像这样更清楚地写出额外的行:
and not (table1.column_a is null and table1.column_b is null and table1.column_c is null);
或者:
and length(table1.column_a||table1.column_b||table1.column_c)>0;