我有下表:
Customers
---------
name text
object_id integer
created_time timestamp with time zone
Indexes:
"my_index" UNIQUE CONSTRAINT, btree (name, object_id, created_time)
唯一索引工作正常,但后来我得到了重复的数据,如:
Name | object_id | created_time
------------------------------------
john | 1 | 2018-02-28 15:42:14.30573+00
JOHN | 1 | 2018-02-28 15:42:14.30573+00
所以我尝试用名称列小写我的所有数据:
UPDATE customers SET name=lower(name) WHERE name != LOWER(name);
但是这个过程产生了错误,因为现在我违反了索引:
ERROR: duplicate key value violates unique constraint "my_index"
DETAIL: Key (name, object_id, created_time)=(john, 1, 2018-02-28 15:42:14.30573+00) already exists.
我可以使用什么样的过程来删除在转换为小写后生成索引违规的行?
如果你在桌子上有'JOHN'
和'John'
而不是'john'
它会变得混乱。这是一个解决方案。
insert into customers
select distinct lower("name") ,object_id,created_time from customers
where name <> lower(name)
and not (lower("name") ,object_id,created_time)
in (select * from customers);
delete from customers where name <> lower(name);
之后考虑:
alter table customers alter column name type citext;