在PostgreSQL 8.4.13中我有2个表和一个程序来填充第二个表:
create table pref_users (
id varchar(32) primary key,
first_name varchar(64),
last_name varchar(64),
female boolean,
avatar varchar(128),
city varchar(64),
login timestamp default current_timestamp,
logout timestamp,
last_ip inet,
vip timestamp,
mail varchar(256)
);
create table pref_rep (
rep_id serial,
id varchar(32) references pref_users(id) check (id <> author) on delete cascade,
author varchar(32) references pref_users(id) on delete cascade,
author_ip inet,
good boolean,
fair boolean,
nice boolean,
about varchar(256),
stamp timestamp default current_timestamp
/* primary key(id, author) */
);
create or replace function pref_update_rep(_id varchar,
_author varchar, _author_ip inet,
_good boolean, _fair boolean, _nice boolean,
_about varchar) returns void as $BODY$
begin
delete from pref_rep
where id = _id and
age(stamp) < interval '1 hour' and
(author_ip & '255.255.255.0'::inet) =
(_author_ip & '255.255.255.0'::inet);
update pref_rep set
author = _author,
author_ip = _author_ip,
good = _good,
fair = _fair,
nice = _nice,
about = _about,
stamp = current_timestamp
where id = _id and author = _author;
if not found then
insert into pref_rep(id, author, author_ip, good, fair, nice, about)
values (_id, _author, _author_ip, _good, _fair, _nice, _about);
end if;
end;
$BODY$ language plpgsql;
pref_users
表包含有关用户的一般信息。
pref_rep
包含有关其他用户(列about
)创建的用户(列id
)的评论(列author
)。
对于第二张表,我忘记申报primary key对(该行在上面评论过)。
我正在尝试在psql提示符处添加该主键,但它失败了 - 可能是因为某些原因(我不知道上面的程序怎么会发生?)我有几条记录同样{{ 1}}多次评论author
:
id
我的问题是如何找到重复的# alter table pref_rep add primary key(id, author);
NOTICE: ALTER TABLE / ADD PRIMARY KEY will create implicit index "pref_rep_pkey" for table "pref_rep"
ERROR: could not create unique index "pref_rep_pkey"
DETAIL: Table contains duplicated values.
和id
对?
我试过了:
author
但那当然不会给我这双......
更新: Catcall的建议(谢谢!)给了我190个这样的重复对:
# select id, count(id) from pref_rep group by id order by count desc limit 5;
id | count
----------------+-------
OK408547485023 | 706
OK261593357402 | 582
DE11198 | 561
DE13041 | 560
OK347613386893 | 556
(5 rows)
但实际上我真正的问题是如何删除重复项的旧版本(通过 id | author | count
------------------------+------------------------+-------
DE10598 | OK495480409724 | 2
DE12188 | MR17925810634439466500 | 3
DE13529 | OK471161192902 | 2
DE13963 | OK434087948702 | 2
DE14037 | DE7692 | 2
......
VK45132921 | DE3544 | 2
VK6152782 | OK261593357402 | 2
VK72883921 | OK506067284178 | 2
(190 rows)
列)?我在psql提示符下尝试了很多查询失败...
答案 0 :(得分:2)
这应该识别重复项。
select id, author
from pref_rep
group by id, author
having count(id) > 1
您可能还必须查看NULL,因为这两列都允许NULL。
答案 1 :(得分:1)
此查询的内容(也在SQL Fiddle上)?
DELETE FROM pref_rep p USING (
SELECT id, author, max(stamp) stamp
FROM pref_rep
GROUP BY id, author
HAVING count(1) > 1) AS f
WHERE p.id=f.id AND p.author=f.author AND p.stamp<f.stamp;
检查count()
功能上的manual。
您可以指定任何表达式。 1
表示将计算所有行,'cos 1
永远不会NULL
。使用count(*)
时效果相同。事实上我更喜欢后者,不知道为什么我这次使用count(1)
:)