postgres:join,group by,order by。我的查询错了吗?

时间:2017-09-28 17:48:31

标签: postgresql join group-by

我的桌子

CREATE TABLE IF NOT EXISTS members(
    group_id BIGINT, 
    amount INT, 
    updated_date timestamp)

CREATE TABLE IF NOT EXISTS supergroups_ref(
    group_id BIGINT PRIMARY KEY, 
    title TEXT DEFAULT NULL, 
    message_date timestamp)

CREATE TABLE IF NOT EXISTS supergroups(
    group_id BIGINT PRIMARY KEY, 
    lang TEXT DEFAULT NULL, 
    last_date timestamp DEFAULT NULL)

成员中的数据示例

(3, 90, date)
(1, 15, date)
(2, 200, date)
(1, 28, date)
(2, 300, date)
(1, 80, date)

supergroups_ref中的数据示例:

(1, 'title', date)
(2, 'title2', date)
(3, 'title3', date)

超组中的数据示例:

(1, 'en', date)
(2, 'it', date)
(3, 'it', date)

我想能够进行查询 从表成员中只获取最新的行,通过'updated_date'DESC对它们进行排序,并将它们连接到超级组,以便超组中的每个group_id都具有最新的成员数。 现在我想在supergroups_ref

中添加另一个左连接标题信息

我希望按成员数量表来订购。

我试图这样做,但我不确定我做得对:

SELECT sub.group_id, sub.amount, sub.title
FROM (
    SELECT s.group_id, m.amount, s_ref.title
    FROM  supergroups AS s
    LEFT OUTER JOIN members AS m
    ON m.group_id = s.group_id
    LEFT OUTER JOIN supergroups_ref AS s_ref
    ON s_ref.group_id = s.group_id
    ORDER BY m.updated_date DESC
) AS sub
GROUP BY sub.group_id, sub.amount, sub.title
ORDER BY sub.amount DESC
我害怕:

  1. 我在查询中做错了什么
  2. 我使查询更加复杂化
  3. 期望的结果将是: [(2, 200, 'title2'), (3, 90, 'title3'), (1, 15, 'title1')] 每个元素的第二个元素是最近的成员数量,因为数量也存储在成员表中的日期。而元组是由那个

    命令的

1 个答案:

答案 0 :(得分:1)

如果我清楚地告诉你,这就是你需要的:

SELECT members.*,supergroups.lang,supergroups_ref.title
FROM
-- Window function to get only de last_date:
    (SELECT last_members.group_id,last_members.amount
    FROM
     (SELECT *,row_number() OVER (PARTITION BY group_id
     ORDER BY updated_date DESC) as row FROM members)last_members
    WHERE last_members.row=1)members
-- Joins with other tables
LEFT JOIN supergroups ON members.group_id=supergroups.group_id
LEFT JOIN supergroups_ref ON supergroups.group_id=supergroups_ref.group_id

这应该返回:

group_id |  amount  |   lang  | title
--------------------------------------
1        |  15      |   en    | title
2        |  200     |   it    | title2
3        |  90      |   it    | title3

要使此查询起作用,您需要使用窗口函数,这里有一篇关于它们的好文章Understanding Window Functions