Question

我有一个像这样的数据库架构：

CREATE TABLE projects (project_id int);
CREATE TABLE project_members (user_id int, project_id int);
CREATE TABLE project_issues (issue_id int, project_id int);

我想要的是找到所有拥有10个或更多成员以及10个或更多问题的项目。我正在努力想出一个查询。

我希望输出类似于：

project_id | members | issues
-----------+---------+-------
65         | 100     | 23
93         | 78      | 45

最好按成员排序然后发布。

我想出了：

SELECT projects.project_id, COUNT(project_members.user_id) FROM 
project_members LEFT JOIN projects ON 
project_members.project_id = projects.project_id GROUP BY projects
ORDER BY count;

但不知道如何在计算问题时将其提升到新的水平。

我正在使用PostgreSQL 9.1

Answer 1

有点简单＆amp;更快：

SELECT p.project_id, m.members, i.issues
FROM   projects p
JOIN  (
    SELECT project_id, count(*) AS members 
    FROM   project_members 
    GROUP  BY project_id
    HAVING count(*) >= 10
    ) m USING (project_id)
JOIN  (
    SELECT project_id, COUNT(*) AS issues 
    FROM   project_issues
    GROUP  BY project_id
    HAVING count(*) >= 10
    ) i  USING (project_id)
ORDER  BY m.members DESC, i.issues DESC;

既然你想要项目，

有10个或更多成员以及10个或更多个问题

..您可以通过JOIN和HAVING提前取消不符合资格的候选人，以获得更快的结果。
还有其他一些小调整。

并将>= 10或> 9设为“10或更多”！

警惕“代理交叉连接”：
Two SQL LEFT JOINS produce incorrect result

Answer 2

这可能会更有效率，但它会给出正确的结果。

SELECT 
  p.project_id,
  m.members,
  i.issues
FROM projects AS p
LEFT JOIN 
  (
      SELECT project_id, COUNT(user_id) AS members 
      FROM project_members 
      GROUP BY project_id
  ) AS m
ON p.project_id = m.project_id
LEFT JOIN 
  (
      SELECT project_id, COUNT(issue_id) AS issues 
      FROM project_issues
      GROUP BY project_id
  ) AS i
ON p.project_id = i.project_id
WHERE members > 10 AND issues > 10
ORDER BY members, issues;

SQL在多个表中计数超过1个值

2 个答案: