Question

有三个表：businesses，categories，categorizations，

CREATE TABLE businesses (
  id SERIAL PRIMARY KEY,
  name varchar(40)
);

CREATE TABLE categories (
  id SERIAL PRIMARY KEY,
  name varchar(40)
);

CREATE TABLE categorizations (
  business_id integer,
  category_id integer
);

所以business有很多categories到categorizations。

如果我想选择没有类别的企业，我会做点什么像这样：

SELECT businesses.* FROM businesses
LEFT OUTER JOIN categorizations
ON categorizations.business_id = businesses.id
LEFT OUTER JOIN categories
ON categories.id = categorizations.category_id
GROUP BY businesses.id
HAVING count(categories.id) = 0;

问题是：如何选择没有类别AND的企业名为＆＃34; Media＆＃34;在一个查询？

Answer 1

您可以使用联合：

SELECT businesses.* 
FROM businesses
LEFT OUTER JOIN categorizations
ON categorizations.business_id = businesses.id
GROUP BY businesses.id
HAVING count(categorizations.business_id) = 0

UNION 

SELECT businesses.* 
FROM businesses
INNER JOIN categorizations
ON categorizations.business_id = businesses.id
INNER JOIN categories
ON categories.id = categorizations.category_id
WHERE categories.name = 'Media';

请注意，在第一个实例（根本没有类别的企业）中，您不需要加入类别 - 您可以检测联结表中缺少类别。如果同一个商家可能多次拥有相同的类别，则需要使用DISTINCT引入第二个查询。

Answer 2

我会尝试：

SELECT b.* FROM businesses b
    LEFT JOIN categorizations cz ON b.business_id = cz.business_id
    LEFT JOIN categories cs ON cz.category_id = cs.category_id
    WHERE COALESCE(cs.name, 'Media') = 'Media';

...希望没有businesses的{{1}}在其连接上获得NULL条目。

Answer 3

双重否定技巧适用于这种选择：

SELECT * FROM businesses b
WHERE NOT EXISTS (
    SELECT *
    FROM categorizations bc
    JOIN categories c ON bc.category_id = c.category_id
    WHERE bc.business_id = b.business_id
    AND c.name <> 'Media'
    );

Has-Many-Through：如何选择没有关系的记录或关系中的某些条件？

3 个答案: