我在INNODB表中有大约20,000行称为'cards',所以FULLTEXT不是一个选项。
请考虑此表:
id | name | description
----------------------------------------------------------
1 John Smith Just some dude
2 Ted Johnson Another dude
3 Johnathan Todd This guy too
4 Susan Smith Her too
5 Sam John Bond And him
6 John Smith Same guy as num 1, another record
7 John Adams Last guy, promise
因此,假设用户搜索“John”,我希望结果集的顺序为:
7 John Adams
6 John Smith
3 Johnathan Todd
5 Sam John Bond
2 Ted Johnson
请注意,我们只拉了一次'John Smith',我们最近参加了他的录取。由于我的数据,所有名字都是针对同一个人,不需要担心两个名叫John Smith的人。 想法?如果我能澄清任何事情,请告诉我。
答案 0 :(得分:110)
版本1:
SELECT max(id) id, name
FROM cards
WHERE name like '%John%'
GROUP BY name
ORDER BY CASE WHEN name like 'John %' THEN 0
WHEN name like 'John%' THEN 1
WHEN name like '% John%' THEN 2
ELSE 3
END, name
第2版:
SELECT max(id) id, name
FROM cards
WHERE name like '%John%'
GROUP BY name
ORDER BY CASE WHEN name like 'John%' THEN 0
WHEN name like '% %John% %' THEN 1
WHEN name like '%John' THEN 2
ELSE 3
END, name