如何使确切的 unicode 字符优先于 ASCII 版本？

Question

我有一个数据库，其中包含德国城镇和城市的名称，例如慕尼黑和明斯特。

如果我这样查询：

SELECT name,
       MATCH(name) AGAINST('+mün*' IN BOOLEAN MODE) AS relevance
FROM place_names
ORDER BY relevance DESC

对于包含 mun、mün 或任何其他在不考虑变音符号的情况下扁平化为 mun 的文本，我得到相同的相关值。换句话说，搜索 mun 或 mün 会得到完全相同的结果。

我如何配置我的数据库，以便搜索 mün 会为实际包含字母 ü 的词提供更高的相关性，但仍将 u 视为匹配项？

Answer 1

CREATE TABLE place_names (id SERIAL PRIMARY KEY, name VARCHAR(255));
CREATE FULLTEXT INDEX idx ON place_names (name);
INSERT INTO place_names (name) VALUES ('Munich'), ('Münster');
SELECT * FROM place_names;

id	name
1	Munich
2	Münster

SELECT name,
       MATCH(name) AGAINST('+mün*' IN BOOLEAN MODE) AS relevance
FROM place_names
ORDER BY relevance DESC;
name relevance

Munich 0.000000001885928302414186

Münster 0.000000001885928302414186

ALTER TABLE place_names ADD COLUMN name2 VARCHAR(255) COLLATE utf8mb4_0900_bin AS (name) STORED;
CREATE FULLTEXT INDEX idx2 ON place_names (name2);

SELECT name,
       MATCH(name) AGAINST('+mün*' IN BOOLEAN MODE) AS relevance,
       MATCH(name2) AGAINST('+mün*' IN BOOLEAN MODE) AS relevance2
FROM place_names
ORDER BY relevance DESC;
name relevance relevance2

Munich 0.000000001885928302414186 0

Münster 0.000000001885928302414186 0.0906190574169159

db<>fiddle here

因此

SELECT name,
       MATCH(name) AGAINST('+mün*' IN BOOLEAN MODE) AS relevance
FROM place_names
ORDER BY MATCH(name2) AGAINST('+mün*' IN BOOLEAN MODE) DESC;

Answer 2

一种方法可能是：

WHERE MATCH(name) AGAINST ('+mün*' IN BOOLEAN MODE) AS relevance
ORDER BY name LIKE '%Mün%' COLLATE utf8mb4_bin DESC, relevance DESC

另一件需要注意的事情是 MySQL 8.0 中存在排序规则 utf8mb4_0900_as_ci——“区分重音和不区分大小写”。（但是，那根本不匹配“Mun”。）

如何使确切的 unicode 字符优先于 ASCII 版本？

How to make exact unicode characters take priority over ASCII versions?

mysql

sql

unicode

full-text-search

diacritics

name	relevance
Munich	0.000000001885928302414186
Münster	0.000000001885928302414186