使用 Chinese_Hong_Kong_Stroke_90_CI_AS 排序规则时 SQL Server 中的模糊匹配

Question

Suppose we create a table as follows:

create table my_table (
id int, 
city nvarchar(256) collate Chinese_Hong_Kong_Stroke_90_CI_AS)

INSERT INTO my_table (id, city)
VALUES (1, 'Shanghai');

INSERT INTO my_table (id, city)
VALUES (2, 'Shandong');

INSERT INTO my_table (id, city)
VALUES (3, 'Shanxi');

INSERT INTO my_table (id, city)
VALUES (4, 'Shaanxi');

There are now four records in my_table:

id  city
1   Shanghai
2   Shandong
3   Shanxi
4   Shaanxi

The following SQL queries return the same number. How to avoid this error?

select top 1 id from my_table order by DIFFERENCE(city, 'Shanghai') desc
select top 1 id from my_table order by DIFFERENCE(city, 'Shandong') desc

Another problem:

select top 1 id from my_table order by DIFFERENCE(city, 'Shannxi') desc 

Returns 3 when the correct answer should be 4.

Answer 1

The issue is caused by the collation of your column. As per the docs SOUNDEX & DIFFERENCE are collation dependent.

A possible solution is:

select top 1 id
from my_table
order by DIFFERENCE(city collate SQL_Latin1_General_CP1_CI_AS, 'Shanghai') desc

select top 1 id
from my_table
order by DIFFERENCE(city collate SQL_Latin1_General_CP1_CI_AS, 'Shandong') desc

I think I would add another column with a SQL_Latin1_General_CP1_CI_AS collation which stores exactly the same value as city .

Although it would be of interest to know exactly what you are trying to accomplish? Because in your current collation those 2 words apparently sound exactly the same.

Its also worth reading Beyond SOUNDEX & DIFFERENCE