collation: add new pinyin collation utf8mb4_zh_pinyin_tidb_as_cs

[xiongjiwei]

What problem does this PR solve?

add a new collation named utf8mb4_zh_pinyin_tidb_as_cs which has collation id 2048
proposal pingcap/tidb#19984
tidb PR pingcap/tidb#20504

What is changed and how it works?

a new method to get collation by given id.

Check List

Tests

• Unit test
• Integration test

Code changes

• Has exported function/method change

[wjhuang2016]

How about rename utf8mb4_general_zh_ci to utf8mb4_zh_ci

[kennytm]

if we want to follow MySQL's naming convention it should be utf8mb4_chinese_ci (similar to big5_chinese_ci, gb18030_chinese_ci, gb2312_chinese_ci, gb2312_chinese_ci).

(note that CLDR also defined the "stroke" and "zhuyin" order, and we could have utf8mb4_chinese_stroke_ci etc.)

[wjhuang2016]

I don't want to follow MySQL's old naming convention, zh is more standard than chinese.
As you can see in the new collation introduced in 8.0, utf8mb4_zh_0900_as_cs uses zh.
Consider that CLDR has defined "pinyin", "stroke" and "zhuyin", we can use "utf8mb4_zh_pinyin", "utf8mb4_zh_stroke", "utf8mb4_zh_zhuyin". If we want to distinguish case-sensitive, we can add extra '_cs' or '_ci'.

[kennytm]

🤷

however this collation is explicitly not following (MySQL's implementation of) UCA, otherwise we will be implementing pingcap/tidb#19747.

given that the naming convention of all 8.0 collations are <charset>[_<language>[_<variant>]]_<unicodeversion>(_<attribute>)+, perhaps

• utf8mb4_zh_pinyin_tidb0500_ci
• utf8mb4_zh_tidb0500_ci ?
• (utf8mb4_zh_stroke_tidb0500_ci)

[wjhuang2016]

LGTM

[kennytm]

assuming it is a typo, rest LGTM.

(btw, are we sure we want case-sensitive ('a' ≠ 'A') not case-insensitive ('a' = 'A') here?)

[bb7133]

LGTM

Thanks for your suggesion @kennytm