Unencoded header links didn't work.

[garmede]

Describe the bug

Unencoded header links (with CJK characters) do not work.

I remember it worked fine in VuePress, but it doesn't work in VitePress.

Reproduction

# Header

This link [한글](#한글-한글) is not work!

This link [한글](#%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF-%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF)  is work.

...

## 한글 한글

...

Expected behavior

Creating links with native headers should also work.

System Info

System:
  OS: Windows 10 10.0.22621
  CPU: (12) x64 AMD Ryzen 5 3600X 6-Core Processor
  Memory: 46.05 GB / 63.92 GB
Binaries:
  Node: 19.3.0 - C:\Program Files\nodejs\node.EXE
  npm: 9.2.0 - C:\Program Files\nodejs\npm.CMD
Browsers:
  Edge: Spartan (44.22621.963.0), Chromium (108.0.1462.54)
  Internet Explorer: 11.0.22621.1
npmPackages:
  vitepress: 1.0.0-alpha.34 => 1.0.0-alpha.34

Additional context

No response

Validations

• Follow our Code of Conduct
• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.

[brc-dd]

Probably has something to do with your browser. We don't manually handle the URL fragments. For me, both are working fine on Chrome 108: https://stackblitz.com/edit/vite-kdiy46?file=docs%2Fexample.md

407cb01215.mp4

[garmede]

In the html source I built, the address is encoded differently as follows.

<!-- [한글](#한글-한글) to -->
<a href="#%ED%95%9C%EA%B8%80-%ED%95%9C%EA%B8%80">한글</a>

<!-- [한글](#%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF-%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF) to -->
<!-- This address was encoded by copy pasting from the browser address bar. -->
<a href="#%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF-%E1%84%92%E1%85%A1%E1%86%AB%E1%84%80%E1%85%B3%E1%86%AF">한글</a>

[brc-dd]

I'm not completely sure, but there is an issue with the text in your first link. Copy paste this: #한글-한글 there (or from my stackblitz link). Both are looking same to me after decoding, but JS tells there is a difference. I don't speak Korean, so not sure what's that:

PS: it looks like second one is using multiple characters to represent a single letter. (VitePress is generating ids having those, even if the source has unicode stuff).

The string normalization is causing this: https://github.com/mdit-vue/mdit-vue/blob/main/packages/shared/src/slugify.ts#L11

[garmede]

Exactly! Both are exactly the same Hangul, but their Unicode is different. Shorter is better.
Unfortunately, VitePress handles Korean characters on the long side.