Heading links improvement

[SleepWalker]

Motivation

• add support for non-latin heading anchors (fixes onPageNav links not working for headings written in Chinese #541)

Have you read the Contributing Guidelines on pull requests?

Yes

Test Plan

I've wrote some test, that should be enough on my opinion.

To check the fixes manually you need write the doc file with the following lines:

## тест тест тест
## 第一章

By rendering on master branch you will get following anchor links:

#
#

After this PR applied:

#тест-тест-тест
#第一章

Things to discuss

Old character mapping in toSlug()

I have found the following filtering rule in lib/core/toSlug.js:

  //  var accents = "àáäâèéëêìíïîòóöôùúüûñç";
  const accents =
    '\u00e0\u00e1\u00e4\u00e2\u00e8' +
    '\u00e9\u00eb\u00ea\u00ec\u00ed\u00ef' +
    '\u00ee\u00f2\u00f3\u00f6\u00f4\u00f9' +
    '\u00fa\u00fc\u00fb\u00f1\u00e7';

  const without = 'aaaaeeeeiiiioooouuuunc';

This code does not cover all possible characters, so I think it can be removed. But if we do so, it may be considered as breaking change, right? As for now I left it as is.

[JoelMarcey]

@SleepWalker Hi. If we go with this, you will need to rebase since CHANGELOG.md has been changed since this PR was made. Maybe just don't worry about making changes to the change log and I can handle that in the next release?

[SleepWalker]

@JoelMarcey pushed changes without touching CHANGELOG

[microbouji]

The unicode handling looks good to me. Great work!

For the unique slugs part, not sure how it's meant to work, but currently this:

# baz
# baz 1
# baz

results in

#baz
#baz-1
#baz-1

I guess that's a negligible edge case.

More importantly though, we'll have to find a way to reuse this in the new on-page nav. anchors.js is a remarkable plugin, while getTOC.js (used by the on page nav) uses markdown-toc. I believe there are good reasons to keep them both like this, but rewriting one or the other to avoid doing the same thing twice could also make sense.

My suggestion is you separate this into two PRs: one for the unicode changes that can land immediately, and another to make slugs unique that will also take into consideration the on-page nav.

[SleepWalker]

For the unique slugs part, not sure how it's meant to work, but currently this:

Oh, thank you. I have missed that case. The correct ids should be the following:

#baz
#baz-1
#baz-2

I have removed commit with headings uniqueness logic from current branch and will submit it as a separate PR including support for on-page nav.

[JoelMarcey]

This is great @SleepWalker. Thanks!

Thank you @microbouji for your review.