Regression? Heading anchors duplicated. Manual overrides ignored in v. 60.0

[kurankat]

What version of Hugo are you using (hugo version)?

$ hugo version
Hugo Static Site Generator v0.60.0/extended windows/amd64 BuildDate: unknown

Does this issue reproduce with the latest release?

Yes

Issues

I maintain a Flora which has long lists of scientific names (Genus species) and long identification keys. I have been using manual override anchors in the format of {#anchor-text} to override the unpredictability of Hugo's automatic anchors

My headers are in the format of:

## 8 SPRENGELIA

### **1 Sprengelia incarnata** Sm., *Kongl. Vetensk. Acad. Nya Handl.* Ser. 2, 15: 263 (1794) {#1-s.-incarnata}

...

## 14 STYPHELIA

### **1 Styphelia adscendens** R.Br., *Prodr. Fl. Nov. Holland.* 537 (1810) {#1-s.-adscendens} 

Previously (v. 58) the resulting HTML was:

<h2 id="8-sprengelia">8 SPRENGELIA</h2>

<h3 id="1-s.-incarnata"><strong>1 Sprengelia incarnata</strong> Sm., <em>Kongl. Vetensk. Acad. Nya Handl.</em> Ser. 2, 15: 263 (1794)</h3>

...

<h2 id="14-styphelia">14 STYPHELIA</h2>

<h3 id="1-s.-adscendens"><strong>1 Styphelia adscendens</strong> R.Br., <em>Prodr. Fl. Nov. Holland.</em> 537 (1810)</h3>

The resulting HTML is now:

<h2 id="8-sprengelia">8 SPRENGELIA</h2>

<h3 id="1-s" class="-incarnata"><strong>1 Sprengelia incarnata</strong> Sm., <em>Kongl. Vetensk. Acad. Nya Handl.</em> Ser. 2, 15: 263 (1794)</h3>

...

<h2 id="14-styphelia">14 STYPHELIA</h2>

<h3 id="1-s" class="-adscendens"><strong>1 Styphelia adscendens</strong> R.Br., <em>Prodr. Fl. Nov. Holland.</em> 537 (1810)</h3>

The overrides are being ignored and the ID anchors (which need to be unique) are being repeated.

[kurankat]

Additional information:

This is related to Goldmark. If I revert to BlackFriday, the issue disappears, and custom anchors are rendered correctly.

With custom anchors:

<h3 id="1-s.-incarnata"><strong>1 Sprengelia incarnata</strong> Sm., <em>Kongl. Vetensk. Acad. Nya Handl.</em> Ser. 2, 15: 263 (1794)</h3>

...

<h3 id="1-s.-adscendens"><strong>1 Styphelia adscendens</strong> R.Br., <em>Prodr. Fl. Nov. Holland.</em> 537 (1810)</h3>

Without custom anchors:

<h3 id="1-sprengelia-incarnata-sm-kongl-vetensk-acad-nya-handl-ser-2-15-263-1794"><strong>1 Sprengelia incarnata</strong> Sm., <em>Kongl. Vetensk. Acad. Nya Handl.</em> Ser. 2, 15: 263 (1794)</h3>

...

<h3 id="1-styphelia-adscendens-r-br-prodr-fl-nov-holland-537-1810"><strong>1 Styphelia adscendens</strong> R.Br., <em>Prodr. Fl. Nov. Holland.</em> 537 (1810)</h3>

[kaushalmodi]

What you are seeing is because of the new attributes support in headings in Goldmark. See https://github.com/yuin/goldmark/blob/master/README.md#headings.

May be this should be a bug report on the Goldmark repo.. you do not have a space before the dot (.) in your id, and still it's parsed as a class attribute specifier. I don't know if that's the expected behavior, so you'll need to open an issue there.

A temporary workaround might be to disable the attribute rendering in Goldmark. See https://gohugo.io/getting-started/configuration-markup/#goldmark.

[kurankat]

Thanks @kaushalmodi,

There is indeed a change in syntax for how Goldmark render custom attributes compared to BlackFriday's way of rendering custom anchor tags on headings. This is not a bug and just a change in syntax required. Unfortunately from Goldmark: "This syntax possibly changes in the future. Headings"

My workaround was to revert to BlackFriday until I can refactor the sites I keep.

The bug is the repeated ID categories. They should be unique, or they are useless for most purposes.

Should I be reporting this as a Goldmark bug?

[kaushalmodi]

The bug is the repeated ID categories. They should be unique, or they are useless for most purposes.

Yes, Blackfriday auto-uniquified the id's even if the user manually provided identical id's.

It would be a very useful feature request on Goldmark repo.

[kurankat]

Thanks @kaushalmodi!

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.