Repository navigation
Auto-generated header links ignore mis-anchor headers that start with a number #527
Description
Activity
Just realized another problem this surfaces - if people have multiple numbered sections (e.g. in a changelog, 3.0, 3.1, 3.2) then adding a new section at the top (e.g, 4.0) will bump the
id#of all subsequent labels, making links to them valid links, but not to the correct section.This may be similar to #195, whereby docutils/sphinx seems to treat such references differently.
Should probably track down in the code exactly where/why it is doing thisI think its treating them as "auto-targets", similar to autofootnote; the only fix may be to stick a prefix onto such slugs e.g. `if target.isdigit(): target = "-" + target"
this would not be in keeping with github slug rules, but may be the only waymaybe they could just be prefixed with
num-or evenn-if they started with a number?
Describe the bug
When section headers start with a number, they are labeled in a confusing way by the auto-header anchor generator. They simply have the structure
id<num>.For example, a document like this:
Will generate anchors like:
I think users would instead expect the ID of the second element to be
2-0Since we describe the behavior as based on the GitHub sluggify, I also quickly checked and it looks like they do support IDs that start with numbers (that's why I'm labeling this as a bug since I think it breaks from expected behavior based on that criteria).
I think the relevant code for this might be here:
https://github.com/executablebooks/mdit-py-plugins/blob/master/mdit_py_plugins/anchors/index.py#L119-L120
originally reported in: jupyterhub/kubespawner#575
Reproduce the bug
Set
myst_heading_anchors = 2Create a page like
Render it and note the heading IDs
List your environment
No response