linkcheck incorrectly splits anchor if # is part of URI

[rsokolewicz]

Describe the bug

I wanted to add the URL https://microsoft.github.io/pyright/#/type-concepts-advanced to my documentation, but when running the link check, sphinx-build -b linkcheck . _build I get an error message:

(           index: line    5) broken    https://microsoft.github.io/pyright/#/type-concepts-advanced - Anchor '/type-concepts-advanced' not found

In this case, the URL is split by # and the entire righthand part is assigned as an anchor, while the # does not designate an anchor but is just part of the directory path.

How to Reproduce

minimal reproducible example:

conf.py

# Add the following import at the top of the file
import sphinx.ext.extlinks

# Project information
project = "My Project"
author = "Your Name"

# Extensions
extensions = [
    "sphinx.ext.extlinks",  # Add the extension here
]

# Define your external link patterns
extlinks = {
    "url": ("%s", None),  # Define the 'url' role pattern
}

# Theme
html_theme = "alabaster"

index.rst

title
-----


:url:`https://microsoft.github.io/pyright/#/type-concepts-advanced`

and running sphinx-build -b linkcheck . _build in the same directory.

Environment Information

running with `sphinx==7.0.1`.

Sphinx extensions

No response

Additional context

No response

[jayaddison]

Although the fragment content (/type-concepts-advanced) looks like a path, it's likely that JavaScript loaded from the base URL (https://microsoft.github.io/pyright/) parses that and loads relevant content by parsing the fragment.

That said, we should still be able to get the linkcheck builder to pass on the base URL (it's a less strong guarantee that the link is correct, but better than indicating failure).

@rsokolewicz could you try configuring the linkcheck_anchors_ignore in your configuration to specify that URLs within https://microsoft.github.io/pyright/ should not be anchor-checked?

[rsokolewicz]

Hi James,

Thanks for the reply. The linkcheck_anchors_ignore is to ignore anchors btw, not base_urls. In the project I was working on I now set

linkcheck_anchors_ignore = ["/type-concepts-advanced"]

which works, so we're not blocked at the moment.

Alternatively, we could also ignore the base_url using

linkcheck_ignore = ["https://microsoft.github.io/pyright/#"]

[jayaddison]

You're absolutely right, my mistake - thanks @rsokolewicz.