convert-links fails to convert link when URLs has "|" or "," characters

Tested on wget2 2.2.0.

I tried:

wget -rEpkH -np https://sites.google.com/tzfonet.org.il/shchanyaschool/home

Which downloads a bunch of files (roughly 413MB), and the top page is stored in the file:

sites.google.com/tzfonet.org.il/shchanyaschool/home.html

When you look at this file, you see that -k (--convert-links) usually works well, for example we see in one place

link href="../../../fonts.googleapis.com/css%3Ffamily=Google%20Sans:400,500%7CRoboto:300,400,500,700%7CSource%20Code%20Pro:400,700&display=swap.css" rel="stylesheet"

Note the relative link into a downloaded file. So far so good.

But the home.html also contains:

 href="https://fonts.googleapis.com/css?family=Montserrat%3A400%2C700%7CBitter%3A400%2C700&display=swap" rel="stylesheet"

This is a bug - this link was not converted! Looking at which files got loaded, I see this file was loaded: The file

fonts.googleapis.com/css?family=Montserrat%3A400,700|Bitter%3A400,700&display=swap.css

does exist! But somehow, the convert-links didn't find it. My guess is that the URL encoding of the "," character (%2C) or the "|" character (%7C) confused it: wget downloaded a URL with %2C and %7C but saved it into a file with "," and "|" in its name - but when it came time to convert-links it expected to find the file with %2C and %7C in its name, failed to find it, and failed the conversion.