Doxygen 1.12.0 mishandles Markdown section header following "<pre>" HTML markup.

[ncbrowns]

Describe the bug
While porting some documentation to use Doxygen 1.12.0 (was using 1.9.7), I discovered a problem in the handling of Markdown content like the following where a top-level parent section ending with <pre> markup is followed by a second-level section header. When the content is rendered in HTML, the "##" prefix for the "Second Level" section header appears to be interpreted as if the first "#" character is literal text at the end of the prior section and the second "#" specifies a top-level section header (like "Top Level").

[TOC]
 
# Top Level
 
1. This is a list item.

2. So is this.

3. So it this, but I also have some
<pre>
preformatted content
</pre>

## Second Level

This is a second level section.

Screenshots
Here's the HTML rendering of the example above:

To Reproduce
Attach a self contained example that allows us to reproduce the problem.

example.zip

This was built from a brand-new Doxyfile pointing at the file "test.md", with no other content.

Expected behavior
The header "Second Level" should be rendered as a child of "Top Level", not as a peer.

Version
Doxygen 1.12.0 (pre-built binaries from zip package on the main site) on Windows 11.

Stack trace
If you encounter a crash and can build doxygen from sources yourself with debug info (-DCMAKE_BUILD_TYPE=Debug), a stack trace can be very helpful (especially if it is not possible to capture the problem in a small example that can be shared).

Additional context
Add any other context about the problem here.

[albert-github]

There is also the extra # in the output.
As far as I can tell an extra newline, before the ##, will work around the problem.

[albert-github]

A bisection gave that the problem is introducend in doxygen 1.8.16 in:

bf1e768 is the first bad commit

commit bf1e768aa86ac5cbca5de4510469dd98334973df (HEAD)

Date:   Sat Jan 19 18:45:26 2019 +0100

    issue #6781 Unable to use math in markdown table headers

    Due to the change of the place where the markdown processing is done the end of the line must be calculated a little bit differently.

    Note: translator.py gave an error due to a strange indentation (did surface now), so had to be corrected as well.

 doc/translator.py |  6 +++---
 src/markdown.cpp  | 36 ++++++++++++------------------------
 2 files changed, 15 insertions(+), 27 deletions(-)

I've just pushed a proposed patch, pull request #11115

[albert-github]

Code has been integrated in master on GitHub (please don't close the issue as this will be done at the moment of an official release).

[ncbrowns]

Thanks for addressing this issue so quickly. I have verified that your proposed workaround of adding a blank line works with our real documentation in addition to the example.

[albert-github]

@ncbrowns Did you also try the real fix (you build the doxygen version based on the current master or get a build from the github actions artifacts, e.g. https://github.com/doxygen/doxygen/actions/runs/10741154727from the bottom of the page).

[ncbrowns]

@albert-github At the time of my previous update, I had tested only the workaround. However, I just tested the Windows MSVC Release build from your GitHub Actions link against documentation with the workaround removed and found that it worked fine.

[doxygen]

This issue was previously marked 'fixed but not released',
which means it should be fixed in doxygen version 1.13.0.
Please verify if this is indeed the case. Reopen the
issue if you think it is not fixed and please include any additional information
that you think can be relevant (preferably in the form of a self-contained example).