A "thematic break":https://spec.commonmark.org/0.29/#thematic-breaks composed of hyphens (e.g. "@---@") breaks the division of Markdown text into individually editable sections.

h2. Steps to reproduce

Configure @markdown@ text formatting, create a Wiki page in a web browser and enter the following content:

# Title
## Heading 2
Preceding CRLF is the default for web-submitted data.

---

End of thematic breaks.

## Heading 2
Nulla nunc nisi, egestas in ornare vel, posuere ac libero.

More in the unit tests in the enclosed patch.

h2. Cause

The reason is that it is confused with a "setext heading":https://spec.commonmark.org/0.29/#setext-headings. Although the current regexp in @extract_sections@ actually tries to restrict setext headings in a way that it must follow a non-empty line, it does not account for a whitespace-only line or even plain CRLF. And as long as the text originates from a web browser, there is always a CRLF. So the problem is pretty common even for carefully formated text.

h2. Fix

Attaching a patch with a fix and corresponding unit tests.

h2. Broader context

The current approach to section extraction is inherently fragile - as shown in the other (skipped) unit test enclosed in the patch. I'd suggest to keep the skipped test there to mark it as a known issue. Will create a dedicated issue for this.