Inconsistent heading levels with AsciiDoc output

[jeffmelville]

When converting fragments to AsciiDoc, heading levels are not converted consistently. It appears that the lowest level heading always gets converted to === in AsiiDoc. See example below. In both examples, I would expect ### third to convert to === third. The issue occurs when exporting with nbconvert, which executes pandoc on one Jupyter cell at a time.

Also see: jupyter/nbconvert#1456

➜  ~ pandoc --version
pandoc 3.1.9
Features: +server +lua
Scripting engine: Lua 5.4
User data directory: /Users/jmelville/.local/share/pandoc
Copyright (C) 2006-2023 John MacFarlane. Web: https://pandoc.org
This is free software; see the source for copying conditions. There is no
warranty, not even for merchantability or fitness for a particular purpose.
➜  ~ echo -e '### third\n' | pandoc -t asciidoc
== third
➜  ~ echo -e '## second\n### third\n' | pandoc -t asciidoc
== second

=== third

Pandoc version?

3.1.9 on MacOS
I also re-ran on 3.3 to confirm.

[jgm]

Asciidoc sections are not supposed to have gaps, e.g.

== one

==== four

If you try this you'll get a warning like

asciidoc: WARNING: my.adoc: line 3: section title out of sequence: expected level 3, got level 4

So, we normalize them before writing. This is by design.

Note: it seems that asciidoc (original) also required the first heading level to be ==, but asciidoctor does not. So we could perhaps change this behavior.

[jeffmelville]

Ah - I guess I see how it gets that way. There aren't gaps in the test vectors per se, but since nbconvert calls pandoc individually with each cell, it's possible for each call to pandoc to not have the right context. Something like this:

Cell 1 (Markdown):

## Some section title

** Cell 2** (Code):

IPython.display.display_image(image_bytes)

Cell 3 (Markdown)

### Subsection Title

#### Sub-subsection Title

will get converted as

== Some section title

image::image_path.png[]

== Subsection Title       # <-  This is one level too low

=== Sub-subsection Title   # <- This is also one level too low, but correctly +1 from its parent in the cell

There another issue with nbconvert and AsciiDoc output (jupyter/nbconvert#2152) but I'm working around that one within nbconvert.

Its hard for me to figure out how to correctly use pandoc AsciiDoc output with Markdown fragments given the current behavior. If I export the whole notebook as Markdown first, I can use pandoc to convert that to AsciiDoc but I lose some other information.

[jgm]

Due to changes in makeSections, pandoc no longer ensures "gaplessness" in the sections anyway. Since asciidoc has changed in caring about the bottom level, I'm just going to remove that feature and have it use whatever level was in the original...