Export MathML in ODT as LaTeX

[memeplex]

I have created the attached and zipped odt by exporting it from google docs. The text is just "Hi x^2" where x^2 is an squared x as an equation. Converting it to markdown using pandoc gives:

pandoc -t markdown Prueba.odt 
Hi ![](./ObjectReplacements/Object 2){width="0.1417in"

Inside the odt Object 2/content.xml is:

cat Object\ 2/content.xml 
<?xml version="1.0" encoding="UTF-8"?>
<math xmlns="http://www.w3.org/1998/Math/MathML" display="block"><semantics><msup><mi>x</mi><mn>2</mn></msup><annotation encoding="StarMath 5.0">{x} ^ {2}</annotation></semantics></math>

that is, the equation was exported as MathML, not as an image. Now, I would have expected this to be exported as LaTeX in the markdown output.

Am I missing something?

Pandoc version: 2.2.1

Prueba.zip

[memeplex]

BTW, this is the output when exporting as docx:

pandoc -t markdown Prueba.docx 
Hi $x^{2}$.

Much better!

[jgm]

I don't think the Odt reader currently parses math elements.
This would be a very useful addition, and it shouldn't be hard -- we already convert mathml to tex in reading docbook, for example.

Note, however, that we'd only be able to parse presentation mathml, not semantic mathml (you use the 'semantics' tag here).

[memeplex]

I barely understand mathml, let alone what semantic/presentation tags are. I can say that it works fine with docx though, so if you're doing the "presentation" stuff there I assume it should work here too.

[memeplex]

Maybe I should have mentioned that equations are represented as mathml in docx too. I have verified this. I don't know about the semantic/presentation thing though.

[mb21]

Might work with #5606...

[jgm]

@memeplex No, docx doesn't use mathml. It uses a different, XML-based math format. texmath and pandoc can convert between these.

[memeplex]

Thanks!

[jgm]

Sorry, it was claimed in the PR that it fixed this issue, but it doesn't.
@BImage was your PR supposed to handle this "object replacement" stuff?
The other issue is math parsing; I don't believe we currently parse mathml in the odt reader, though we could.