Regression when image is in a `Shape Format`

[fsoedjede]

Explain the problem.

Since pandoc 3.1.10, when a image is inside a Shape Format, the text is extracted but the image is not.

Source file: image-issue.docx

1. With pandoc 3.1.9, the file produce:

   
   *Lorem*

{width="5.10625in" height="3.2333333333333334in"}L'espace entre l'homme et la nature

2. With `pandoc 3.1.10`, the file produce:
```markdown
*Lorem*

L'espace entre l'homme et la nature

Homme

Nature

The PR which changes that is: https://github.com/jgm/pandoc/pull/9223 and issue is https://github.com/jgm/pandoc/issues/9214.

I detected the bug while trying to upgrade from pandoc 3.1.8 to pandoc 3.1.12.3. As my tests failed, I've installed intermediary version and been able to find that the issue occurres when passing from pandoc 3.1.9 to pandoc 3.1.10.

In Word, as the user, I can change the content to avoid having shape format around images but I think the image should be kept anyway

Proposed solution: Keep the image along with the text from the Shape Format Example (the goal is to have the content kept, even if it's not well formatted):

*Lorem*

Homme

Nature
![](media/image1.png){width="5.10625in"
height="3.2333333333333334in"}L'espace entre l'homme et la nature

Pandoc version? What version of pandoc are you using, on what OS? (If it's not the latest release, please try with the latest release before reporting the issue.)

$ pandoc -v
pandoc 3.1.10
Features: +server +lua
Scripting engine: Lua 5.4
User data directory: /Users/me/.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.

[jgm]

@StephanMeijer since you made PR #9223, maybe you have a sense for why it affected this and what would be needed to restore original behavior in this case?

[StephanMeijer]

@jgm Yes, I do. Let me type out a case soon.