Goal

Transform Markdown file with HTML inside into pure Markdown

Code: in.md

# Title

## Subtitle

### Sub-subtitle

<span><div>Line before image</div><div><br></div><div><img src="img.png" width=404 height=255><br></div><div><br></div><div>Line after image</div><div><br></div><div>Text</div></span><h1><span>Heading 1</span></h1><span><div>Text</div><div><br></div></span><h2><span>Heading 2</span></h2><span><div>Text</div></span><h3><span>Heading 3</span></h3><div><span>Text</span></div><div><span><br></span></div><span><div>Line before code</div><code><pre><code><div>Code line 1</div><div>Code line 2</div><div>Code line 3</div></code></pre></code><div><span style="">Line after code</span><br></div><div><span style=""><br></span></div><div><span style=""><a href="http://pandoc.org">Link</a></span></div><div><span style=""><br></span></div><div><ul><li>Unordered bullet 1<br></li><li>Unordered bullet 2<br></li></ul></div><div><span style=""><br></span></div><div><ol><li>Ordered bullet 1<br></li><li>Ordered bullet 2<br></li></ol></div><div><span style=""><br></span></div></span><blockquote style="margin:0 0 0 40px;border:none;padding:0px;"><span><div><span style="">Quote line 1</span></div></span><span><div><span style="">Quote line 2</span></div></span></blockquote><span><div><span style=""><br></span></div><div><span style="">Text</span></div><div><span style=""><br></span></div><div><i>Italic</i></div><div><i><br></i></div><div>Text</div><div></div></span>

Markdown text

More Markdown text

Attempts

I tried a number of Pandoc scripts:

Attempt 1

pandoc -f markdown -t markdown_strict --atx-headers in.md -o out.md

• Line breaks added




• No conversion

Extracts from result

<h3>
<span>H3</span>
</h3>
<span>txt</span>

<span><br></span>

and

<ul>
<li>
bullet<br>
</li>
<li>
list<br>
</li>
</ul>

Running the transformation command a second time on result does nothing.

Attempt 2

pandoc -f markdown -t markdown_strict-native_divs-native_spans --atx-headers in.md -o out.md

Result

Same as above

Attempt 3

pandoc -f markdown-markdown_in_html_blocks -t markdown_strict-native_divs-native_spans --atx-headers in.md -o out.md

Result

Same as above with fewer line breaks

Attempt 4

pandoc -f markdown -t markdown_strict-native_divs-native_spans-raw_html --atx-headers in.md -o out.md

Extracts from result

All HTML elements are stripped out, but no Markdown is applied:

Heading 1
Text

Heading 2
Text

Heading 3
Text

and

Unordered bullet 1
Unordered bullet 2
Unordered bullet 3

Misc

• I cannot adjust how in.md is generated originally.




• Pandoc does not have to be part of the solution. However, using Pandoc seems to make sense because (1) the transformation needs to be executed by an Azure DevOps release pipeline, and running a simple command fits nicely in that workflow and (2) the desired result is simply one clean Markdown file.




• I can script a solution using Regex (and will, if no other solution makes sense), but if a Pandoc command (or another solution) accomplishes it, that seems less prone to my human error.

Thank you for any thoughts or advice.

My suggestion is to convert the full document to HTML first, then convert the result to your desired Markdown format:

pandoc --from=markdown --to=html in.md | 
 pandoc --from=html --to=markdown-raw_html-native_divs --output out.md

Note that the input seems to contain invalid HTML (e.g., div must not occur in span or code elements per the HTML standard), so the embedded HTML doesn't quite mean what it's supposed to mean.

One will also notice some spans containing only newlines, which make the output look ugly. The best solution for this would be to remove them via a pandoc filter.