Convert HTML iframes to links

[marph91]

I want to convert HTML iframes to links (and finally convert the document to Markdown). My approach was to use the input format html+raw_html to get the iframe in AST and convert the iframe by a lua filter (like echo '<iframe src="https://google.com" />' | pandoc -f html+raw_html -t markdown_strict --lua-filter iframes_to_links.lua. The filter is not written yet, but I believe it should work.

However, in another part of the code, I rely on the SVG parsing. SVG tags are converted to Image in AST. This breaks when using +html.

Is there a way to get the iframes in AST and keep the SVG-to-Image-conversion at the same time? Or would you suggest another approach?

[jgm]

You could keep +raw_html with the result that SVGs become raw blocks, and then use a filter to identify the SVG raw blocks and run them through the HTML reader (with raw_html disabled) to produce an Image?

[marph91]

$ echo '<iframe src="https://google.com" /> <svg />' | pandoc -f html+raw_html -t html --lua-filter iframes_to_links.lua | pandoc -f html -t markdown_strict
<https://google.com> ![](data:image/svg+xml;base64,PHN2Zz4KPC9zdmc+)

iframes_to_links.lua:

function RawBlock(el)
    if el.text:match("^<iframe") ~= nil then
        local iframe_src = el.text:match('^<iframe.*src="?([^> "]*)')
        if not iframe_src then
            return {}
        end

        local iframe_title = el.text:match('^<iframe.*title="?([^"> ]*)')
        if not iframe_title then
            iframe_title = iframe_src
        end

        return pandoc.Plain(pandoc.Link(pandoc.Str(iframe_title), iframe_src))
    end

    if el.text:match("^</iframe") ~= nil then
        return {}
    end

    return el
end

Thanks, the snippet above yields the expected result. But I wonder if it's possible to avoid the second call to Pandoc. I. e. Is it possible to apply the HTML reader directly on the SVG raw blocks (inside the lua filter)?

[jgm]

Are you looking for this?
https://pandoc.org/lua-filters.html#pandoc.read

[marph91]

Yes, this works. Thanks!

FTR, I added this snippet:

    if el.text:match("^<svg") ~= nil then
        local document = pandoc.read(el.text, "html")
        return document.blocks[1]
    end
    if el.text:match("^</svg") ~= nil then
        return {}
    end