Is there a way to generate HTML and PDFs with selectable text in pandoc?

In case anyone is also trying to do this, I was able to make the text selectable in HTML sites by writing a pandoc filter that uses the object tag instead of the img tag. I wrote it in Haskell, but it can probably be easily converted to lua or python.

#!/usr/bin/env runhaskell

-- svg.hs
--    Use the `<object>` tag instead of `<img>` tag for svgs. Using `<object>`
--    allows for interactive SVGs.
--
-- Do `cabal install --lib pandoc-types` to install the pandoc Haskell module.
--
-- Compile with `ghc -package pandoc -package text svg.hs`
--
-- Then `pandoc --filter ./svg <rest of options>`

{-# LANGUAGE OverloadedStrings #-}

import Data.Text
import Text.Pandoc.JSON
import Text.Pandoc.Shared

imgToObject :: Inline -> Inline
imgToObject (Image attr alt (src, title))
    | ".svg" `isSuffixOf` src = 
        RawInline "html" $
            "<object data=\"" <> src <> "\" type=\"image/svg+xml\" title=\"" <> title <> "\">"
            <> (stringify alt)
            <> "</object>"
    | otherwise = (Image attr alt (src, title))

imgToObject x = x

main :: IO ()
main = toJSONFilter imgToObject

As for the PDF output, pandoc uses the \includesvg from the svg package. While looking through the documentation, I saw an example of exactly what I was trying to do (see https://ctan.org/pkg/svg?lang=en Section 4. Example). Sure enough, using their example svg, the text is selectable in the output pdf. This package looks to have been made with Inkscape in mind, but I was able to make an svg diagram in Inkscape which I exported as a plain svg (not an Inkscape svg) and it still had selectable text. I think might just be a draw.io issue.