How to include a figure in PDF format?

[roaldarbol]

Description

When including a PDF as an image, which is quite common for academic papers, it renders as a scroll box with little height. The basic example could look like this (I have also attempted to add e.g. {height=100%} to no avail):

:::{#fig-myfigure}

![](figure-1.pdf)

My caption
:::

...which renders like this:

I guess the underlying issue is that it is inserted as in a PDF reader; I would expect it to render as a normal figure when it is inside a #fig div. Is there anyway to make it so?
Thanks in advance!

[mcanouil]

Please provide a fully reproducible example not a snippet out of context.

I will guess, the format is HTML.
I'm afraid a PDF is not an image, it's a binary format, thus it requires third party to be read which leads Pandoc to use iframe in HTML.
That's beyond Quarto or Pandoc control, that's how HTML works.

[roaldarbol]

It is indeed HTML. MyST converts PDFs to images with imagemagick before rendering, so I imagined a similar functionality to be present in Quarto, since many journals recommend submitting figures in PDF (e.g. nature, science, journal of experimental biology) hence the question. Thanks for the answer. :-)

[mcanouil]

Quarto does not depend on imagemagick and will very unlikely do so.
You can make a Lua filter to do that if you want.

PS: I am perfectly aware of journals guidelines (I spent more than 10 years in academia).

[roaldarbol]

Oh, that wasn't meant to sound like I questioned you - apologies of it sounded that way; it was just to ensure context, just as you explained PDF format although I've also spent 10 years in academia and know what it is - doesn't mean I didn't think it was relevant information. :-)

Fair game regarding the imagemagick dependency, I may have a look at how to make a Lua filter then.

[mcanouil]

Oh, that wasn't meant to sound like I questioned you - apologies of it sounded that way; it was just to ensure context, just as you explained PDF format although I've also spent 10 years in academia and know what it is - doesn't mean I didn't think it was relevant information. :-)

All is fine, I was only giving context on my background ;)

Fair game regarding the imagemagick dependency, I may have a look at how to make a Lua filter then.

See https://github.com/search?q=lua%20imagemagick&type=repositories to get you started.

https://quarto.org/docs/extensions/filters.html

[roaldarbol]

Thanks for the links! I'll report back if I get around to making one. ;)

[roaldarbol]

Ended up creating a Lua filter which converts PDF files to .webp format with some help from ChatGPT, which does this quite nicely. You can tweak the imagemagick settings a to your liking.

pdf-to-webp.lua, which I've for now just placed in the base of my workspace:

1. Converts .pdf to .webp in the source folder.
2. Except if a webp file is already generated
3. Checks whether the source .pdf file has changed, in which case it re-converts

---

EDIT: I've made the filter into an extension which can be found here: https://github.com/roaldarbol/pdf-to-image

---

-- Lua filter for Quarto to convert PDF images to WebP only in HTML output

-- Check if the current output format is HTML
local is_html = (quarto and quarto.doc.is_format("html"))

-- Function to check if a file exists
local function file_exists(name)
  local f = io.open(name, "r")
  if f then f:close() end
  return f ~= nil
end

-- Function to get the modification time of a file using a shell command
local function file_mod_time(name)
  local handle = io.popen('stat -f "%m" "' .. name .. '"') -- macOS or BSD
  local result = handle:read("*a")
  handle:close()
  
  return tonumber(result)
end

function Image(elem)
  -- Only proceed if output format is HTML and the image is a PDF
  if is_html and elem.src:match("%.pdf$") then
    -- Define new filename for the WebP version
    local webp_src = elem.src:gsub("%.pdf$", ".webp")

    -- Check if WebP file exists and compare modification times
    local pdf_time = file_mod_time(elem.src)
    local webp_time = file_exists(webp_src) and file_mod_time(webp_src)

    -- If WebP doesn't exist or the PDF is newer, convert PDF to WebP
    if not webp_time or (pdf_time and pdf_time > webp_time) then
      -- Run the ImageMagick command to convert PDF to WebP
      os.execute(string.format('magick -density 600 -colorspace RGB "%s" "%s"', elem.src, webp_src))
    end

    -- Update image source to the new or existing WebP file
    elem.src = webp_src
  end

  return elem
end

And added it in _quarto.yml:

format:
  html:
    filters:
      - pdf-to-webp.lua

[bastienchassagnol]

Could this feature, @roaldarbol and @mcanouil, could it be considered integrated within quarto, as a quarto extension? Since there ar emore and more image files avalaible on the web as a webp extension?

[mcanouil]

I can'ttell.
But this filter could at least be shared as a Quarto extension at first.

You can see extensions as some kind of incubator. If an extension is not used, integrating it makes not much sense.

[bastienchassagnol]

Actually, that was the exact reciprocal problem I had, namely converting webp images to pdf or png (any recommendations on the favoured format), when rendering pdf or docx outputs. I dealt (partially) with that issue also creating my own (local only) extension for converting webp images using magick convert:

function Image(img)
  if img.src:match("%.webp$") then
    local png_src = img.src:gsub("%.webp$", ".png")
    os.execute("magick convert " .. img.src .. " " .. png_src)
    img.src = png_src
  end
  return img
end

Could I consider tailor and integrate this code to yours, @roaldarbol, in order to create a more generic quarto extension for automatically handling distinct kinds of image formats depending on the output format?

[mcanouil]

Be aware that you script might not work across OS, so you should be sure your filter handle this well.

Note that the author edited his post to show that the filter is available as a Quarto extension: https://github.com/roaldarbol/pdf-to-image (which you can also find on https://m.canouil.dev/quarto-extensions/)

[roaldarbol]

Hi @bastienchassagnol!
As @mcanouil mentioned, the filter is already available as an extension. I won't have time to develop it into a more general extension - however, feel free to play around with it and make a PR if you want! :-) You're also free to make make a new extension out of it if that is easier. :-)