pandoc for the people error

[cmadland]

I very regularly encounter this error. Most recently trying to convert from .docx to .pdf

Does this have to do with permissions on the docx file?

'Conversion failed: Typst: failed to load file (access denied)'

[jgm]

I have not encountered this issue. Is it a reproducible issue (same file always gives the error) or intermittent? If I could reproduce it I might be able to figure out the problem.

[cmadland]

Conference Research Paper Template 2022.10.06-OTH-OTH (2).docx

Doesn't seem to be any particular file. This one threw the error for me.

Version: 7.9.3970.50 (Official Build) (arm64)
Chromium Version: 146.0.7680.189
Channel: Official Build
Platform / OS: macOS Version 14.8 (Build 23J21)
Executable Path: /Applications/Vivaldi.app/Contents/MacOS/Vivaldi
Command Line:  /Applications/Vivaldi.app/Contents/MacOS/Vivaldi --flag-switches-begin --flag-switches-end --disable-smooth-scrolling

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/146.0.0.0 Safari/537.36
Revision: 565d1eea6aaa255888b366d41c39d54dd40f286b
Profile Path: /Users/colin.madland/Library/Application Support/Vivaldi/Default
JavaScript: V8 14.6.202.33

[jgm]

This file works fine when I try it. What options are you using for the conversion?
(Target format, other things you've selected...)
Also, try opening the JavaScript console (developer tools) and see if any relevant error messages appear.

[cmadland]

Thanks for looking into this.

Trying to convert from docx to pdf with default settings, although it happened earlier with other file types (epub to pdf although I was able to convert epub to LaTeX, then LaTeX to pdf).

From the console:

v8c78df7c7c0f484497ecbca7046644da1771523124516:1  Failed to load resource: net::ERR_BLOCKED_BY_CLIENT
index.js?sha1=c1d3469c4f0f95728b0e4cb0beae500ac87d32a0:492 Pandoc converter initialized (pandoc.wasm loading in background)
index.js:3150 using deprecated parameters for the initialization function; pass a single object instead
Qg @ index.js:3150
petite-vue:1 TypeError: Cannot read properties of undefined (reading 'size')
    at eval (eval at Le (petite-vue:1:8640), <anonymous>:3:51)
    at Be (petite-vue:1:8587)
    at Me (petite-vue:1:8511)
    at get (petite-vue:1:13856)
    at W.fn (petite-vue:1:8110)
    at W.run (petite-vue:1:2309)
    at me (petite-vue:1:6108)
Be @ petite-vue:1
petite-vue:1 TypeError: Cannot read properties of undefined (reading 'size')
    at eval (eval at Le (petite-vue:1:8640), <anonymous>:3:51)
    at Be (petite-vue:1:8587)
    at Me (petite-vue:1:8511)
    at get (petite-vue:1:13856)
    at W.fn (petite-vue:1:8110)
    at W.run (petite-vue:1:2309)
    at me (petite-vue:1:6108)
Be @ petite-vue:1
petite-vue:1 TypeError: Cannot read properties of undefined (reading 'size')
    at eval (eval at Le (petite-vue:1:8640), <anonymous>:3:51)
    at Be (petite-vue:1:8587)
    at Me (petite-vue:1:8511)
    at get (petite-vue:1:13856)
    at W.fn (petite-vue:1:8110)
    at W.run (petite-vue:1:2309)
    at me (petite-vue:1:6108)
Be @ petite-vue:1
petite-vue:1 TypeError: Cannot read properties of undefined (reading 'size')
    at eval (eval at Le (petite-vue:1:8640), <anonymous>:3:51)
    at Be (petite-vue:1:8587)
    at Me (petite-vue:1:8511)
    at get (petite-vue:1:13856)
    at W.fn (petite-vue:1:8110)
    at W.run (petite-vue:1:2309)
    at me (petite-vue:1:6108)
Be @ petite-vue:1
petite-vue:1 TypeError: Cannot read properties of undefined (reading 'size')
    at eval (eval at Le (petite-vue:1:8640), <anonymous>:3:51)
    at Be (petite-vue:1:8587)
    at Me (petite-vue:1:8511)
    at get (petite-vue:1:13856)
    at W.fn (petite-vue:1:8110)
    at W.run (petite-vue:1:2309)
    at me (petite-vue:1:6108)
Be @ petite-vue:1

pandoc.org-1776269102399.log

[jgm]

That console output is irrelevant, I think.
But now that I know you're going to PDF, it makes more sense. I am guessing that this docx has images in it?
In going to PDF via typst, pandoc will try to save the images from the docx in a temporary file and then refer to them in the typst document. But this isn't possible in the wasm sandbox we're using.

[jgm]

Relevant issue: #10801
If we supported this, we wouldn't need to write the images to external files, and it would work.

[cmadland]

I see, that makes sense. Thank you for your help and your work on this.

[jgm]

In fact, if you use @tarleb's filter from #10801, it works!
Save as embed_images.lua:

function Pandoc (doc)
  doc = pandoc.mediabag.fill(doc)
  return doc:walk {
    Image = function (img)
      local mimetype, contents = pandoc.mediabag.lookup(img.src)
      if mimetype and contents then
        img.src = pandoc.mediabag.make_data_uri(mimetype, contents)
        return img
      end
    end
  }
end

Then click on the Filters tab in the pandoc app and upload this as a Lua filter.
Your docx -> pdf conversion should then work.

[jgm]

We should make this happen automatically on WASM, perhaps.

[cmadland]

will give that a shot

[cmadland]

YATTA! The lua filter worked!

[jgm]

I modified the web app so that the filter is automatically injected for pdf-via-typst. So, it should now "just work."