Making DOI Ready PDFs from Discourse

This is all helpful. I just blogged about the use case I am after here:

The challenge I’ve now seen from using this is several fold.

It does have to be a static file that can be uploaded to Zenodo. The “promise” behind DOIs is that they will always be accessible in perpetuity, and immutable. Uploading to Zenodo guarrantees this, but we have to upload. It need not be a PDF. A well packaged HTML file could work too, but PDF’s are viewable on the website. The key thing is that it must be static.

That would be a great feature. Though it seems downstream from some other problems. Here is what I see are the big issues:

  1. Links are not handled well, nor are crosslinks.
  2. Most of the problems are hackishly solvable with custom css added to the template.
  3. Links should ideally be handled as footnotes, but this is not easy to do with standard CSS. I should be done on the backend.
  4. It is currently untested on things like “spoiler” tags or Latex.
  5. There is a ton of nuisance elements that need to be removed (such as cross link indicators after posts). If someone wants to help with this, I can produce a definitive list.
  6. Process is pretty manual. Ideally there is an Admin button one could click which would interface directly with the Zenodo API, including getting the versioning correctly, etc. This admittedly is a lot of work. So I imagine will take a while.

I can provide the CSS I am using if someone wants to make it default. All it does is add some CSS in print and page css sections to clean things up for print. That is probably the easiest thing to do right now, perhaps adding some configuration options to discourse to tune it. Is there interest in that?

The HTML link to footnote is a real problem that does need to be addressed. It is most likely solvable only on the backend, as CSS is not set up for this. That, it seems, it the most important hard problem to solve quickly.

Is this something anyone can help with?

3 Likes