How to collect info from many chapters then dump at end

[pragdave]

I'm creating a book with exercises. Many exercises have answers. I want to collect all the answers and put them into their own appendix, cross linked with the questions.

I have this working fine for PDF output. I have a pre_quarto filter that matches div.answer and moves them into their own list.

However, it doesn't work for HTML. The problem is that filters are loaded just once for a PDF documents, but they seem to be loaded for each chapter for HTML. This means that I don't seem to be able to traverse the entire document in a filter. I've confirmed that the .lua files are reinitialized for each chapter (see below).

Is there a trick for maintaining document-global state for lua filters when the target is HTML? Am I reinventing the wheel?

Cheers

Dave

---

To show that filters are reloaded for each chapter:

# _quarto.yml
project:
  type: book

book:
  chapters:
    - index.qmd
    - one.qmd
    - two.qmd
format:
  html:
    filters:
      - report_when_loaded.lua

-- report_when_loaded.lua
quarto.log.output("Loading filter")

And the chapters are all just # some title

Run this:

$ quarto render
[1/3] index.qmd
Loading filter
[2/3] one.qmd
Loading filter
[3/3] two.qmd
Loading filter

Output created: _book/index.html

(source attached)

reload.zip