Project Highlights pipeline

Turns the PowerPoint decks in files/ProjectHighlightsSource/ into the thumbnails, PDFs, and data that back the /highlights page.

Adding a new highlight slide

Drop the .pptx into files/ProjectHighlightsSource/. It should be a single slide whose speaker notes hold the talking points – that’s what this pipeline expects (only slide 1 of each file is processed).
Run:
```
python3 scripts/build_highlights.py
```
This renders a thumbnail (files/highlights/<slug>/thumbnail.png) and a full-size PDF (files/highlights/<slug>/slide.pdf), and adds/updates a matching entry in _data/highlights.yml.
Open _data/highlights.yml, find the new entry (its caption starts with TODO:), and write a short, plainly-factual caption. Keep the tone even-handed – this is public-facing text, and the talking points in PowerPoint speaker notes are written for a live audience, not a general web audience. In particular:
- Avoid absolute claims of credit (“critical,” “essential,” “the only”).
- Credit the broader community/consortium (CASS, HPSF, the Spack/E4S communities, etc.) rather than framing outcomes as PESO’s alone.
- Keep specific numbers (benchmarks, growth stats) but caveat one-off measurements (e.g., “in one test on a laptop…”) rather than stating them as general guarantees.
Commit the .pptx, the new files/highlights/<slug>/ folder, and the updated _data/highlights.yml.

Re-running the script is always safe: it regenerates thumbnails/PDFs from the source decks but never overwrites a title or caption that’s already been hand-edited, so you can re-run it after touching up a slide’s artwork without losing caption edits.

One-time setup (macOS)

brew install --cask libreoffice   # gives `soffice`, used for .pptx -> PDF
brew install poppler              # gives `pdftoppm`, used for PDF -> PNG
pip3 install python-pptx pyyaml

sips (thumbnail resizing) ships with macOS already.