Paperclip

Local tool to capture web papers/articles, parse them into clean text + sections + references, and export in a few useful formats.

Quickstart

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
python run.py

Open:

• Library: http://127.0.0.1:8000/library/
• Collections: http://127.0.0.1:8000/collections/

Chrome extension

1. Open chrome://extensions
2. Enable Developer mode
3. Load unpacked → select extensions/chrome/

If your server isn’t on http://127.0.0.1:8000, edit extensions/chrome/background.js (API_ENDPOINT).

How it works

• Extension posts captures to POST /api/captures/ (URL + full HTML + best-effort main content + metadata).
• Server parses with a site-aware parser (PMC/OUP/Wiley/…) and falls back to generic heuristics.
• Data is stored locally:
  • SQLite: data/db.sqlite3
  • Artifacts: data/artifacts/<capture_id>/

Artifacts

Each capture has a folder at data/artifacts/<capture_id>/, typically containing:

• page.html, content.html
• article.json, reduced.json
• sections.json, references.json
• paper.md (deterministic bundle)

Exports

• BibTeX: /exports/bibtex/
• RIS: /exports/ris/
• Master Markdown: /exports/master.md/
• Papers JSONL: /exports/papers.jsonl/

Add ?collection=<collection_id> to export a specific collection.

Dev

pip install -r requirements-dev.txt
pytest