n8n PDF Purchase Order Parsing

I already let my users upload supplier purchase orders (always in PDF) from within my SaaS app. What I now need is a complete n8n workflow that intercepts that upload, extracts every line of the purchase order, and returns a clean JSON payload I can consume to pre-fill my internal “commande” object.

The JSON must contain, for each line of the order:
• Quantité
• Prix unitaire
• Description du produit

In addition, the header-level data I must receive are:
• Nom du fournisseur
• Numéro du bon de commande
• Date de la commande

I’m open to whichever OCR or parsing nodes you feel are most reliable (native PDF parsers, Tesseract, Google Vision, etc.), as long as the result is accurate and the flow remains self-hostable inside n8n. Please wire everything so the file comes in through my existing webhook, gets parsed, and the final JSON is returned to my app via HTTP response (or a simple POST to my API endpoint—whichever is cleaner).

Acceptance will be based on:
1. A shareable n8n workflow file (.json) I can import.
2. Clear mapping of each required field.
3. Successful end-to-end test with at least two real purchase-order PDFs showing correct values in my app.