eperx
Fiat group parts catalogue — Fiat, Lancia, Alfa Romeo, Abarth, LCV, Chrysler
About
A reimplementation of ePER 8.3 (edition 83, data release 4147), reverse-engineered from the shipped Java application and its data files. It reads the original ePER data client-side with no backend, and ships without data.
No warranty
The catalogue data is Fiat's, is not redistributed here, and nothing in the repository is derived from it. Which parts fit which vehicle is reconstructed and never verified against a real car.
Getting data in
A fresh visit lands on the source picker, because there is nothing to show until you supply some. Two of the options need nothing but a folder.
| Source | What it does |
|---|---|
| Over HTTP | Any static host honouring Range — eperx serve, or a URL. Nothing stored in the browser. |
| A folder on this machine | Read in place, nothing copied. A service worker answers the reads, so 5.7 GB of drawings stay put. Chromium only. |
| Stored in this browser | Copied into the origin private file system: opens on reload with no prompt and no disc mounted. |
| Import a disc | Build the tree in the browser from a mounted disc — no CLI at all. |
A tree can be opened straight from a link (?data=<url>), so a link handed to a colleague sets their source up rather than working once. The choice is remembered.
What works
- Browsing — marque → model → catalogue → group → subgroup → drawing, with the diagram, its numbered callout list, part-number search, and every drawing a part appears on
- Parts by vehicle, chosen by version or by VIN. Drawings and callouts are scored fits / does not fit / not determined — a 1.3 JTD Panda narrows a subgroup from 13 drawings to 3
- VIN lookup — a chassis number resolves to its exact sold version, engine number, build date, and the options and characteristics it left the factory with
- Importing without the CLI — the Settings → Data wizard reads a mounted disc and writes the tree into browser storage, same importer and same output as the CLI
- A parts bin, keyed by part number, with provenance recorded and a CSV export
Stack: TypeScript, Svelte, csfs, pnpm + Turborepo