open-rando

Train station-to-station hiking on French GR paths -- car-free hiking made easy.

A Python pipeline fetches hiking routes and train station data from open sources, computes walkable multi-step hikes between stations, and exports a catalog. An Astro static website displays hikes on a map with filters and GPX download.

How it works

BUILD TIME                                    RUNTIME (static files)
+-----------------------------------------+   +----------------------------+
| Python Pipeline                         |   | Astro Static Site          |
|                                         |   |                            |
| Overpass API --+                        |   |  Leaflet Map + List/Filter |
| OSM stations --+-> fetch -> match ->    |   |         |                  |
| SRTM tiles ---+   step graph -> DFS ->  |   |    catalog.json            |
|                    export               |   |    /gpx/{id}.gpx           |
|               data/catalog.json         |   |    /geojson/{id}.json      |
|               data/gpx/*.gpx            |   +----------------------------+
|               data/geojson/*.json       |
+-----------------------------------------+

The pipeline:

1. Fetches a GR hiking relation from the Overpass API (resolves super-relations recursively)
2. Fetches active train stations near the route (filters out disused/abandoned)
3. Matches stations to the trail using Shapely (keeps those within 5km)
4. Builds a step graph with edges between station pairs 8-18km apart
5. Finds all maximal hike chains via DFS, deduplicates sub-paths
6. Fetches nearby accommodation (hotel, guest house, hostel, camping) for each station
7. Exports GPX files, GeoJSON, and a catalog.json consumed by the website

Getting started

Prerequisites

• Python 3.13+ with uv
• Bun (or Node.js)

Run the pipeline

cd pipeline
uv sync --all-extras
uv run python -m open_rando

This fetches data from OpenStreetMap and writes output to data/.

Run the website

cd website
bun install
bun run dev

The dev server copies data/ into public/data/ automatically.

To build for production:

bun run build

Static output goes to website/dist/.

Features

• Interactive Leaflet map with trail visualization
• Filter hikes by distance, duration, step count, and max step length
• Filter by accommodation: show only hikes where every overnight stop has a hotel or camping
• Hike detail pages with step breakdown, station list, and dedicated map
• GPX download for each hike
• Only active train stations (disused/abandoned filtered out)
• Accommodation info (hotel/camping) displayed per station

Project structure

open-rando/
  pipeline/           Python data pipeline
    src/open_rando/
      cli.py           Entry point
      config.py        Constants (step distances, API config, search radius)
      models.py        Station, Accommodation, HikeStep, Hike dataclasses
      fetchers/        Overpass API, station fetcher, accommodation fetcher
      processors/      Station-to-trail matching, step graph + DFS
      exporters/       GPX, GeoJSON, catalog.json writers
  website/            Astro static site
    src/
      pages/           Index (map + filters + list) and hike detail
      components/      HikeMap, HikeCard, HikeFilters, HikeList
      lib/catalog.ts   TypeScript types + data loading
  data/               Pipeline output (gitignored)
  docs/               Architecture, data sources, deployment

Stack

Layer	Tech
Pipeline	Python 3.13+, Shapely, requests, gpxpy
Website	Astro, Leaflet, Tailwind CSS
Data sources	OpenStreetMap Overpass API, SNCF open data, SRTM elevation tiles

Data sources

All data comes from open sources under the ODbL license.

• Hiking routes: OpenStreetMap via Overpass API (route=hiking, ref~"^GR")
• Train stations: OpenStreetMap (railway=station / railway=halt), filtered for active service
• Accommodation: OpenStreetMap (tourism=hotel|guest_house|hostel|camp_site) within 2km of each station
• Elevation (planned): SRTM .hgt tiles

Development

# Lint and format (Python)
cd pipeline
uv run ruff check src/
uv run ruff format src/

# Type check
uv run mypy src/

License

Code and website: GPLv3

Data: ODbL (OpenStreetMap)