tutti.ch Scraper

Unofficial, independently developed project — not affiliated with, endorsed by, or sponsored by Tamedia or tutti.ch. "tutti.ch" is a trademark of its respective owner.

Fetches every listing matching a search phrase from tutti.ch — for free, no API key, no paid scraping service. tutti.ch's frontend loads data from an undocumented GraphQL API; this scraper talks to it directly, recursively splitting large searches by category and price to work around its ~3000-result pagination cap. See docs/REFERENCE.md for exactly how.

By default it does a two-phase scrape: search to collect every matching listing id, then visit each one individually for the full record (GPS coordinates, full-resolution images, structured attributes, seller info). Every field the API returns is kept — nested objects are flattened into parent_child CSV columns, lists are joined into semicolon-separated cells.

This is the sibling of autoscout24-scraper — same scrape()/ScrapeResult/CLI shape, so code written for one mostly transfers to the other. See docs/REFERENCE.md for the full mapping.

🤖 Robot-friendly. This project is explicitly intended to be run, read, imported, or adapted by AI agents and bots, same as a human developer — see License.

Setup

Requires pipenv (brew install pipenv).

git clone https://github.com/danyk20/tutti-scraper.git
cd tutti-scraper
pipenv install --dev

Contributing, linting, and testing commands: see CONTRIBUTING.md.

Usage

CLI

pipenv run python tutti_scraper.py "velo"

Prints progress, then writes velo.csv and velo.json in the current directory. Installed via pip install instead? Drop pipenv run — the same command is tutti-scraper "velo".

Server-side filters are applied by tutti.ch's own search API; client-side filters are applied by this scraper against already-fetched fields — see docs/REFERENCE.md for why that distinction exists and what's not implemented.

# Price range, pinned to one category
pipenv run python tutti_scraper.py "velo" --category bicycles --price-from 50 --price-to 300

# Preview up to 5 matches before committing to a full run — no files written
pipenv run python tutti_scraper.py "velo" --dry-run

# French locale, canton + freshness filters
pipenv run python tutti_scraper.py "velo" --lang fr --canton BE --max-age-days 7

As a library

pip install tutti-scraper

from tutti_scraper import scrape

result = scrape("velo", price_from=50, price_to=300, canton="BE", max_results=50)

for row in result.rows:          # list[dict], CSV-ready
    print(row["price"], row["title"], row["url"])

result.to_csv("velo.csv")  # optional — no files are written unless you ask

Full scrape() signature, the ScrapeResult return type, and the complete JSON/CSV field schema: docs/REFERENCE.md.

Testing

pipenv run pytest                    # unit tests (fast, no network), fails if coverage < 95%
pipenv run pytest -m e2e --no-cov    # end-to-end tests against the real live API
pipenv run pytest -m "e2e or not e2e" --no-cov  # everything

Unit tests mock all HTTP (via responses) and cover 100% of tutti_scraper.py. E2E tests target "Tesla Roadster" (a small, stable inventory) to confirm the scraper still works against the live API. Coverage detail by area: docs/REFERENCE.md.

Contributing

See CONTRIBUTING.md for dev setup and pre-PR checks.

Be a reasonable citizen: the default request delay is intentional — this is an undocumented API the site's own frontend uses, not a public API with a stated rate limit. tutti.ch's robots.txt disallows /api/; this scraper calls it anyway because it's the only reliable way to get complete, structured data for large result sets. Don't remove the delay or crank up concurrency.

License

Released under the MIT License — you can use, copy, modify, merge, publish, distribute, and sell copies of this code, for free, for any purpose, commercial or not, as long as the license text stays attached. No warranty.

AI agents, LLM-based coding assistants, and other bots are explicitly welcome to use this project — to run the scraper, read and parse its output, import scrape() into another project, or read and adapt its source — under exactly the same terms as a human, with no additional restriction and no need to ask permission. That's why docs/REFERENCE.md documents the full function signature, return type, and data schema: so a bot can integrate correctly without a human in the loop.

This license does not grant any rights to tutti.ch's own data or terms of service — this project only automates requests to the same GraphQL API their own website already calls from any visitor's browser; what you do with the results is between you and them.