AutoScout24 Scraper · CH

Overview

This scraper extracts vehicle listings from AutoScout24 Switzerland (autoscout24.ch) — a separate product operated by Auto Scout 24 Schweiz AG (part of Swiss Marketplace Group). It is not the same backend as .de / .com and other AutoScout24 EU markets — those are operated by a different company and need a different scraper.

Looking for listings from any other AutoScout24 market (.de / .com / .it / .es / .fr / .be / .nl / .at / .lu)? Use the sibling actor: AutoScout24 Scraper · DE · AT · IT · ES · FR · BE · NL · LU · .com.

The scraper talks to the public api.autoscout24.ch REST API:

Endpoint	Method	Purpose
POST /v1/listings/search?language={de|fr|it|en}	POST	Paginated vehicle search (max page size 20)
GET /v1/listings/{id}?language={...}	GET	Full structured detail JSON for one listing

No authentication needed. The API host is not behind Cloudflare (unlike the public www.autoscout24.ch HTML pages), so a proxy is not strictly required.

Features

• Filter mode — paginate through the entire car catalog and pull matching listings.
• Detail-URL mode — paste one or more https://www.autoscout24.ch/{lang}/d/{id} URLs.
• Optional detail enrichment — fetches the JSON detail endpoint for each match (extra fields like cylinders, doors, drive type, full consumption breakdown).
• All four UI languages: de, fr, it, en.
• Five vehicle categories: car, camper, truck, motorcycle, trailer.

Server-side filtering

The POST /v1/listings/search endpoint honours all the major filters server-side, using the same shape the website's own SSR uses (extracted from initialSearchState in a Wayback Machine snapshot):

Filter	Field in request body	Notes
Make / model	makeModelVersions: [{ makeKey, modelKey }]	Nested — not top-level makeKey
Fuel	fuelTypes: [string]	Plural array; singular fuelType is silently ignored
Body	bodyTypes: [string]	Plural array
Condition	conditionTypes: [string]	Plural array
Price (CHF)	priceFrom, priceTo	Int
Year	firstRegistrationYearFrom, firstRegistrationYearTo	Int
Mileage (km)	mileageFrom, mileageTo	Int
Power (HP)	horsePowerFrom, horsePowerTo	Int — not powerFrom/powerTo
Vehicle category	vehicleCategories: [string]	Plural array
Sort	sort: [{ type, order, variant: 'v1' }]	Server-side

Result: a query that matches 341 listings on the website also returns 341 from the API (no over-fetching). A matchesClientSideFilters pass still runs as a defense-in-depth safety net but is normally a no-op.

Still client-side only (server-side shape not yet confirmed)

• ZIP code + radius (zipCode / radiusKm) — sent but silently ignored
• Warranty / accident (hasWarranty / hasAccidentHistory) — sent but ignored

If you set these, the actor pages through results and applies them client-side. Send an example URL that uses one of these filters and we can extend support.

Input

Provide either startUrls (detail URLs) or filter fields. URLs always win when both are set.

Filter-mode example

{
  "filterLanguage": "de",
  "filterVehicleCategory": "car",
  "filterMake": "bmw",
  "filterModel": "x1",
  "filterPriceFromCHF": 20000,
  "filterPriceToCHF": 50000,
  "filterYearFrom": 2020,
  "filterFuelType": "hybrid",
  "sortType": "RELEVANCE",
  "sortOrder": "DESC",
  "enrichDetails": false,
  "maxItems": 100,
  "proxy": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Detail-URL example

You can paste full URLs or bare numeric ids — both are accepted in the same startUrls array.

{
  "filterLanguage": "de",
  "startUrls": [
    { "url": "https://www.autoscout24.ch/de/d/20458677" },
    { "url": "https://www.autoscout24.ch/fr/d/20462882" },
    { "url": "20464837" }
  ],
  "maxItems": 100
}

For bare ids the language defaults to filterLanguage (which itself defaults to de).

Input field reference

Field	Type	Description
startUrls	[{ url }]	Detail-page URLs (/de/d/{id}, /fr/d/{id}, …) or bare numeric ids ("12345678"). Search/list URLs are not parsed — use filters instead.
filterLanguage	de|fr|it|en	Response locale & detail-page URL prefix. Default de.
filterVehicleCategory	car|camper|truck|motorcycle|trailer	Works server-side. Default car.
filterMake / filterModel	string	Make/model key (lowercase, hyphen-separated; e.g. bmw, mercedes-benz, x1). Applied client-side.
filterConditionType	new|pre-registered|demonstration|used	Client-side.
filterFuelType	electric|petrol|diesel|hybrid|mhev-petrol|mhev-diesel|plugin-hybrid	Client-side.
filterBodyType	suv|saloon|estate|coupe|cabriolet|small-car|pickup|minivan|bus	Client-side.
filterPriceFromCHF / filterPriceToCHF	int	CHF, inclusive. Client-side.
filterYearFrom / filterYearTo	int	First-registration year, inclusive. Client-side.
filterMileageFrom / filterMileageTo	int	km, inclusive. Client-side.
filterPowerFromHP / filterPowerToHP	int	HP, inclusive. Client-side.
filterZipCode / filterRadiusKm	string + int	Sent to the API; not verified client-side (no geo lookup).
sortType	RELEVANCE|PRICE|MILEAGE|YEAR|POWER	Works server-side. Default RELEVANCE.
sortOrder	DESC|ASC	Default DESC.
enrichDetails	bool	When true, fetch /v1/listings/{id} per result for extra fields. Doubles request count.
maxItems	int	Hard cap on output rows. Default 1000. Non-paying users are capped at 60.
maxConcurrency / minConcurrency / maxRequestRetries	int	Crawlee tuning knobs.
proxy	object	Apify proxy config. Residential recommended only for very heavy runs.

Output

Each pushed dataset row has this shape (selected fields shown):

{
  "id": 20458677,
  "url": "https://www.autoscout24.ch/de/d/20458677",
  "language": "de",
  "title": "M2 Coupé",
  "price": 63890,
  "priceCurrency": "CHF",
  "make": "BMW", "makeKey": "bmw", "makeId": 9,
  "model": "M2", "modelKey": "m2", "modelId": 999,
  "firstRegistrationDate": "2024-…",
  "firstRegistrationYear": 2024,
  "mileage": 28900,
  "fuelType": "petrol",
  "transmissionType": "automatic",
  "bodyType": "coupe",
  "conditionType": "used",
  "powerHp": 460,
  "powerKw": 338,
  "consumptionCombined": 9.6,
  "drive": "rear",
  "hadAccident": false,
  "hasWarranty": true,
  "warrantyType": "from-delivery",
  "sellerId": 123456, "sellerName": "…", "sellerType": "professional",
  "sellerCity": "Zürich", "sellerZipCode": "8001",
  "imageCount": 17,
  "images": ["https://listing-images.autoscout24.ch/listing/…", "…"],
  "source": "list",         // or "detail" when enrichDetails=true
  "raw": { "...": "full listing object as returned by the API" }
}

Architecture

• src/main.ts — input validation, dual-mode resolution, crawler setup. Free-user clamp + ACTOR_MAX_PAID_DATASET_ITEMS respected.
• src/routes.ts — two handlers:
  • LIST — POSTs to /v1/listings/search, applies matchesClientSideFilters, pushes rows, enqueues next page until maxItems or end.
  • DETAILS — GETs /v1/listings/{id}, merges with basicInfo from the list pass.
• src/lib/buildSearchQuery.ts — payload builder + client-side filter predicate.
• src/lib/utils.ts — URL classification (isDetailPageUrl, extractListingId, extractLanguageFromUrl, …).

Tests

npm test         # 61 unit tests, fully offline (sub-second)
npm run test:live  # 2 integration tests against the real CDN (~1 s, requires internet)

The offline suite covers URL classification, language extraction, filter matching, payload construction, start-URL/bare-id normalisation, and image-URL construction.

The live suite checks that the image-CDN host we construct (listing-images.autoscout24.ch) still resolves to real images. Run it after any AutoScout24-side change to catch silent breakage early. Not invoked by default so flaky networks don't break local dev.

Out of scope

• Pasting search/list URLs (/de/s/...) is not supported — the URL → API mapping isn't published and parsing the query-string syntax (mk-bmw, mo-320, …) is fragile. Use the filter fields instead.
• The Swiss site has a separate iOS app (App Store ID 295835058, Android package ch.autoscout24.autoscout24). The actor does not currently use any iOS-app-specific endpoints — those would need a mitmproxy capture to discover.

Support

For custom solutions, schema changes, or to report a broken endpoint: muhameddidovic@gmail.com