Scrape car listings from Kleinanzeigen.de, Germany's largest classifieds site: make, model, year, mileage, price, postcode and city, photo and URL. No proxy needed, about 27 cars per page.
Store page only; the extraction code is unchanged and the output fields keep their names.
README rewritten for the Store: data, use cases, input/output examples, pricing, limits and
FAQ, including the warning about the per-IP limit on large runs without a proxy.
dataset_schema.json is now a real JSON schema (nullable types, required core fields) with
overview and all_fields views, wired through storages.dataset in actor.json.
Input: prefill of 1 page; proxy prefill and default are "no proxy" (the Apify datacenter
proxy gets HTTP 403); sections and clearer descriptions.
Default memory lowered from 512 to 256 MB (measured peak 59-64 MB).
Output schema in the current actorOutputSchemaVersion format, linking the overview view.
0.2 (2026-11-14, beta)
Fix: 0 listings since Kleinanzeigen moved the result list to a new layout
(<article data-adid> cards; km/EZ in data-dhl-promotion spans; new price
and location markup). The parser now reads the new cards, including titles
wrapped in <span data-url>.
Fix: the configured proxy was never passed to the HTTP client; each page now
goes through a fresh proxy URL when a proxy is configured. With no proxy
configured the actor connects directly (as 0.1 did in practice): Apify
datacenter proxy gets HTTP 403. Console prefill changed from RESIDENTIAL to
no proxy.
The run now fails with a clear status message when it extracts nothing
(block, HTTP error or layout change) instead of ending SUCCEEDED with 0 items.
Top ads repeated on later pages are emitted once per run.
Make detection is word-bounded, picks the earliest make in the title, and
skips the "Benz" in "Mercedes Benz"; a few makes added (output spellings kept).