Autogidas.lt Car Ads Extractor

Turns publicly listed used-car ads on Autogidas.lt into a structured dataset — price, year, mileage, engine, gearbox, photos and the full parameter table from each ad.

Only publicly available data is returned. Everything this actor collects is what any visitor sees on Autogidas' own public listing and ad pages. It signs in to nothing, holds no account, and reaches nothing behind a login, a paywall or a private area of the site. That includes sellerName, which is simply the dealer or seller name the ad itself displays.

Start URLs are the primary input

Build the search on autogidas.lt, then paste the URL into Start URLs.

That is the intended way to run this actor. The site's own search supports far more than the filter fields below it — equipment, seller type, damage status, colour, and every combination of them — and pasting the URL gets all of it for free, exactly as the site interprets it. What you see in the browser is what the dataset will contain.

Several URLs are fine. Their results are merged and de-duplicated, so two overlapping searches never produce the same ad twice.

{
  "startUrls": [{ "url": "https://autogidas.lt/en/skelbimai/automobiliai/bmw/" }],
  "scrapeDetails": true,
  "maxItems": 1000
}

Lithuanian, Russian and Latvian URLs all work — they are rewritten to the English edition before they are fetched, so the labels come back in English.

The filter fields are a fallback, and not recommended

makes, priceFrom/To, yearFrom/To, mileageFrom/To, fuelType, bodyType, gearbox, city, validInspectionOnly, withVinOnly and hideAuctions exist for callers that have to construct input programmatically and cannot obtain a URL. They rebuild an Autogidas search from scratch and reach only a subset of what the site's search can express — everything else it offers, such as "only cars in Lithuania", is reachable only by pasting a URL.

They are used only when Start URLs is empty. Set a start URL and every filter is discarded — no error, because there is nothing wrong with the input; the run logs a warning naming each field it ignored. Apify input forms cannot disable one field because another is filled, so this is the clearest signal the platform allows.

One thing to know if you do use them: Autogidas puts make and model in the URL path, so each make/model combination becomes its own search and the actor runs them all. The make dropdown is a suggestion rather than a whitelist — a make the site's own picker omits can still be typed in — and models are free text, because the site publishes no per-make model list.

Settings that apply either way

Field	What it does
scrapeDetails	Visit each ad for the full parameter set (description, defects, consumption figures, all photos). Turn it off for ~20x fewer requests and the results-page fields only.
maxItems	Hard cap on ads pushed to the dataset.
proxyConfiguration	Residential proxy pinned to Lithuania, by default.

Keep the Lithuania-pinned residential proxy. Cloudflare fronts the site and 403s a share of foreign exit IPs. With the defaults left alone, a measured platform run made 64 requests with zero failures and zero retries.

Unlike the sibling Autoplius actor there is no pagination cap — the full ~28,000-ad catalogue is reachable from a single search.

Output

One dataset record per ad. Known parameters are mapped to typed fields — price, year, mileageKm, powerKw, cylinders, gears, steeringSide, defects, firstRegistrationCountry, fuel consumption and imageUrls among them — and every label/value row from the ad is also kept verbatim in rawParameters, so a change on the site adds unmapped keys rather than silently dropping data.

Note that Autogidas masks the VIN (WVWZZZAU...), so the field is reported as vinPartial rather than vin.