Yandex Search Scraper

Collect complete public organic web results from Yandex with one ordinary Chromium browser. The Actor does not mask navigator.webdriver, inject browser fingerprints, rotate sessions, retry challenges, intercept private APIs, or return fabricated fallback rows.

Input

{
  "query": "machine learning",
  "maxItems": 10,
  "maxPages": 1,
  "language": "en",
  "lr": "",
  "timeRange": "",
  "requestDelayMs": 0,
  "proxyConfiguration": { "useApifyProxy": false }
}

• query: 1–200 characters.
• maxItems: 1–100 complete organic results.
• maxPages: 1–5 public pages.
• language: two-letter language code.
• lr: optional numeric Yandex region code.
• timeRange: any time, past day (1), week (7), or month (30).
• requestDelayMs: fixed 0–3000 ms delay before reading each page.
• proxyConfiguration: optional explicit Apify Proxy configuration; direct is the default.

Dataset

Each dataset item has a visible result title and a public external HTTPS URL. Stable URL-derived IDs, continuous positions, exact domains, query/filter context, page provenance, connection mode, and fingerprintApplied: false are always present. Optional snippets, displayed paths, date text, images, breadcrumbs, and site links are omitted when absent. Ads and Yandex redirect/internal URLs are not counted.

Run-level CAPTCHA, access, timeout, navigation, empty-page, and incomplete-card evidence is written to the OUTPUT key rather than the paid dataset. A zero-row run exits unsuccessfully.

Access behavior

Yandex can show CAPTCHA or access-limited pages depending on network and region. The Actor fails closed at that boundary. It does not retry with a changed identity or silently downgrade a requested proxy. Start with one page and a small item limit.