Yep Image Discovery Scraper

Yep removed its former /images search vertical; that URL now returns HTTP 404 and the surviving legacy API returns ordinary web pages. Version 2 therefore uses Yep's public web ranking to discover relevant source pages and extracts genuine image assets from Open Graph, Twitter Card, and DOM image markup. It never maps a webpage URL into imageUrl.

The dataset schema declares 22 fields. Each output contains a unique image URL, source-page provenance and rank, extraction method, the query, and an ISO-8601 timestamp. Optional image dimensions, format, alt text, and source-page metadata are included only when present. Source metadata includes the page description, declared HTML language, and publisher/site name when the page exposes them. Signed or credential-bearing URLs are excluded.

The Chrome review confirmed that the public https://www.goldengate.org/ source page loads and visibly identifies the Golden Gate Bridge, Highway and Transportation District through its page title, heading, and footer. The Actor reads description, language, and publisher from source-page metadata; it does not synthesize those fields from visible branding when the metadata is absent. Yep's /web page still showed ordinary web results and a generated TLDR rather than an image-results vertical, so records represent image assets found on the ranked source pages.

Inputs and limitations

• query is required.
• maxItems defaults to 25; maxSourcePages defaults to 10.
• safeSearch is sent to Yep's search response.
• Size filters use only dimensions exposed by source-page markup; unknown-size images are excluded when filtering by size.
• GIF filtering uses the asset path. Color, clipart, and line-art filters cannot be verified after the target change and are rejected explicitly.
• proxyConfiguration supports authorized Apify Proxy or custom proxy settings. The Actor does not solve or bypass CAPTCHA/access controls.

Run locally with apify run --purge --input-file qa-inputs/baseline.json.