Yahoo Videos Scraper

Search public Yahoo Video Search and export normalized, API-ready video cards. The Actor reads Yahoo's server-rendered result cards, extracts up to 60 cards per page, follows b=1,61,121… pagination, and completes typical runs without launching a browser.

Features

• Single-query and deduplicated multiple-query modes
• Fair interleaving with global and per-query limits
• Up to 20 pages per query, currently about 60 source cards each
• Verified US, UK, and Canada regional Yahoo Video hosts
• Source-backed short, medium, and long duration filters
• Platform filters for YouTube, Vimeo, Dailymotion, TikTok, Twitch, Facebook, X, and other sources
• Yahoo relevance, shortest-first, and longest-first ordering
• Stable IDs and global video URL deduplication
• Real video and canonical embed URLs with platform video IDs
• Titles, Yahoo card descriptions, durations, displayed view-count labels, thumbnails, reported dimensions, aspect ratio, and platform/domain metadata
• Direct, Apify Residential/Datacenter, and custom proxy operation
• Bounded retries, strict response validation, block detection, and clean no-results handling

Input

Provide query, queries, or both. Duplicate queries are removed case-insensitively.

Field	Description	Default
query	One Yahoo Video query	—
queries	Up to 20 queries	—
maxItems	Maximum videos across all queries	100
maxItemsPerQuery	Optional cap per query	fair share of maxItems
maxPages	Maximum pages per query, 1–20	10
country	Two-letter country code	us
language	Language or language-country locale	en
duration	any, short, medium, or long	any
platform	Any or a supported source platform	any
sortBy	relevance, durationAsc, or durationDesc	relevance
maxConcurrency	Concurrent Yahoo page requests, 1–10	5
maxRequestRetries	Temporary retries, 0–3	2
requestTimeoutSecs	Per-request timeout	30
proxyConfiguration	Apify or custom proxy configuration	direct

Example batch input:

{
  "queries": ["machine learning", "artificial intelligence documentary"],
  "maxItems": 20,
  "maxItemsPerQuery": 10,
  "maxPages": 3,
  "duration": "short",
  "platform": "youtube",
  "sortBy": "durationAsc",
  "proxyConfiguration": {
    "useApifyProxy": true,
    "apifyProxyGroups": ["RESIDENTIAL"],
    "apifyProxyCountry": "US"
  }
}

Set maxItems to 1 for one search item.

Output

The 53-field dataset contract can include:

• stable ID and global, query, page, and page-card positions
• title, full Yahoo description, real video URL, domain, and source platform
• Yahoo's displayed compact view-count label, preserved as text
• platform video ID and canonical embed URL
• Yahoo thumbnail URL, alternative text, title, width, height, aspect ratio, and host
• duration label, numeric seconds, and derived short/medium/long category
• exact query, regional search host, country, language, filters, ordering, provenance, and aggregate search metadata

Unavailable optional values are omitted rather than emitted as null, empty strings, or fabricated placeholders. Current result cards expose a platform/domain label and compact view-count text; the Actor preserves the displayed label without converting it into an exact count. The current Chrome observation did not expose a distinct channel name or publication age/date, so those values are not claimed.

Filter behavior

Yahoo's former age and videodur URL parameters no longer change current video result cards. The Actor enforces duration and platform filters after extraction using Yahoo's displayed duration and real destination domain.

Yahoo cards expose no publication date. Non-empty legacy timeRange input therefore fails validation instead of silently returning unfiltered data.

Duration categories are:

• short: under 5 minutes
• medium: 5 minutes through 19:59
• long: 20 minutes or longer

Duration sorting covers the candidates fetched within maxPages; it does not claim to sort Yahoo's entire index.

Regional behavior

us, gb, and ca use verified regional Yahoo Video Search hosts. Other valid two-letter country codes use the global host with locale context because the tested AU and IN video subdomains currently return empty card layouts.

Performance and resilience

Yahoo served 60 result cards in the inspected desktop page. The Actor fetches independent pages and queries concurrently using one stable authorized proxy session for the run. It scopes extraction to the labeled Video Search Results list, excluding related-search suggestions, and validates HTTP status, HTML content type, final Yahoo host, response size, challenge content, and current card structure before parsing or storing anything.

Challenge pages are never retried or stored. Ordinary transient network and server failures use bounded exponential backoff on the same route. Rate limits, access denials, redirects, verification/login gates and TLS/proxy-authentication failures stop new requests across all workers. Logs never include raw HTML, result URLs, tracking/beacon data, cookies, or proxy credentials.

Proxy notes

Use Apify Residential or Datacenter proxies when direct access is rate-limited. GOOGLE_SERP cannot be used because it supports selected Google Search/Shopping requests, not Yahoo.

Local run

npm ci
apify run --purge --input-file INPUT.json

Results are written to storage/datasets/default.

Limitations

Some queries return overlapping videos on adjacent Yahoo pages; the Actor removes them globally. Yahoo can also return genuine fallback videos for very narrow queries. A structural no-results page is distinguished from those legitimate fallback cards. The Actor does not bypass authentication, CAPTCHAs, paywalls, or access controls.

Bounded preview and run status

The UI and INPUT.json reproduce the verified historical machine learning input: five records, one page, one worker, no retries, 25-second request timeout and direct access. Omitted API inputs keep their existing defaults. Historical results do not certify current availability.

OUTPUT reports SUCCEEDED, NO_RESULTS, PARTIAL, BLOCKED or FAILED with completed, failed, skipped and incomplete query/page counts. Successfully completed and in-flight pages survive a later failure. Empty failures and rejected dataset/summary writes exit with code 1. Article/query words such as login or CAPTCHA do not establish an access gate.

Canonical YouTube/Vimeo embed URLs may be derived from verified platform URLs and video IDs; they do not prove that playback or embedding is permitted. isEmbeddable is omitted unless the source explicitly supplies a boolean. Missing dimensions and invalid duration labels do not become invented media measurements. Platform filters match verified domains or exact source labels rather than incidental hostname substrings.

Duration caveat: Yahoo search cards can display abbreviated or inconsistent lengths. duration is its visible label; durationSeconds and durationCategory only parse that label, and duration filters use those source-reported values. Records with a duration carry durationSource: "yahoo_search_card" and durationIsVerified: false. The Actor does not open video destinations to verify playback length or infer missing hours from titles or other search engines.