Contact

For any inquiries or issues, please contact tnot2652@gmail.com.

YouTube Advanced Scraper

This actor scrapes regular YouTube video data from:

• text search queries
• direct watch URLs and youtu.be links
• channel URLs such as https://www.youtube.com/@Google, youtube.com/@Google, or @Google

It returns one dataset item per video with:

• title
• views
• url
• publication_date
• subscribers
• channel_link
• channel_name
• comments
• related_videos
• data_date
• video_id
• duration_seconds
• thumbnail_url

Pricing transition

This Actor migrates from monthly rental to pay per event. Apify shows the effective date and the price for each run in Console.

Apify plan	Price per 1,000 videos
Free and Bronze	$0.50
Silver	$0.45
Gold, Platinum, and Diamond	$0.40

The startup charge is $0.00005 per started GB of allocated memory, with a minimum of one event. Each video saved to the default dataset counts as one result. Comments and related links inside a video record have no separate event charge. Runs with no results can still incur the startup charge. The Actor stops when the remaining budget cannot cover another video.

Current Behavior

• The default input is the public video “Me at the zoo”.
• Channel root URLs are normalized to the channel's /videos tab automatically.
• Search and channel pagination use YouTube continuation endpoints, so higher max_results values work.
• Search and channel scraping now builds base dataset items from list renderers first. Watch pages are opened only for direct video inputs or when comment / related-video enrichment is requested.
• Search results now do lightweight subscriber enrichment with one channel-page request per unique channel, using a bounded in-memory cache to avoid unbounded growth on large runs.
• Duplicate input entries are deduplicated before scraping.
• Unsupported YouTube URLs, such as playlist pages, cause an input error.
• If an input fails, the run reports failure and preserves the videos already saved.
• The RUN_SUMMARY record in the key-value store lists completed, empty, and failed inputs.
• A valid search with no matching videos completes with zero results.

Input

The actor accepts a JSON object with these fields:

• queries: Required. A mix of search queries, direct video URLs, or channel URLs. Bare handles such as @Google and scheme-less channel URLs such as youtube.com/@Google are accepted.
• sort_by: Optional. Search sorting for text queries only. One of relevance, view, date, rating. Default is relevance.
• max_results: Optional. Maximum number of videos to collect per search query or channel URL. Default is 10.
• num_comments: Optional. Requested comment count per video. Default is 0.
• related_videos: Optional. Whether to try collecting related video links from watch pages. Default is false.

Legacy note: The old clean input is still accepted for backwards compatibility, but it is ignored by the current actor version.

Proxy Setup

Do not place proxy credentials in actor input.

Configure the proxy through actor environment variables instead:

• YOUTUBE_PROXY_URL
• or the split form: YOUTUBE_PROXY_HOST, YOUTUBE_PROXY_PORT, YOUTUBE_PROXY_USERNAME, YOUTUBE_PROXY_PASSWORD, and optional YOUTUBE_PROXY_SCHEME

Example:

YOUTUBE_PROXY_URL=http://USERNAME:PASSWORD@HOST:PORT/

The actor redacts credentials in its own logs and only logs the proxy host and port.

Output

Example dataset item:

{
  "title": "Me at the zoo",
  "views": 355000000,
  "url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
  "publication_date": "2005-04-24",
  "subscribers": 3210000,
  "channel_link": "https://www.youtube.com/@jawed",
  "channel_name": "jawed",
  "comments": [],
  "related_videos": [
    "https://www.youtube.com/watch?v=QH2-TGUlwu4",
    "https://www.youtube.com/watch?v=oHg5SJYRHA0"
  ],
  "data_date": "2026-04-18T00:00:00+00:00"
}

Output Schema

The actor always returns these keys for every dataset item:

{
  "type": "object",
  "properties": {
    "title": { "type": "string" },
    "views": { "type": ["integer", "null"] },
    "url": { "type": "string" },
    "publication_date": { "type": ["string", "null"] },
    "subscribers": { "type": ["integer", "null"] },
    "channel_link": { "type": ["string", "null"] },
    "channel_name": { "type": ["string", "null"] },
    "comments": {
      "type": "array",
      "items": { "type": "string" }
    },
    "related_videos": {
      "type": "array",
      "items": { "type": "string" }
    },
    "data_date": { "type": "string" },
    "video_id": { "type": "string" },
    "duration_seconds": { "type": ["integer", "null"] },
    "thumbnail_url": { "type": ["string", "null"] }
  },
  "required": [
    "title",
    "views",
    "url",
    "publication_date",
    "subscribers",
    "channel_link",
    "channel_name",
    "comments",
    "related_videos",
    "data_date"
  ]
}

Notes

• comments are best effort. The actor now attempts to collect top-level comments through YouTube's watch-page continuation flow. Some videos can still return an empty array when comments are disabled or YouTube blocks the continuation requests.
• related_videos are best effort. Some watch pages block continuation requests and return an empty array.
• Search and channel items can now survive blocked watch pages. In those cases, the actor logs the watch-page problem and keeps the list-page fallback item.
• subscribers are now hydrated from direct watch pages, channel headers, or one lightweight channel-page request per unique channel when available. They can still be null if YouTube does not expose the count.
• sort_by does not affect direct video URLs or channel URLs.
• publication_date preserves the available precision. List results can contain relative text, such as 3 months ago.
• Direct video pages provide a calendar date when YouTube exposes it. The Actor does not convert relative text into an invented date.
• duration_seconds and thumbnail_url are null when unavailable. Live videos can have no duration.
• If watch-page enrichment fails, a list result can include enrichment_errors with the failure reason.

Machine Specs

256 MB is enough for smaller runs, but larger query lists and high max_results values benefit from more memory and time.