TikTok Creator Videos Scraper

Enter one or more TikTok creator usernames and get every public video from each profile, with full metadata and engagement stats.

---

中文说明

输入一个或多个 TikTok 达人用户名，抓取每位达人的全部公开视频，含完整元数据和互动指标。

输入参数

字段	类型	说明
creators	数组（必填）	达人用户名（可带或不带 @）或主页链接，每行一个
maxVideos	整数	每个达人最多抓取的视频数；0 = 全部
maxConcurrency	整数	同时抓取的达人数量，默认 2
maxRequestRetries	整数	每个达人失败后的重试次数，默认 2
requestTimeoutSecs	整数	页面导航和关键请求超时，默认 45 秒
proxyConfiguration	对象	默认使用 Apify Proxy DEFAULT 非住宅池；不会自动切住宅代理

输出（Output schema）

每条记录对应一个视频，主要字段：

字段	说明
id / url	视频 ID / 链接
text	文案
createTime / createTimeUnix	发布时间（ISO / 时间戳）
durationSeconds	时长（秒）
cover / dynamicCover	静态 / 动态封面
width / height	分辨率
playCount	播放量
diggCount	点赞
commentCount / shareCount / collectCount	评论 / 分享 / 收藏
hashtags	话题标签数组
music	音乐：{ id, title, authorName, isOriginal }
isPinned	是否置顶
authorUsername / authorNickname / authorId / authorVerified	达人信息

控制台 Output 标签会以带封面缩略图的表格呈现以上结果。

工作原理：用真实浏览器打开达人主页，先读取服务端渲染的视频数据；再调用 TikTok 自身的 item_list 接口翻页——用页面内置的签名函数 byted_acrawler.frontierSign 对每次请求重新签名（X-Bogus），跟随 cursor 翻到底，从而拿到达人全部公开视频，不依赖任何第三方 actor。页面没有主动发出 item_list 请求时，actor 会用页面里的 secUid 构造同等参数模板并签名请求，提升非住宅代理下的结果获取率。

注意：只能抓取公开视频；返回数量可能略少于主页显示的总数（差额为私密/地区限制/已删除）。默认不会使用住宅代理，也不会自动切换到 RESIDENTIAL；如果用户在代理配置里手动选择 RESIDENTIAL，actor 会尊重用户选择。

---

Input

Field	Type	Description
creators	array (required)	Usernames (with or without @) or full profile URLs.
maxVideos	integer	Max videos per creator. 0 = all available.
maxConcurrency	integer	Number of creators scraped in parallel. Default 2.
maxRequestRetries	integer	Retries per creator. Default 2.
requestTimeoutSecs	integer	Navigation and key request timeout in seconds. Default 45.
proxyConfiguration	object	Defaults to Apify Proxy DEFAULT non-residential pool. No automatic residential fallback.

{
    "creators": ["mrbeast", "@khaby.lame", "https://www.tiktok.com/@zachking"],
    "maxVideos": 0,
    "maxConcurrency": 2,
    "maxRequestRetries": 2,
    "requestTimeoutSecs": 45,
    "proxyConfiguration": { "useApifyProxy": true }
}

Output

One dataset record per video:

Field	Description
id, url	Video ID and canonical URL
text	Caption / description
createTime, createTimeUnix	Publish time (ISO + epoch)
durationSeconds	Video length
cover, dynamicCover	Thumbnail images
width, height	Resolution
playCount	Views
diggCount	Likes
commentCount, shareCount, collectCount	Engagement
hashtags	Array of hashtag titles
music	{ id, title, authorName, isOriginal }
isPinned	Whether the video is pinned on the profile
authorUsername, authorNickname, authorId, authorVerified	Creator info

Example record

{
    "id": "7300000000000000000",
    "url": "https://www.tiktok.com/@zachking/video/7300000000000000000",
    "text": "Magic trick #magic",
    "createTime": "2024-11-01T12:00:00.000Z",
    "durationSeconds": 12,
    "playCount": 12000000,
    "diggCount": 800000,
    "commentCount": 5400,
    "shareCount": 22000,
    "hashtags": ["magic"],
    "music": { "id": "...", "title": "original sound", "authorName": "zachking", "isOriginal": true },
    "isPinned": false,
    "authorUsername": "zachking",
    "authorVerified": true
}

How it works

The actor opens each profile in a real Chromium browser (via Crawlee's PlaywrightCrawler) and reads the server-rendered video data. To page through the creator's entire back-catalog it calls TikTok's item_list API directly:

1. It captures the exact item_list request the page issues itself (which carries TikTok's full expected parameter set), or builds an equivalent synthetic /api/post/item_list/ template from the page's secUid when the page does not emit the request quickly.
2. For each subsequent page it swaps the cursor on that template and re-signs the request with TikTok's own in-page signer (window.byted_acrawler.frontierSign, producing the required X-Bogus), then fetches from inside the page so the request carries the page's cookies (msToken).
3. It follows the server-provided cursor until hasMore is false.

This is fully self-contained — no dependency on any third-party actor — and returns public videos beyond the initial profile batch. Browser fingerprinting, shorter empty path waits, and signed synthetic pagination improve non-residential success rates.

Notes

• Only public profiles/videos are accessible. The returned count may be slightly below the profile's stated total — the difference is private, region-locked, or deleted posts.
• By default, the actor uses Apify Proxy DEFAULT non-residential pool and does not switch to RESIDENTIAL automatically. If you explicitly select RESIDENTIAL, the actor uses your selection.
• TikTok rotates its request-signing internals periodically. If a run ever returns zero, the signer hook in src/main.js (frontierSign) is the place to update.