Douyin Watermark-Free Media Scraper

Problem

When processing Douyin content, the hard part is often not downloading a single video. A shared link may point to a pure video, an image post, a mixed media work, or a slides/animated work, and each type can require different parsing and file handling. A video-only downloader may miss images or produce inconsistent directories and filenames. This skill addresses that: given a Douyin link or shared text, it detects the work type and saves the downloadable media into a consistent local structure.

How It Works

The core workflow combines multi-source fallback parsing and unified media output. It first tries iesdouyin, then falls back to Douyin web parsing, and continues with API-based resolution when needed, which improves compatibility across work types. The skill detects the content category and handles each case separately:

• Image-only posts: downloads all images
• Video posts: obtains a watermark-free video link and saves the file
• Mixed works: saves both images and video files
• Slides/animated works: saves frame images plus video files

Output is written under douyin-scraper-downloads/YYYYMMDD_author_title/, with filenames such as author_title or author_title_01, making it easier to organize by date, author, and title. For video content, it can also extract spoken text when speech-to-text is enabled; that capability requires configuring DOUYIN_API_KEY or API_KEY.

Scope And Limits

This skill is best suited for link parsing, media download, and local archiving, not full account-level crawling or real-time data pipelines. Supported inputs include short links, video pages, image pages, and shared text that contains a link. Output filenames strip emojis, spaces, punctuation, hashtags, and other unsupported characters, and author/title fields are length-limited. Speech transcript extraction depends on an external API, so without a configured key the base download workflow should be used.