SaveNowX: How It Downloads X Videos Without Server Storage
An in-depth look at SaveNowX, a free tool that downloads X videos without accounts or server storage, built using Next.js, FastAPI, and Cloudflare Workers.
Stock photo for illustration only, not from the actual event
- SaveNowX is a free tool to download X videos without user accounts or watermarks.
- Designed under a strict rule: the server must never store video files or download history.
- Uses a Cloudflare Worker to handle cross-origin downloads without proxying through the backend.
- Supports HLS streaming conversion and MP3 extraction with temporary files instantly deleted after delivery.
When developers build social media video downloaders, the naive approach is to fetch the video file onto their own server before streaming it to the user. However, for SaveNowX, the creator set a strict architectural constraint from day one: nothing touches the server disk, ever. No stored video files, no download history, and nothing left behind that could leak or require manual cleanup.
This design constraint forced several crucial architectural decisions, transforming a basic utility into a highly optimized, privacy-first application that minimizes backend workload and maximizes processing speed.
Stock photo for illustration only, not from the actual event
The first step begins when a user pastes a link into the GET /api/resolve?url= endpoint. This route fetches metadata and CDN URLs directly without downloading a single byte of actual video data. This keeps the resolution step extremely fast and lightweight under high load. Results are cached using Redis or an in-process dictionary with two distinct Time-To-Live (TTL) durations:
- 1 hour for successfully resolved links.
- 5 minutes for genuine 'not found' or 'no video' errors, preventing bots or repeated bad requests from hammering the syndication endpoint.
For standard posts containing direct MP4 links from X's CDN (video.twimg.com), instead of having the backend fetch and stream the file, SaveNowX directs requests through a lightweight Cloudflare Worker sitting at the edge. The Worker's sole responsibility is setting the proper download header, allowing browsers to download the file directly instead of navigating to it. As a result, the video travels straight from X's CDN to the user's device, while the FastAPI backend gets completely out of the way.
"Nothing touches disk on the server, ever. No stored video files, no download history, nothing to clean up or leak."
Dev.to
Leveraging Edge Computing like Cloudflare Workers to handle HTTP headers is an effective modern web design pattern. It drastically offloads compute overhead from the primary origin server, bypasses cross-origin download restrictions naturally, and significantly enhances application scalability without requiring costly infrastructure expansions.
Nevertheless, a major hurdle arises with HLS-only posts, which consist of a .m3u8 playlist and numerous small .ts segments rather than a progressive MP4 file. For these scenarios, the application triggers a specialized job system utilizing yt-dlp and ffmpeg to fetch, remux, and stream the finished file while reporting real-time progress. True to its core promise, every temporary file generated during this process is deleted the exact moment it is sent.
Source: Dev.to
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment