https://github.com/stanislav-janu/bunny-stream-uploader
Native macOS app for fast parallel uploads of large video files to Bunny Stream over TUS concatenation
https://github.com/stanislav-janu/bunny-stream-uploader
bunny-stream bunnycdn macos parallel-upload swift swiftui tus uploader video-upload
Last synced: 14 days ago
JSON representation
Native macOS app for fast parallel uploads of large video files to Bunny Stream over TUS concatenation
- Host: GitHub
- URL: https://github.com/stanislav-janu/bunny-stream-uploader
- Owner: stanislav-janu
- License: mit
- Created: 2026-06-27T17:12:22.000Z (26 days ago)
- Default Branch: main
- Last Pushed: 2026-06-27T18:53:39.000Z (26 days ago)
- Last Synced: 2026-06-27T19:15:36.537Z (26 days ago)
- Topics: bunny-stream, bunnycdn, macos, parallel-upload, swift, swiftui, tus, uploader, video-upload
- Language: Swift
- Size: 1.11 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Bunny Stream Uploader
[](https://github.com/stanislav-janu/bunny-stream-uploader/actions/workflows/build.yml)
A native macOS app for uploading large video files to [Bunny Stream](https://bunny.net) as fast as the link allows. It splits a single file into many parts and uploads them in parallel over TUS `concatenation`, reaching multiples of single-stream throughput.
On a test link a single 9 GB file went from ~56 minutes (single stream) to ~5 minutes (64 parallel threads), roughly 11x faster.
## Features
- **Parallel single-file upload.** One file is split into N parts, each uploaded over its own TCP connection, then merged on Bunny via `Upload-Concat: final`.
- **Automatic thread count** based on file size (configurable).
- **Drag and drop**, plus a Finder **Quick Action / "Open with"** for common video formats (mp4, m4v, mov, mkv, webm, avi, flv, wmv, ts, mpeg).
- **Resumable single-stream** fallback for small files (via TUSKit).
- **Live throughput** and per-upload progress, with cancel.
- **Credentials stored in the macOS Keychain**, never shown or logged.
- **Localized** in English, Czech, Hungarian, Polish, and German.
## Requirements
- macOS 26 or later
- Xcode 26 toolchain (Swift 6) to build
- A Bunny Stream library (Library ID + Stream API key)
## Build
```sh
swift build -c release
./scripts/make-app.sh release
open BunnyUploader.app
```
`scripts/make-app.sh` builds the binary, assembles `BunnyUploader.app` (icon, localizations, document types), code-signs it, and registers it with LaunchServices.
### Code signing (optional but recommended)
A stable signing identity keeps the Keychain "Always Allow" decision valid across rebuilds. The script auto-detects the first available codesigning identity, or you can pin one:
```sh
BUNNY_SIGN_IDENTITY="Apple Development: you@example.com (TEAMID)" ./scripts/make-app.sh release
```
Without any identity it falls back to ad-hoc signing, and the app still runs locally (the Keychain will prompt on each rebuild).
For the Finder Quick Action and the app icon to behave reliably, move `BunnyUploader.app` into `/Applications`.
## Usage
1. Open **Settings** (⌘,) and enter your Bunny **Library ID** and **Stream API key**. They are stored in the Keychain.
2. Upload a video three ways:
- **Drag and drop** an `.mp4`/`.mov` onto the window.
- **Right-click** a video in Finder → **Quick Actions / Services → Upload to Bunny**.
- **Open with → Bunny Stream Uploader**.
3. Watch the aggregate throughput and per-file progress. Uploads can be cancelled mid-flight.
By default the app picks the thread count by file size (≥ 1 GB → 64, 500 MB–1 GB → 32, 100–500 MB → 16, 50–100 MB → 8, under 50 MB → single stream). You can switch to a manual thread count in Settings.
If a new Service does not appear in the Finder menu, enable it under **System Settings → Keyboard → Keyboard Shortcuts → Services**.
## How it works
Bunny's TUS endpoint advertises the `concatenation` extension and runs over HTTP/1.1, so parallel requests use independent TCP connections (each with its own congestion window). For one file the app:
1. Creates the video via the REST API to get a `videoId`.
2. Signs the upload: `SHA256(libraryId + apiKey + expiration + videoId)`.
3. Creates N partial uploads (`Upload-Concat: partial`) and PATCHes each part's bytes in parallel, reading from disk with `pread` off the cooperative pool.
4. Merges them with `Upload-Concat: final;`.
Single TCP throughput to Bunny is bounded by round-trip time; aggregating independent connections is what multiplies it. See [docs/performance.md](docs/performance.md) for the measured scaling curve.
## License
MIT. See [LICENSE](LICENSE). You are free to use, modify, and distribute this software.
---
Built by [stanislav-janu](https://github.com/stanislav-janu), co-authored with Claude (Anthropic).