Proof of concept and demo of using WebRTC data channels to create a browser-based P2P file sharing network.
This project is maintained by rohansingh
This is a proof of concept and demo of using WebRTC data channels to create a browser-based P2P file sharing network. The actual implementation is just a proof-of-concept, and not actually optimized for any real world usage.
So far, this has only been tested with Chrome 26. In order to work, the Enable RTCDataChannel flag must be enabled in Chrome on the chrome://flags configuration page.
Since this is a proof-of-concept, it has a number of limitations that mean it cannot be used for real life file sharing:
Only the first 100 bytes of any file are shared. There is a size limit on each message on a data channel. Unfortunately, Chrome 26 doesn't support reliable transport, so the data channel is UDP-like rather than TCP-like. This means that building a reliable transport for transferring an entire file in multiple blocks would require some work, and is beyond the scope of this proof-of-concept.
File requests are sent to all available peers. If there are multiple peers with the same file available, they will all respond to the request.
Each peer automatically connects to all other peers. This is clearly not scalable, but works for the purposes of this demonstration.
Install with npm install and start with npm start. The server will
start at http://localhost:8000.
Browse to the address above in two or more tabs, and register some files in each tab (preferrably text files).
Copy a file hash from one tab into another tab's Request File box and request the file. The first 100 bytes of the request file should be transferred.
Files are transferred directly between peers, but a server is needed for two things:
WebRTC requires a server or some sort of intermediary to exchange signaling and session data. This must be done before a connection between two peers can be established.
In this case, each peer communicates with the server using WebSockets to exchange signaling data with other peers.
rtc-p2p uses a web worker to compute the SHA1 hashes of registered files in the background. Web workers run on the client, but must be served from a server.
Slides from my presentation on this project at sthlm.js are available at: http://www.rvl.io/rohansingh/p2p