nfstream: a flexible network data analysis framework

nfstream is a Python package providing fast, flexible, and expressive data structures designed to make working with online or offline network data both easy and intuitive. It aims to be the fundamental high-level building block for doing practical, real world network data analysis in Python. Additionally, it has the broader goal of becoming a common network data processing framework for researchers providing data reproducibility across experiments.

Live Demo Notebook
Latest Release
Citation
Downloads
Supported Platforms
Supported Versions
Build Status
Documentation Status
Code Quality
Code Coverage
Discussion Channel

Main Features

• Performance: nfstream is designed to be fast (x10 faster with pypy3 support) with a small CPU and memory footprint.
• Layer-7 visibility: nfstream deep packet inspection engine is based on nDPI. It allows nfstream to perform reliable encrypted applications identification and metadata extraction (e.g. TLS, QUIC, TOR, HTTP, SSH, DNS, etc.).
• Flexibility: add a flow feature in 2 lines as an NFPlugin.
• Machine Learning oriented: add your trained model as an NFPlugin.

How to use it?

• Dealing with a big pcap file and just want to aggregate it as network flows? nfstream make this path easier in few lines:

   from nfstream import NFStreamer
   my_awesome_streamer = NFStreamer(source="facebook.pcap") # or network interface (source="eth0")
   for flow in my_awesome_streamer:
       print(flow)  # print it, append to pandas Dataframe or whatever you want :)!

    NFEntry(
        id=0,
        first_seen=1472393122365,
        last_seen=1472393123665,
        version=4,
        src_port=52066,
        dst_port=443,
        protocol=6,
        vlan_id=0,
        src_ip='192.168.43.18',
        dst_ip='66.220.156.68',
        total_packets=19,
        total_bytes=5745,
        duration=1300,
        src2dst_packets=9,
        src2dst_bytes=1345,
        dst2src_packets=10,
        dst2src_bytes=4400,
        expiration_id=0,
        master_protocol=91,
        app_protocol=119,
        application_name='TLS.Facebook',
        category_name='SocialNetwork',
        client_info='facebook.com',
        server_info='*.facebook.com,*.facebook.net,*.fb.com,*.fbcdn.net,*.fbsbx.com,*.m.facebook.com,*.messenger.com,*.xx.fbcdn.net,*.xy.fbcdn.net,*.xz.fbcdn.net,facebook.com,fb.com,messenger.com',
        j3a_client='bfcc1a3891601edb4f137ab7ab25b840',
        j3a_server='2d1eb5817ece335c24904f516ad5da12'
    )

• From pcap to Pandas DataFrame?

    my_dataframe = NFStreamer(source='devil.pcap').to_pandas()
    my_dataframe.head(5)

• Didn't find a specific flow feature? add a plugin to nfstream in few lines:

    from nfstream import NFPlugin

    class my_awesome_plugin(NFPlugin):
        def on_update(self, obs, entry):
            if obs.raw_size >= 666:
                entry.my_awesome_plugin += 1
		
    streamer_awesome = NFStreamer(source='devil.pcap', plugins=[my_awesome_plugin()])
    for flow in streamer_awesome:
       print(flow.my_awesome_plugin) # see your dynamically created metric in generated flows

• More example and details are provided on the official documentation.
• You can test nfstream without installation using our live demo notebook.

Installation

Using pip

Binary installers for the latest released version are available:

    python3 -m pip install nfstream

Build from sources

If you want to build nfstream from sources on your local machine:

On Linux

    sudo apt-get install autoconf automake libtool pkg-config libpcap-dev
    git clone https://github.com/aouinizied/nfstream.git
    cd nfstream
    python3 -m pip install -r requirements.txt
    python3 setup.py install

On MacOS

    brew install autoconf automake libtool pkg-config
    git clone https://github.com/aouinizied/nfstream.git
    cd nfstream
    python3 -m pip install -r requirements.txt
    python3 setup.py install

Contributing

Please read Contributing for details on our code of conduct, and the process for submitting pull requests to us.

Authors

Zied Aouini created nfstream and these fine people have contributed.

Ethics

nfstream is intended for network data research and forensics. Researchers and network data scientists can use these framework to build reliable datasets, train and evaluate network applied machine learning models. As with any packet monitoring tool, nfstream could potentially be misused. Do not run it on any network of which you are not the owner or the administrator.

License

This project is licensed under the GPLv3 License - see the License file for details