JoyfulJay - Encrypted Traffic Feature Extraction

JoyfulJay is a Python library for extracting standardized, ML-ready features from encrypted network traffic. It operates on PCAP files and live network interfaces, producing feature vectors that capture timing, size, and protocol metadata patterns - all without decrypting any traffic.

Features

• Encrypted Traffic Focus: Extract features proven effective for classifying TLS, QUIC, VPN, and Tor traffic
• ML-Ready Output: Pandas DataFrames, NumPy arrays, CSV, JSON, or Parquet - ready for scikit-learn, PyTorch, etc.
• Streaming Architecture: Process multi-GB PCAPs without loading them into memory
• Live Capture: Real-time feature extraction from network interfaces
• Remote Capture: Stream packets from remote devices over secure WebSocket (TLS/WSS)
• Protocol Metadata: TLS handshake parsing, JA3/JA3S fingerprints, QUIC metadata
• Traffic Fingerprinting: Detect Tor, VPN, and DoH traffic patterns
• Tranalyzer Compatible: 387 features across 21 extractors, matching research-grade tools
• Enterprise Ready: Kafka streaming, Prometheus metrics, mDNS discovery

Installation

pip install joyfuljay
# or
uv pip install joyfuljay

For optional features (same syntax works with uv pip):

# Fast parsing with dpkt
pip install joyfuljay[fast]

# High-speed capture with libpcap
pip install joyfuljay[libpcap]

# Kafka streaming output
pip install joyfuljay[kafka]

# Prometheus metrics
pip install joyfuljay[monitoring]

# mDNS server discovery
pip install joyfuljay[discovery]

# Connection graph analysis
pip install joyfuljay[graphs]

# All optional features
pip install joyfuljay[fast,kafka,monitoring,discovery,graphs]

Quick Start

Python API

from joyfuljay import extract_features_from_pcap

# Extract features from a PCAP file
features_df = extract_features_from_pcap("capture.pcap")

print(features_df.shape)
print(features_df.columns.tolist())
print(features_df.head())

Command Line

# Extract features to CSV
jj extract capture.pcap -o features.csv

# Live capture for 60 seconds
jj live eth0 --duration 60 -o live_features.csv

# Output as JSON
jj extract capture.pcap -o features.json --format json

Feature Groups

Remote Capture

Stream packets from a remote device (e.g., Android phone, Raspberry Pi) to your analysis machine:

# On the capture device - start server with TLS
jj serve wlan0 --tls-cert server.crt --tls-key server.key --announce

# On your machine - discover and connect
jj discover                    # Find servers on LAN
jj connect jj://192.168.1.50:8765?token=xxx&tls=1 -o features.csv

Kafka Streaming

Stream features directly to Kafka for real-time pipelines:

from joyfuljay.output.kafka import KafkaWriter

with KafkaWriter("localhost:9092", topic="network-features") as writer:
    for features in extract_features_streaming("capture.pcap"):
        writer.write(features)

Prometheus Metrics

Export processing metrics for monitoring:

from joyfuljay.monitoring import PrometheusMetrics, start_prometheus_server

metrics = PrometheusMetrics()
start_prometheus_server(9090)  # Scrape at http://localhost:9090/metrics

Requirements

• Python 3.10+
• scapy >= 2.5.0
• pandas >= 2.0.0
• numpy >= 1.24.0

Cross-Platform Support

Check your system status with:

jj status

Documentation

Full documentation: docs.joyfuljay.com

Citation

If you use JoyfulJay in academic research, please cite:

@software{joyfuljay2025,
  title = {{JoyfulJay}: Encrypted Traffic Feature Extraction Library},
  year = {2025},
  publisher = {GitHub},
  url = {https://github.com/cenab/joyfuljay}
}

License

MIT License - see LICENSE for details.