Linkturn
is a specialized command-line utility for extracting URLs and URIs from customized JavaScript files.
It leverages a Convolutional Neural Network (CNN) - NLP model trained on a curated dataset of JavaScript patterns. The model tokenizes and classifies potential matches to minimize false positives and false negatives.
-
AI-Powered Pattern Recognition
Uses a CNN model to learn URL/URI patterns from real-world JavaScript code. -
False Positive Reduction
Token-based classification ensures thatdata:
URIs, base64 blobs, and unrelated strings are excluded. -
Easy Maintained Training
Easily add new variations of invalid/valid sample on thelib/samples
folder to be trained. You need morethan 10 variation of false-positive case to be trained well. -
Lightweight CLI Tool
Minimal dependencies, designed for speed in security testing pipelines.
git clone https://github.com/redspr/linkturn.git
cd linkturn
pip install -r requirements.txt
python linkturn.py <URL>
- Create patterns variation of some invalid or valid sample, paste it on corresponding ".txt"
- Run the command
python trainmodel.py