TrackerSift

TrackerSift, a powerful tool designed for security researchers to easily crawl websites and capture various API calls, network requests, and dynamic features of JavaScript, including stack traces. TrakerSift uses selenium to automatically crawl and navigate through a website, capturing all relevant information using purposely-built chrome extension. Chrome extension can identify and collect data from various sources, including APIs, database queries, server responses, and user interactions. With TrakerSift, you can easily analyze the behavior of a website's JavaScript code, including the different function calls, event listeners, and error messages. The tool is designed to capture stack traces, which can provide valuable insights into the code execution path and help identify potential issues and vulnerabilities.TrakerSift exports your findings in JSON format.

TrackerSift, Abdul Haddi Amjad, Zubair Shafiq, Muhammad Ali Gulzar, IMC'21

The published version of the manuscript is avaibable at : TrackerSift

Installation

We are working on docker and soon we will release the DockerFile

Meanwhile follow these:

Requirements

1. NodeJS
2. Python Conda
3. Xvfb

Steps

Crawling

1. Clone this repository and move inside the repo directory using cd command
2. Open two terminals inside repository directory ---- First Terminal ----
3. In first terminal, run cd server and then node server.js -- this will start the local-host server at Port:3000 which communicates with chrome-extension to save the captured data inside server/output directory. ---- Second Terminal ----
4. In second terminal, create conda environment using requirement.txt -- conda create --name envname --file requirements.txt
5. Now if you want to crawl a specific website e.g, livescore.com you can add any list of websites mentioned in ten.csv which is specified in sele.py - line 20
6. After specifying the websites, simply activate the conda env and run python sele.py -- this will crawl the website and store the respective data in output/server/{website}/ directory e.g, for livescore.com you can find in output/server/livescore.com/.

See SCHEMA.md for each file.

Labelling network request using EasyList and EasyPrivacy

7. In the same conda env you can run python label.py to label the network requests for all crawled websites. This will create label_request.json inside each website directory.

Generating graph

7. In the same conda env you can run python graph-plot/main.py to create graph for each website. This will create label_request.json inside each website directory.

Extracting features from graph

8. In the same conda env you can run python graph-plot/makeFeatures.py to create graph for each website. This will create multiple files inside website directory, see SCHEMA.md for each file.

Research papers that used this instrumentation

1. TrackerSift
2. Blocking JavaScript without Breaking the Web

Citation

@inproceedings{Amjad22TrackerSift, title = {TrackerSift}, author = {Abdul Haddi Amjad, Zubair Shafiq, Muhammad Ali Gulzar}, booktitle = {ACM Internet Measurement Conference (IMC)}, year = {2021} }

Contact

Please contact Hadi Amjad if you run into any problems running the code or if you have any questions.