In a second step, Bundle Scanner identifies the specific segment of the bundle code that has been matched by each library and excludes libraries that match intervals that are already taken by other libraries with a better match score. Finally, libraries with a match score above a certain threshold are shown to the user as being part of the bundle.
In benchmark tests, approximately 15% of libraries that are actually inside the bundle are not identified, and around 5% of libraries identified are false positives. The goal is of course to get these percentages closer to zero. The biggest challenge is the large amounts of duplicated code on NPM. Many library authors show a strong devotion to the principles of copy-paste-driven development which makes it hard for Bundle Scanner to distinguishing between libraries.
Bundle Scanner is developed by me - Markus Englund. I'm a developer based in Gothenburg doing mostly Node.js and frontend development. If you have any questions you can email me at
email@example.com. You can check out some of my open source work on GitHub.