Code Fingerprint Plagiarism Detector

Automated code checkers in AI-driven education platforms don't diff raw text — they tokenize submissions by grammatical role, hash overlapping k-grams, and keep only local-minimum hashes via the winnowing algorithm, then compare the resulting fingerprint sets with a Jaccard index. This simulator renders two code submissions as two rows of tokens in real 3D: pick how submission B was derived from A — an identical copy, a variable rename, a block reorder, or a full rewrite — and watch the k-gram scanner sweep both rows, selecting fingerprints and drawing gold arcs between every match. Live readouts track the similarity score, the matching-fingerprint count, and the token counts on each side, so you can see exactly why a rename fools no one while a genuine rewrite doesn't get flagged.