Instructors & TAs
Automate structural similarity checks across dozens of programming assignments. Quickly identify suspicious submissions with clear, visual evidence.
Hybrid Code
Similarity Analyzer
Detect Java source code clones and structural plagiarism instantly. By leveraging AST-aware hybrid scoring, LCS alignment, and tokenized K-gram Winnowing, CodeSniff uncovers Type 1–4 clones even when logic is heavily disguised.
<200ms
Avg Response Time
AST+Token
Hybrid Scoring
Type 1–4
Clone Detection
Structural Analysis
AST parsing and tokenized k-gram fingerprints resist renaming, statement reordering, and comment-removal tricks. Safely detects disguised plagiarism that text matchers miss.
Hybrid Scoring
Combines Winnowing fingerprint similarity, coverage metrics, LCS statement alignment, and AST structural comparison for reliable, nuanced similarity scores.
History & Exports
Side-by-side highlighting makes reviews fast. Authenticated users can track analysis history, search past runs, and export reports in PDF, CSV, or HTML.
Who can use our structural analyzer
When it comes to coding assignments or software auditing, originality and deduplication are key. Our structural similarity detector shines when you need fast, reliable checks and clear guidance for evaluating code overlaps.
Students & Self-Checkers
Check your own code before submission. Accidental overlap happens, and a structural check prevents embarrassing academic integrity mistakes.
Developers & Auditors
Audit large codebases for duplicated logic, refactor opportunities, and unintentional copy-pasting using scalable batch analysis.
Discover the benefits of using CodeSniff
- Advanced Detection: Catches identifier renames and operator divergence (e.g., max-finding vs min-finding) using AST and Winnowing algorithms.
- Contextual Explanations: Plain-English clone-type classifications (e.g., "Type-3 Clone (Logic Mutation)").
- Cloud Persistence: Optional secure authentication to save analysis history to a PostgreSQL database for later review.
- Multi-Format Exports: Generate and download structured similarity reports in CSV, HTML, or PDF formats.
CodeSniff operates as a fast, in-memory structural comparison engine. It detects exact matches, logic mutations, and heavy structural rearrangements. Standard framework boilerplate is detected and handled correctly so that common setups are not flagged as plagiarism by default. The engine looks at the actual AST and execution logic, not just string similarity.
Similarity ranges from intentional theft to accidental overlap. We flag both: deliberate obfuscation (renamed variables, reordered loops, inverted conditionals) and unintentional duplication. The tool helps users understand the true overlap by identifying problem areas and categorizing the clone type.
The consequences of plagiarism can be severe. Regular structural checks limit risk and teach good habits. Use CodeSniff as part of your grading workflow or codebase assessment to remain confident in your code's originality and maintain DRY principles.
How a structural checker helps reviewers
- Rely on AST-aware algorithms, not fragile exact-text matching.
- Use the batch heatmap to quickly isolate suspicious clusters.
- Review the plain-English clone classification to understand the intent.
- Export reports to keep clear, objective records for offline grading.
Analyze Your Code
Drag & drop source files here
or
Welcome Back
Forgot password?
New account? Sign
Up
Set New Password
Dashboard
Analyses (Last 30 Days)
0
Total Analyses
0
Usage History (Last 30 Days)
Loading your history...
Profile
Settings
Personal Details
Help & FAQ
How does CodeSniff work? (Placeholder)
CodeSniff uses a hybrid approach leveraging AST (Abstract Syntax Tree) parsing and tokenized k-gram fingerprinting to accurately detect Type 1 to 4 code clones.
How do I read similarity scores? (Placeholder)
The similarity score is a percentage reflecting the structural and token overlap between the queried file and the reference codebases. Higher percentage indicates higher similarity.
What do the verdict levels mean? (Placeholder)
Verdicts are categorized as Low (green), Medium (orange), and High (red) similarity. High generally indicates a strong likelihood of cloned code.