https://github.com/anubislms/mayat
Experimental AST-Based Source Code Similarity Detection Tool
https://github.com/anubislms/mayat
anticheat antiplagiarism ast c education plagiarism-checker plagiarism-detection python
Last synced: 2 months ago
JSON representation
Experimental AST-Based Source Code Similarity Detection Tool
- Host: GitHub
- URL: https://github.com/anubislms/mayat
- Owner: AnubisLMS
- License: mit
- Created: 2022-02-16T15:45:23.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2024-04-10T06:22:50.000Z (about 1 year ago)
- Last Synced: 2025-04-08T06:47:29.778Z (2 months ago)
- Topics: anticheat, antiplagiarism, ast, c, education, plagiarism-checker, plagiarism-detection, python
- Language: C
- Homepage:
- Size: 863 KB
- Stars: 23
- Watchers: 4
- Forks: 3
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Mayat
**Mayat** is a code similarity detection tool developed by [Tian(Maxwell) Yang](https://github.com/AlpacaMax). It works by comparing the Abstract Syntax Trees of students' code solutions and generate a similarity score for each pair of students' code.
## Install
```
pip install mayat
python -m mayat.install_langs
```
## Usage
Let's say we need to check all students' `uniq.c` for homework1. The path for each `uniq.c` has the format `homework1//user/uniq.c`. All we need to do is run:
```
python -m mayat.frontends.TS_C homework1/*/user/uniq.c
```
If we only want to check the `main` function, we can do:
```
python -m mayat.frontends.TS_C homework1/*/user/uniq.c -f main
```
Additionally, we can pass more optional arguments for `C.py`:
- `--threshold`: Specify the granularity for the matching algorithm. Default to `5`. A smaller value will cause it to check trivial details, which increases the similarity score of two code even though they might not be similar. A larger value will cause it to overlook some common cheat tricks such as swapping two function definitions.
## Supported Languages
- **C**:
- `mayat.TS_C`
- `mayat.C`(Legacy)
- **Python**:
- `mayat.TS_Python`
- `mayat.Python`(Legacy)
- **Java**:
- `mayat.TS_Java`
## Implement a New PL's frontend
We implement a new programming language's frontend by using classes and functions defined in `mayat`. They are:
- `mayat.AST.AST`: The base class for Abstract Syntax Tree. For a new PL you should inherit this and implement the `AST.create(path)` class method, which takes the path of a program as a parameter and returns the AST representation of that program. Currently it is preferred to use `tree-sitter` parsers to implement language frontends, whose corresponding file should be prefixed with `TS_`.
- `mayat.args.arg_parser`: A `argparse.ArgumentParser` object. We need to use this object to retrieve command arguments. We can add new arguments if needed.
- `mayat.driver.driver`: The driver function that takes the inherited AST class and the parsed arguments as parameters and run the plagiarism detection algorithm.
An example of this can be find in `mayat/frontends/TS_C.py`, which is a C frontend implemented using `tree-sitter-c` parser.
## Testing
```
cd tests
python test.py -v
```
## Limitations
This tool will never work for assembly code as the code has to be written in a high level programming language that can be converted into an AST. We can potentially figure out a way to automatically reverse engineer assembly code back to C and then convert it to AST. However, there's no guarantee that the reverse-engineered code can be a good representation for its assembly counterpart.