joe12138/jplag

Token-Based Software Plagiarism Detection

JPlag - Detecting Software Plagiarism

JPlag is a system that finds similarities among multiple sets of source code files. This way it can detect software plagiarism and collusion in software development. JPlag currently supports various programming languages, EMF metamodels, and natural language text.

Supported Languages

In the following, a list of all supported languages with their supported language version is provided. A language can be selected from the command line using the -l <cli argument name> argument.

Language	Version	CLI Argument Name	state	parser
Java	17	java	mature	JavaC
C/C++	11	cpp	legacy	JavaCC
C/C++	14	cpp2	beta	ANTLR 4
C#	8	csharp	beta	ANTLR 4
Go	1.17	golang	beta	ANTLR 4
Kotlin	1.3	kotlin	beta	ANTLR 4
Python	3.6	python3	legacy	ANTLR 4
R	3.5.0	rlang	beta	ANTLR 4
Rust	1.60.0	rust	beta	ANTLR 4
Scala	2.13.8	scala	beta	Scalameta
Scheme	?	scheme	unknown	JavaCC
Swift	5.4	swift	beta	ANTLR 4
EMF Metamodel	2.25.0	emf	alpha	EMF
Text (naive)	-	text	legacy	CoreNLP

Download and Installation

You need Java SE 17 to run or build JPlag.

Downloading a release

• Download a released version.
• In case you depend on the legacy version of JPlag we refer to the legacy release v2.12.1 and the legacy branch.

Via Maven

JPlag is released on Maven Central, it can be included as follows:

<dependency>
  <groupId>de.jplag</groupId>
  <artifactId>jplag</artifactId>
</dependency>

Building from sources

1. Download or clone the code from this repository.
2. Run mvn clean package from the root of the repository to compile and build all submodules. Run mvn clean package assembly:single instead if you need the full jar which includes all dependencies.
3. You will find the generated JARs in the subdirectory cli/target.

Usage

JPlag can either be used via the CLI or directly via its Java API. For more information, see the usage information in the wiki. If you are using the CLI, you can display your results via jplag.github.io. No data will leave your computer!

CLI

Note that the legacy CLI is varying slightly.

positional arguments:
  rootDir                Root-directory with submissions to check for plagiarism

named arguments:
  -h, --help             show this help message and exit
  -new NEW [NEW ...]     Root-directory with submissions to check for plagiarism (same as the root directory)
  -old OLD [OLD ...]     Root-directory with prior submissions to compare against
  -l {cpp,csharp,emf,go,java,kotlin,python3,rlang,rust,scala,scheme,swift,text}
                         Select the language to parse the submissions (default: java)
  -bc BC                 Path of  the  directory  containing  the  base  code  (common  framework  used  in  all
                         submissions)
  -t T                   Tunes the comparison sensitivity by adjusting the  minimum token required to be counted
                         as a matching section. A smaller <n>  increases  the sensitivity but might lead to more
                         false-positives
  -n N                   The maximum number of comparisons that will  be  shown  in the generated report, if set
                         to -1 all comparisons will be shown (default: 100)
  -r R                   Name of the directory in which the comparison results will be stored (default: result)

Advanced:
  -d                     Debug parser. Non-parsable files will be stored (default: false)
  -s S                   Look in directories <root-dir>/*/<dir> for programs
  -p P                   comma-separated list of all filename suffixes that are included
  -x X                   All files named in this file will be ignored in the comparison (line-separated list)
  -m M                   Comparison similarity threshold [0.0-1.0]:  All  comparisons  above this threshold will
                         be saved (default: 0.0)

Clustering:
  --cluster-skip         Skips the clustering (default: false)
  --cluster-alg {AGGLOMERATIVE,SPECTRAL}
                         Which clustering algorithm to use. Agglomerative  merges similar submissions bottom up.
                         Spectral clustering is  combined  with  Bayesian  Optimization  to  execute the k-Means
                         clustering  algorithm  multiple   times,   hopefully   finding   a   "good"  clustering
                         automatically. (default: spectral)
  --cluster-metric {AVG,MIN,MAX,INTERSECTION}
                         The metric used for clustering. AVG  is  intersection  over  union, MAX can expose some
                         attempts of obfuscation. (default: MAX)

Java API

The new API makes it easy to integrate JPlag's plagiarism detection into external Java projects:

Language language = new de.jplag.java.Language();
Set<File> submissionDirectories = Set.of(new File("/path/to/rootDir"));
File baseCode = new File("/path/to/baseCode");
JPlagOptions options = new JPlagOptions(language, submissionDirectories, Set.of()).withBaseCodeSubmissionDirectory(baseCode);

JPlag jplag = new JPlag(options);
try {
    JPlagResult result = jplag.run();
     
    // Optional
    ReportObjectFactory reportObjectFactory = new ReportObjectFactory();
    reportObjectFactory.createAndSaveReport(result, "/path/to/output");
} catch (ExitException e) {
    // error handling here
}

Contributing

We're happy to incorporate all improvements to JPlag into this codebase. Feel free to fork the project and send pull requests. Please consider our guidelines for contributions.

Contact

If you encounter bugs or other issues, please report them here. For other purposes, you can contact us at [email protected] . If you are doing research related to JPlag, we would love to know what you are doing. Feel free to contact us!

More information can be found in our Wiki!