Copydetect

Copydetect is a code plagiarism detection tool that uses the winnowing algorithm (similar to the MOSS platform) to identify copied code slices.

Key features:

• Input: A list of directories containing code.
• Output: An HTML report displaying detected copied slices.
• Language Support: Leverages Pygments for tokenization, supporting over 500 languages.
• Performance: Uses numpy for efficient result generation.

Note: Copydetect is designed to detect likely instances of plagiarism; it is not a guarantee of plagiarism and can be evaded by dedicated users.

Adjust these parameters to control how strictly the tool detects plagiarism:

• Noise Threshold (-n, --noise-thresh): The minimum number of matching characters (after tokenization/filtering) to consider plagiarism. Default is 25. Note that tokenization replaces variable names with V, functions with F, etc.
• Guarantee Threshold (-g, --guarantee-thresh): The number of matching characters for which the detector is guaranteed to find a match. This must be $\ge$ the noise threshold. Default is 30. For maximum speed, set this equal to the noise threshold.
• Display Threshold (-d, --display-thresh): The similarity percentage cutoff for showing results in the HTML report. Default is 0.33.

Install copydetect using pip. Requires Python 3.7 or greater.

pip install copydetect

When performing plagiarism detection, you often need to distinguish between student submissions, previous semesters' work, and provided starter code:

• Reference Directories (-r, --ref-dirs): Use this to provide a list of directories containing files you want to compare the test files against (e.g., submissions from previous years). If not provided, the test directories themselves are used as reference directories.
• Boilerplate Directories (-b, --boilerplate-dirs): Use this to specify directories containing provided starter code. Matches found within boilerplate code will not be flagged as plagiarism.

Example usage:

# Test current semester files against current and previous semesters, excluding boilerplate
copydetect -t PA01_F20 -r PA01_F20 PA01_S20 PA01_F19 -b PA01_Boilerplate

To distinguish between files being checked and files used for comparison:

• Use -r (--ref-dirs) to provide a list of reference directories (e.g., previous semesters' submissions). If not provided, test directories are used as references.
• Use -b (--boilerplate-dirs) to specify directories containing boilerplate code. Matches found within boilerplate code will not be flagged as plagiarism.

copydetect -t PA01_F20 -r PA01_F20 PA01_S20 PA01_F19

To run a basic plagiarism check, use copydetect -t DIRS where DIRS is a space-separated list of directories to search. The tool will recursively search these directories and compare every file against every other file.

To limit the check to specific file extensions, use the -e flag followed by a space-separated list of extensions.

# Search all files in student_code directory
copydetect -t student_code

# Search only for C++ and header files in student_code
copydetect -t student_code -e cc cpp h

Once installed, you can generate reports using the copydetect command.

• Linux/macOS: Use copydetect <directories>.
• Windows: Use copydetect.exe <directories>.
• Alternative (if not in PATH): If the scripts folder is not in your PATH, run the module using py.exe -m copydetect <directories>.

# Standard usage
copydetect <directories>

# Windows alternative if not in PATH
py.exe -m copydetect <directories>

You can run the tool using the copydetect command. On Windows, use copydetect.exe. If the scripts folder is not in your PATH, you can run it as a module:

py.exe -m copydetect

copydetect -t DIRS

Instead of long command-line strings, you can use a JSON configuration file. Specify the file using the -c flag.

Supported JSON keys include:

• test_directories (-t)
• reference_directories (-r)
• boilerplate_directories (-b)
• extensions (-e)
• noise_threshold (-n)
• guarantee_threshold (-g)
• display_threshold (-d)
• force_language (-o)
• same_name_only (-s)
• ignore_leaf (-l)
• disable_filtering (-f)
• disable_autoopen (-a)
• truncate (-T)
• out_file (-O)
• encoding (--encoding)
• css (--css)

copydetect -c configuration.json

Install Copydetect using pip. Ensure you are using Python version 3.6 or greater.

pip install copydetect

Use the -e flag followed by a space-separated list of extensions to limit the search to specific file types.

copydetect -t student_code -e cc cpp h