CloneWorks vs copydetect

CloneWorks prepares source code into fragments and detects code clones in those fragments. It supports Linux, macOS, and Windows, with a token-based detection method and repository-level scan scope. Its input builder supports Java, C, C++, C#, and Python, and can extract blocks, functions, or files. The detector accepts a minimum similarity threshold from 0.0 to 1.0; the repository describes a Type-3 clone-detection test run. When an index exceeds available memory, a partitioned partial-index mode is available, though the README says its partition size must be determined experimentally. CloneWorks requires FreeTXL 10.6 or later, gcc, make, and Java 7 or later. The input builder runs more slowly on Cygwin because process creation is slower on Windows. It is free to use, with acknowledgement required; copies, modifications, and derivatives must remain under the license and be distributed without charge. The license also provides the software as-is without warranty.

copydetect is a free tool for comparing code directories and producing an HTML report that highlights matching code slices. It uses Winnowing document fingerprinting, an approach also used by MOSS, and Pygments tokenization for the 500-plus languages Pygments can detect and tokenize. Users can compare test directories with reference directories such as previous submissions, and exclude boilerplate directories from matches. Noise, guarantee and display thresholds adjust what is detected and shown in reports. Install the package with pip; it requires Python 3.7 or later. The command-line tool recursively searches supplied directories, while its Python API supports comparisons and HTML report generation. Reports can be saved to a specified path, highlighted code can be truncated, and custom CSS can be linked. Configuration is available through command-line arguments or a JSON file. The package is MIT-licensed and classified for Education with Beta development status. Its documentation cautions that it may not catch plagiarism designed to evade detection and that a match does not establish that plagiarism occurred.