VULDEFF: Vulnerability detection method based on function fingerprints and code differences

The dataset of VULDEFF. For more information, visit our GitHub. Due to the lack of large-scale, real, and uniformly formatted vulnerability datasets in the field of software security, we realize the automatic collection and processing of vulnerability patches and vulnerability source code and build a reliable and continuously updated patch and vulnerability source code dataset. The dataset includes 9,546 patch files of CVE and 24,920 unpatched vulnerable source code files, covering 464 programs, and can be extended to support other open source projects.