michoo42/Patchnoisseur

Patchnoisseur是一个CVE × 补丁 × CWE的Parquet数据集，它整合了NVD（国家漏洞数据库）中的每个CVE条目，包括其修复提交的差异（当能找到时）、NVD描述以及相关的CWE（通用弱点枚举）信息（ID、名称、简短描述）。数据集以单个Parquet文件形式提供，包含351,884个CVE，其中25,015个带有真实的git差异。数据涵盖了从CVE-1999到CVE-2026的漏洞，磁盘占用约744 MB（zstd压缩的Parquet格式，分片存储，每片约300 MB）。对于没有可检索补丁的CVE，数据集仍保留其行（has_patch = false，补丁列为空），因此也可用作包含CWE元数据的完整NVD快照。数据集支持漏洞分类、补丁摘要、漏洞定位和安全研究等多种应用场景。Patchnoisseur is a CVE × Patch × CWE Parquet dataset that joins every NVD CVE to its fixing-commit diff (when one could be found), its NVD description, and its associated CWE(s) (id, name, short description) — served as a single Parquet dataset. It contains 351,884 CVEs, with 25,015 having a real git diff attached, covering CVE-1999 to CVE-2026, and occupies ~744 MB on disk (zstd-compressed Parquet, sharded ~300 MB each). CVEs without a retrievable patch are kept (with has_patch = false and empty patch columns), making the dataset also usable as a complete NVD snapshot enriched with CWE metadata. It is designed for use cases such as vulnerability classification, patch summarization, vulnerability localization, and empirical security research.