Automatic Rust program defect detection method and system based on feature extraction

The invention discloses a method and system for automatically detecting defects of a Rust program based on feature extraction, and the method comprises the steps: forming each type of unsafe operation data set as the input of feature extraction by analyzing the main use scene of limited unsafe Rust in the Rust and statically analyzing the essential unsafe reason of an unsafe code; analyzing specific reasons, sources and use scenes of unsafety of the unsafe Russes, and constructing a defect data set containing each type of unsafe operation; analyzing the lexical rules of the unsafe operations in the defect data set, and obtaining keywords, function names and operator information of each type of unsafe Rust operations; and designing defect detection based on feature recognition according to the features of the limited class unsafe Rust. The probability of false positive and false negative is effectively reduced, and the defect detection efficiency is effectively improved through the automation technology. 本发明公开