Robust Parsing of Cloned Token Sequences

Token-based clone detection techniques are known for their scalability, high recall, and robustness against syntax errors and incomplete code. They, however, may yield clones that are syntactically incomplete and they know very little about the syntactic structure of their reported clones. Hence, their results cannot immediately be used for automated refactorings or syntactic filters for relevance. This paper explores techniques of robust parsing to parse code fragments reported by token-based clone detectors to determine whether the clones are syntactically complete and what kind of syntactic elements they contain. This knowledge can be used to improve the precision of token-based clone detection.