SOURCE CODE SIMILARITY DETECTION USING DIGITAL FINGERPRINTS

摘要

Similarity in source code is identified by searching digital fingerprints representing at least control flow of blocks of programming statements. At least some of the source code is converted into a plurality of respective tokens. Each of the tokens is associated with a plurality of blocks. Tokens are modified by normalizing at least one value in at least one of the blocks and/or by defining at least one abstraction. Thereafter, a representation of control flow is created, and a digital fingerprint representing at control flow of a token is generated. Thereafter, source code within at least one block of a given token is determined and identifiable as being a duplicate of source code stored in a repository by comparing at least one of the generated digital fingerprints and at least one previously generated digital fingerprint.