Interactive fault localization for Python with CharmFL

Abstract

We present a plug-in called “CharmFL” for the PyCharm IDE. It employs Spectrum-based Fault Localization to automatically analyze Python programs and produces a ranked list of potentially faulty program elements (i.e., statements, functions, etc.). Our tool offers advanced features, e.g., it enables the users to give their feedback on the suspicious elements to help re-rank them, thus improving the fault localization process. The tool utilizes contextual information about program elements complementary to the spectrum data. The users can explore function call graphs during a failed test. Thus they can investigate the data flow traces of any failed test case or construct a causal inference model for the location of the fault. The tool has been used with a set of experimental use cases.