Method and system for code analysis to identify causes of code smells

A system for code analysis to identify causes of code smells, the system comprising:

• a memory storing instructions;
• one or more input/output (I/O) interfaces;
• and
• one or more hardware processors coupled to the memory via the one or more I/O interfaces, wherein the one or more hardware processors are configured by the instructions to:
• detect the code smells in a first version of an application and a second version of the application using a code smell detection technique, wherein the detected code smells in the first version are stored in a first eXtensible Markup Language (XML) file and the detected code smells in the second version are stored in a second XML file;
• compare the first XML file with second XML file using a XML-difference technique, to classify each code smell among the detected code smells from the first version and the second version of the application to a category among a plurality of categories, wherein the plurality of categories comprise common code smells, added code smells and removed code smells, and wherein
• the common code smells are code smells detected in the first version and the second version, and are tagged as least significant code smells;
• the added code smells are code smells detected in the second version and absent in the first version of the application, wherein the added code smells correspond to at least one of new module additions and modifications to existing functions of the first version of the application, wherein the added code smells associated with the modifications to the existing function of the first version of the application are tagged as significant added code smells;
• and
• the removed code smells are code smells detected in the first version and absent in the second version of the application, wherein the removed code smells correspond to at least one of removal of an existing module and partial removal of code from the existing functions of the first version of the application, wherein the removed code smells associated with the partial removal of code from the existing functions of the first version of the application are tagged as significant removed code smells;
• map from the first XML file of the first version and the second XML file of the second version, files that are associated with the significant added code smells or the significant removed code smells to generate a plurality of pairs of mapped functions corresponding to the significant added code smells or the significant removed code smells of the first version and the second version of the application;
• generate a customized Program Dependence Graph (PDG) of each pair among the plurality of pairs of the mapped functions to determine data dependencies and control dependencies for each pair;
• perform an isomorphic sub-graph mapping on the customized PDG of each pair to identify a similar code in a code associated with each pair, wherein the isomorphic sub-graph mapping on the customized PDG considers a syntactic structure and data flow of the code associated with each pair to:
• generate n-ary tree for each mapped function in each pair by performing edge splitting into equivalence classes, wherein each node of each n-ary tree comprises two child nodes with a left child node as data dependency and a right child node as control dependency;
• and
• perform node comparison and node to node mapping on the generated n-ary trees for each mapped function by comparing corresponding tree vertices to match maximum vertices to determine mapped vertices between the generated n-ary trees;
• and
• compute a function difference between each mapped function of each pair, wherein the function difference comprises at least one of:
• added part of the code, which is cause of the significant added code smells, wherein the added part of the code is obtained by eliminating matched vertices from the second version;
• and
• removed part of the code, which is cause of the significant removed code smells, where the removed part of the code is obtained by eliminating the matched vertices of function from the first version, wherein the computed function difference is analyzed further to identify root causes of the code smells.

One or more one or more non-transitory machine readable information storage media comprising instructions which, when executed by one or more hardware processors, cause the one or more hardware processors to execute a method comprising:

• detecting code smells in a first version of an application and a second version of the application using a code smell detection technique, wherein the detected code smells in the first version are stored in a first eXtensible Markup Language (XML) file and the detected code smells in the second version are stored in a second XML file;
• comparing the first XML file with the second XML file using a XML-difference technique, to classify each code smell among the detected code smells of the first version and the second version of the application to a category among a plurality of categories, wherein the plurality of categories comprise common code smells, added code smells, and removed code smells, and wherein
• the common code smells are code smells detected in the first version and the second version of the application, and are tagged as least significant code smells;
• the added code smells are code smells detected in the second version and absent in the first version of the application, wherein the added code smells correspond to at least one of new module additions and modifications to existing functions of the first version of the application, wherein the added code smells associated with the modifications to the existing function of the first version of the application are tagged as significant added code smells;
• and
• the removed code smells are code smells detected in the first version and absent in the second version of the application, wherein the removed code smells correspond to at least one of removal of an existing module and partial removal of code from the existing functions of the first version of the application, wherein the removed code smells associated with the partial removal of code from the existing functions of the first version of the application are tagged as significant removed code smells;
• mapping from the first XML file of the first version and the second XML file of the second version, files that are associated with the significant added code smells or the significant removed code smells to generate a plurality of pairs of mapped functions corresponding to the significant added code smells or the significant removed code smells of the first version and the second version of the application;
• generating a customized Program Dependence Graph (PDG) of each pair among the plurality of pairs of the mapped functions to determine data dependencies and control dependencies for each pair;
• performing an isomorphic sub-graph mapping on the customized PDG of each pair to identify a similar code in a code associated with each pair, wherein the isomorphic sub-graph mapping on the customized PDG considers a syntactic structure and data flow of the code associated with each pair to:
• generate an n-ary tree for each mapped function in each pair by performing edge splitting into equivalence classes, wherein each node of each n-ary tree comprises two child nodes with a left child node as data dependency and a right child node as control dependency;
• and
• perform node comparison and node to node mapping on the generated n-ary trees for each mapped function by comparing corresponding tree vertices to match maximum vertices to determine mapped vertices between the generated n-ary trees;
• and
• computing a function difference between each mapped function of each pair, wherein the function difference comprises at least one of:
• added part of the code, which is cause of the significant added code smells, wherein the added part of the code is obtained by eliminating matched vertices from the second version;
• and
• removed part of the code, which is cause of the significant removed code smells, where the removed part of the code is obtained by eliminating the matched vertices of function from the first version, wherein the computed function difference is analyzed further to identify root causes of the code smells.