Efficient Call Path Detection for Android-OS Size of Huge Source Code

摘要

Today most developers utilize source code written by other parties. Because the code ismodified frequently, the developers need to grasp the impact of the modification repeatedly. Acall graph and especially its special type, a call path, help the developers comprehend themodification. Source code written by other parties, however, becomes too huge to be held inmemory in the form of parsed data for a call graph or path. This paper offers a bidirectionalsearch algorithm for a call graph of too huge amount of source code to store all parse results ofthe code in memory. It refers to a method definition in source code corresponding to the visitednode in the call graph. The significant feature of the algorithm is the referenced information isused not in order to select a prioritized node to visit next but in order to select a node topostpone visiting. It reduces path extraction time by 8% for a case in which ordinary pathsearch algorithms do not reduce the time.