Targeted execution of Android applications (apps) remains a longstanding challenge for software testing and analysis. Although recent advances in GUI testing have substantially improved app exploration, existing approaches remain largely coverage-driven and struggle to steer execution reliably toward specific methods of interest. This limitation hinders important downstream tasks, such as vulnerability detection, program comprehension, and dynamic analysis. We present GAPS (Graph-based Automated Path Synthesizer), the first approach to bridge static program analysis and dynamic GUI exploration, enabling targeted execution in Android apps. Given a target method, GAPS performs a lightweight static backward traversal of the call graph, enriched with data-flow reasoning, to identify feasible execution paths that lead to the target. It then translates these paths into actionable execution strategies that guide dynamic interactions with the app. We evaluate GAPS on the AndroTest benchmark and show that it reconstructs paths toward 88.24% of the target methods, requiring only 12.67 seconds per app on average. During dynamic execution, GAPS reaches 56.93% of the target methods, outperforming state-of-the-art alternatives. In comparison, the model-based GUI testing tool APE reaches only 11.12%, the hybrid exploration framework GoalExplorer reaches 4.75%, and the recent LLM-driven approach Guardian reaches 34%. To assess scalability in real-world settings, we further evaluate GAPS across the 50 most-downloaded apps on the Google Play Store. GAPS reconstructs execution paths toward 62.03% of the target methods, averaging 278.9 seconds per app, and dynamically reaches 54.80% of them. These results show that combining static path synthesis with guided dynamic exploration provides an effective and scalable solution for targeted method execution in Android apps.

翻译：暂无翻译