Building an environment for automated test code generation using a similar source code search technique

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K20240
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 60050:Software-related
• Research Institution: Kyoto Institute of Technology
• Principal Investigator: Choi Eunjong 京都工芸繊維大学, 情報工学・人間科学系, 助教 (90755943)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 類似ソースコード / テストコード自動生成 / ソフトウェア品質

Outline of Final Research Achievements

This study presents a method and condition for automatically transplanting any given test method while preserving its executability . Furthermore, this study investigates open-source software projects developed in Java. It reveals several relevant findings, including 1,862 projects that can serve as sources for reusing test code, and the majority of test code in the investigated projects have at most two dependencies on the code under test. Based on these findings, this study developed an environment for automated test code generation using a similar source code search technique. Finally, the effectiveness of this environment was evaluated, and its results show that the test code generated using this environment exhibited a smaller number of test failures and higher overall quality.

Free Research Field

ソフトウェア工学

Academic Significance and Societal Importance of the Research Achievements

本環境を用いることで，開発者はテストコードを再利用し，テストコードを生成することで，開発者が理解しやすいかつ高いカバレッジを持つテストコードを生成できると期待される．また，本緩急を使用することで開発者がより迅速にテストを実施できる．最後に，テストコードの再利用と自動生成により，ソフトウェアの品質向上が期待できる．