CRII: SHF: Automatic Building of Software Projects to Support Analysis of Open Software Repositories

随着越来越多的软件开发人员选择在Sourceforge和Github等开放软件存储库上托管他们的项目，这些大型存储库已经成为开发新软件工程技术的宝贵资源。它们既可以作为新兴的基于挖掘的软件工程技术的数据集，也可以作为实验评估和实证研究的主题。然而，阻止充分利用开放软件存储库的一个主要障碍是，尽管许多现有的程序分析工具需要成功构建的项目作为它们的输入，软件项目通常只将它们的源代码存储在存储库中，并且源代码通常不能自动构建。因此，通常很难将现有的程序分析工具应用到大量的软件项目中，这限制了大型开放软件存储库提供的好处。本研究项目旨在通过提供在开放软件存储库中自动构建软件项目的技术来克服上述障碍。具体而言，项目主要探索以下方向：(1)分析构建配置文件中的路径表达式和文件流，为构建配置工具提供合适的参数和环境变量；(2)通过在线依赖库搜索解决缺失的软件依赖，并在搜索结果中识别兼容的软件依赖文件；(3)根据软件项目的特点，将软件项目分配到合适的构建平台。该项目有助于发现关于自动软件构建的可行性和有效性的知识，并且可能受益于利用开放软件存储库的各种软件工程技术。

As more and more software developers choose to host their projects at open software repositories such as Sourceforge and Github, these large repositories have become valuable resources for the development of novel software engineering techniques. They can serve as either data sets for the emerging mining-based software-engineering techniques or subjects for experimental evaluations and empirical studies. However, a major obstacle preventing full utilization of open software repositories is that, while many existing program analysis tools require successfully built projects as their input, software projects typically have only their source code stored in the repositories, and the source code often cannot be built automatically. Therefore, it is often difficult to apply existing program analysis tools to a large number of software projects, which limits the benefit provided by large open software repositories. This research project aims to overcome the above obstacle by providing technique for automatic building of software projects in open software repositories. Specifically, the project mainly explores the following directions: (1) analyzing path expressions and file flows in build configuration files to provide proper parameters and environment variables for build configuration tools (2) resolving missing software dependencies by searching online dependency repositories, and identifying the compatible software dependency files among the search results; and (3) assigning a software project to an appropriate building platform according to its features. The project helps to discover knowledge on the feasibility and effectiveness of automatic software building, and may benefit a large variety of software engineering techniques that take advantage of open software repositories.