Searching, Selecting, and Synthesizing Source Code Components

As programmers develop software, they instinctively sense that source code exists that could be reused if found --- many programming tasks are common to many software projects across different domains. oftentimes, a programmer will attempt to create new software from this existing source code, such as third-party libraries or code from online repositories. Unfortunately, several major challenges make it difficult to locate the relevant source code and to reuse it. First, there is a fundamental mismatch between the high-level intent reflected in the descriptions of source code, and the low-level implementation details. This mismatch is known as the concept assignment problem , and refers to the frequent case when the keywords from comments or identifiers in code do not match the features implemented in the code. Second, even if relevant source code is found, programmers must invest significant intellectual effort into understanding how to reuse the different functions, classes, or other components present in the source code. These components may be specific to a particular application, and difficult to reuse.;One key source of information that programmers use to understand source code is the set of relationships among the source code components. These relationships are typically structural data, such as function calls or class instantiations. This structural data has been repeatedly suggested as an alternative to textual analysis for search and reuse, however as yet no comprehensive strategy exists for locating relevant and reusable source code. In my research program, I harness this structural data in a unified approach to creating and evolving software from existing components. For locating relevant source code, I present a search engine for finding applications based on the underlying Application Programming Interface (API) calls, and a technique for finding chains of relevant function invocations from repositories of millions of lines of code. Next, for reusing source code, I introduce a system to facilitate building software prototypes from existing packages, and an approach to detecting similar software applications