Novel and applied algorithms in a search engine for Java code snippets

Abstract

Programmers often look for a snippet, that is, a small piece of example code, to remind themselves of how to solve a problem or to quickly learn about a new resource. However, existing tools such as general-purpose search engines and code-specific search engines do not deal well with searches for snippets. In this chapter, we present a prototype search engine designed to work with code snippets. Our approach is based on using the non-code text on a web page as metadata for the snippet to improve indexing and retrieval. We discuss some implementation issues that we encountered, which lead to lessons learned for others who follow. These issues include: extracting snippets from web pages, selecting and indexing metadata, matching query terms with multiple metadata indexes, and identifying a text summary to be used in the presentations of results.