Over the past decade, users have come to expect software to be highly intelligent when searching data. It is no longer enough to simply make searches case-insensitive, look for keywords as substrings, or other such basic SQL tricks.
Today, when a user searches the product catalog on an e-commerce site, he or she expects keywords to be evaluated across all the data points. Whether a term matches the model number of a computer or the ISBN of a book, the search should still find all the possibilities. To help the user sort through a large number of results, the search should be smart enough to somehow rank them by relevance.
A search should be able to parse words and understand how they might be connected. If you search for the word development, then the search should somehow understand that this is related to developer, even though neither of the words is a substring of the other.
Above all else, a search should be nice. When we post something in an online forum and mistake the words "there", "they're", and "their", people might only criticize our grammar. By contrast, a search should simply understand our typos and be cool about it! A search is at its best when it pleasantly surprises us, seeming to understand the real gist of what we're looking for better than we understood it ourselves.
The purpose of this book is to introduce and explore Hibernate Search, a software package for adding modern search functionality to our own custom applications, without having to invent it from scratch. Because coders usually learn best by looking at real code, this book revolves around an example application. We will stick with this application as we progress through the book, fleshing it out as new concepts are introduced in each chapter.
The true brain behind this search functionality is Apache Lucene, an open source software library for indexing and searching data. Lucene is an established Java project with a rich history of innovation, although it has been ported to other programming languages as well. It is widely adopted across a variety of industries, with high-profile users ranging from Disney to Twitter.
Lucene is often discussed interchangeably with Apache Solr, a related project. From one perspective, Solr is a standalone search server based on Lucene. However, the dependency relationship can flow both ways. Solr subcomponents are often bundled along with Lucene to enhance its functionality when embedded in other applications.
The following diagram shows the relationship between all of these components:
Ultimately, Hibernate Search serves two roles:
- First, it translates Hibernate data objects into information that Lucene can use to build search indexes
- Going in the other direction, it translates the results of Lucene searches into a familiar Hibernate format
From a programmer's perspective, he or she is mapping data with Hibernate in the usual way. Search results come back in the same form as normal Hibernate database queries. Hibernate Search hides most of the low-level plumbing with Lucene.