In this section we walk through the sources behind the basic Lucene Web Application demo. Where to find it, its parts, and their function. This section is intended for Java developers wishing to understand how to use Jakarta Lucene in their applications or for those involved in deploying web applications based on Lucene.

Relative the directory created when you extracted Lucene or retreived it from CVS, you should see a directory called "src" which in turn contains a directory called "jsp". This is the root for all of the Lucene web demo.

Within this directory you should see the index.jsp class. Bring this up in vi or your editor of choice.

This jsp page is pretty boring by itself. All it does is include a header, display a form and include a footer. If you look at the form, it has two fields: query (where you enter your search criteria) and maxresults where you specify the number of results per page. If you look at the form tag, you'll notice it uses the get method as opposed to the post. While this is considered deprecated functionality by the latest w3c specs, its unlikely to go away due to the usefulness of being able to bookmark things like searches. By the structure of this JSP it should be easy to customize it without even editing this particular file. You could simply change the header and footer. Let's look at the header.jsp (located in the same directory) next.

The header is also very simple by itself. The only thing it does is include the configuration.jsp (which you looked at in the last section of this guide) and set the title and a brief header. This would be a good place to put your own custom HTML to "pretty" things up a bit. We won't cover the footer because all it does is display the footer and close your tags. Let's look at the results.jsp, the meat of this application next.

The results.jsp had a lot more functionality. Much of it is for paging the search results we'll not cover this as its commented well enough. It does not peform any optimizations such as caching results, etc. as that would make this a more complex example. The first thing in this page is the actual imports for the Lucene classes and Lucene demo classes. These classes are loaded from the jars included in the WEB-INF/lib directory in the final war file.

You'll notice that this file includes the same header and footer as the "index.jsp". From there the jsp constructs an IndexSearcher with the "indexLocation" that was specified in the "configuration.jsp". If there is an error of any kind in opening the index, it is diplayed ot the user and a boolean flag is set to tell the rest of the sections of the jsp not to continue.

From there, this jsp attempts to get the search criteria, the start index (used for paging) and the maximum number of results per page. If the maximum results per page is not set or not valid then it and the start index are set to default values. If only the start index is invalid it is set to a default value. If the criteria isn't provided then a servlet error is thrown (it is assumed that this is the result of url tampering or some form of browser malfunction).

The jsp moves on to construct a StandardAnalyzer just as in the simple demo, to analyze the search critieria, it is passed to the QueryParser along with the criteria to construct a Query object. You'll also notice the string literal "contents" included. This is to specify the search should include the the contents and not the title, url or some other field in the indexed documents. If there is any error in constructing a Query object an error is displayed to the user.

In the next section of the jsp the IndexSearcher is asked to search given the query object. the results are returned in a collection called "hits". If the length property of the hits collection is 0 then an error is displayed to the user and the error flag is set.

Finally the jsp iterates through the hits collection and displayed properties of the "Document" objects we talked about in the first walkthrough. These objects contain "known" fields specific to their indexer (in this case "IndexHTML" constructs a document with "url", "title" and "contents"). You'll notice that these results are paged but the search is repeated every time. This is an area where optimization could improve performance for large result sets.

There are additional sources used by the web app that were not specifically covered by either walkthrough. For example the HTML parser, the IndexHTML class and HTMLDocument class. These are very similar to the classes covered in the first example, however they have properties sepecific to parsing and indexing HTML. This is beyond our scope; however, by now you should feel like you're "getting started" with Lucene.

There are a number of things this demo doesn't do or doesn't do quite right. For instance, you may have noticed that documents in the root context are unreachable (unless you reconfigure Tomcat to support that context or redirect to it), anywhere where the directory doesn't quite match the context mapping, you'll have a broken link in your results. If you want to index non-local files or have some other needs this isn't supported, plus there may be security issues with running the indexing application from your webapps directory. There are a number of things left for you the implementor or developer to do.

In time some of these things may be added to Lucene as features (if you've got a good idea we'd love to hear it!), but for now: this is where you begin and the search engine/indexer ends. Lastly, one would assume you'd want to follow the above advice and customize the application to look a little more fancy than black on white with "Lucene Template" at the top. We'll see you on the Lucene Users' or Developers' mailing lists!

Please resist the urge to contact the authors of this document (without bribes of fame and fortune attached). First contact the mailing lists. That being said feedback, and modifications to this document and samples are ever so greatly appreciatedThey are just best sent to the lists so that everyone can share in them. Certainly you'll get the most help there as well. Thanks for understanding.