Hello, I need to add a completely client-side search mechanism to a help system that I'm implementing for a Google Summer of Code project for DocBook. I use an Ant script to run xslts that generate html from DocBook XML. I can use a Java indexer to index the html, but when performing the search, I need to access the index via JavaScript. There is no way of invoking a Java code and NO server-side calls can be used. This allows users to use the documentation even through local filesystem without any problem. I do understand it's a difficult requirement to achieve, but there is no other option. I'm willing to code on low-level with use of Lucene API for this.
My questions are: - Is there a way to build the index with a structure I define? Lucene creates its index in a binary format, which I cannot access using JavaScript. Therefore, I would have to edit Lucene backend classes to have a customized index in a plain text format. If this lets me benefit from Lucene's many strengths (stemming, cjk support), It's well worth my time to do it. So my question is: what java files do I need to edit? I see that the IndexFiles class does the indexing. I need to find what specific classes are involved when writing out the index. - I know that there is no JavaScript implementation available for Lucene QueryParser (or if there's one, please let me know!). So which classes should I study to reimplement them in JavaScript? - Are there any other concerns regarding this approach? I'm not sure that this will ultimately be feasible. I have about one month to do the implementation, so I'm trying to determine if this approach is worth persuing or if I should take another approach. Lucene is a great tool. I'm happy to contribute back whatever I can do in making use of it. Thanks for any help you can give me to get me started in this. /KasunBG
