Some tips to implement this in Lucene

[Hussein Baghdadi]

Hi.
I collect many Java sources from the web (java.sun.com , java.net , dw , dev2dev ...) and it became really painful to search for a specific title.
So I decided to build my own search utility based upon SWT and Lucene and my question is about Lucene (I'm not expert in this IR engine).
This application should be to recognize if I added/removed articles to/from where I store my resources.
Which means at least to me that this application should update its index repository each time it run.
1. How to implement this in Lucene (to update its repository) ?
2. Any other ideas or suggestions ?

[Ulf Dittmer]

What platform are you working on? On Unix/Linux, Windows and OS X it would be relatively easy to put together a script that searches through multiple directories.

If you're set to use Lucene, I'd rather create the index from scratch every night or so, instead of trying to keep track of which file may have changed and updating the index every time. While it's possible to update indices with Lucene, the previously added version must be removed first, which introduces some complexities.

[Hussein Baghdadi]

Windows XP SP2

[Ulf Dittmer]

You can install Unxutils, which will give you many of the Unix command line utilities. grep in particular would be very helpful in performing searches like you describe.

Another option (which does not require to install Unxutils) would be AGREP, which can also perform approximate searches. That's handy if one can't remember the exact title to search for.