Recently I had to set up an intranet search engine to crawl trough thousands of PDF files. There are a ton of commercial solutions (read: \(\)) out there on the market, ranging from Google Search Appliance to IBM’s OmniFind. There are also a few good Open Source engines, such as Apache’s Lucene. The problem is that these are primarily intended for enterprises with server farms full of data. That’s really not what I was looking for. I was looking something simple that was easy to set up and maintain. That’s when I came across Xapian. It’s Open Source and lightweight. Combine Xapian with Omega and you got exactly what I was looking for — A lightweight intranet search engine.
This howto will walk you trough how to set up Xapian with Omega on FreeBSD. The version I used was FreeBSD 8.1, but I’m sure any recent version of FreeBSD (7.x>) will do. Please note that I do expect you to know your way around FreeBSD, so I’m not going to spend time on simple tasks like how to edit files etc. I also assume you already got your system up and running.
I’ve called the path we’re going to index (recursively) ‘/path/to/something’. This can be either a local path or something mounted from a remote server. Also, as you’ll see below, a lot of dependencies are installed. This is to increase the number of file-format Xapian will index. It should be able to index PDF-files, Word-files, RTF-files, in addition to plain-text files.
Let’s get started.
Note: If you don’t have the ports-tree installed (/usr/ports), you can download it by simply running:
portsnap fetch extract
echo -e “\\napache22_enable=\\”YES\\”” >> /etc/rc.conf
Install Xapian with Xapian-Omega
Make sure to uncheck X11 and DRAW
Next we need to edit Apache’s config-file (/usr/local/etc/apache22/httpd.conf)
ScriptAlias /cgi-bin/ “/usr/local/www/apache22/cgi-bin/”
ScriptAlias /cgi-bin/ “/usr/local/www/xapian-omega/cgi-bin/”
We also need to create a new config-file for Xapian. Create the file /usr/local/etc/apache22/Include/xapian.conf
Alias /something /path/to/something
Allow from all
Allow from all
With all Apache configuration being done, let’s fire up Apache:
Create the holding directory
mkdir -p /usr/local/lib/omega/data/
Copy over the templates. For some reason FreeBSD doesn’t do this by default.
cp -rfv /usr/ports/www/xapian-omega/work/xapian-omega-*/templates /usr/local/lib/omega/
We also need to tell Xapian-Omega where to look for the files. Create the file /usr/local/www/xapian-omega/cgi-bin/omega.conf
\# Directory containing Xapian databases:
\# Directory containing OmegaScript templates:
\# Directory to write Omega logs to:
\# Directory containing any cdb files for the $lookup OmegaScript command:
Create a search page. I’ll just use index.html in Apache’s default DocumentRoot (/usr/local/www/apache22/data/index.html).
Match any word
Match all words