[Xapian-discuss] PHP functions in Ubuntu packages

[Francis Irving]

On Sun, Nov 19, 2006 at 05:46:47AM -0500, Jim Lynch wrote:
> Francis Irving wrote:
> >/**
> > * This drupal module links drupal and the result of a xapian search.  
> > * The main point of it is to provide the xapian_do_search() function.  
> > * I've also added some settings, to make it a bit more pluggable.
> > */
> >  
> So you are working on a drupal search using xapian too?  Small world.  
> My clients site take about 5 minutes to search using Drupal search, but 
> the same data takes less than a second using xapian.

Well, we're using both Drupal and Xapian in the same project. 

We're actually not storing the (large and bulky, externally sourced)
data at all in an SQL database, just indexing the XML (well, actually
it is structured HTML, but it may as well be XML) files with Xapian.

Then we're making the returned data appear to Drupal to be normal
nodes, so we can link comments to it and so on using conventional
Drupal modules.

At least, this is the plan. We've only just started. I think that
you might be thinking of a Drupal search module which uses Xapian, but
searches data which is directly stored in Drupal; a slightly different
thing. That could perhaps become a standard Drupal module. What we're
doing probably couldn't.

Oh yes, there's one interesting thing that using Xapian as the main
database has bought up. I'm making more extensive use of "metadata"
(i.e.  prefixed keywords such as Da60pv94) using QueryParser. For some
meta-data would really like to have punctuation in keywords. e.g.
A/60/PV.57 instead of a60pv57. Sometimes the punctuation matters, and
I don't like the hack of replacing it with "x".

Am I correct that keywords can only contain alphanumeric characters,
even if (for example) quoted? Is this something that has bothered
other people, or just me?

I'm quite happy about it for the normal search keywords, it is just in
meta data where I've needed it.

Francis