Posted by drm on May 2, 2010 at 10:34pm
Is anybody familiar with this module? I need to do a search that searches node titles and taxonomy only, and not the node body. I have the field indexer module to index the title, but don't see how to exclude the body. One way would be to tell the search to only use the title and taxonomy facets. I suppose another might be to build the index without the body, but I don't see any way to do that.
Comments
Solr
With the ApacheSolr search integration, you have pretty granular control over what gets given weight in a search.
You could also set up a views filter that would do this - expose the title and taxonomy fields as exposed filters for your view.
Depending on your use case, Solr might be overkill, but the functionality it brings is pretty useful.
From what I've read on Faceted Search (and we haven't used it since D5) it's pretty resource intensive.
FunnyMonkey
I tried the faceted search
I tried the faceted search module once on a D6 site on shared hosting, and it was extremely resource-intensive in my experience, too. I am under the impression that either Apache Solr would work better or, if you are using shared hosting or have restrictions on your server configuration, the Search Lucene API would be a better fit.
It's a also cache killer
Also not that faceted search and hierarchical select both seem to use cookies somewhat indiscriminately. They both apparently set cookies on anon users when not needed and don't clean up after themselves by deleting them. This really interferes with caching if you're using a reverse proxy cache like Varnish or Squid. Solr doesn't have that problem.
Thanks for the comments . . .
I've used ApacheSolr elsewhere and like it, but this end-user doesn't have a setup where it can be used and doesn't want to get into a subscription service for it. It's also an e-commerce site, so the focus is presenting a very well-themed and well-controlled and focused search result to the user. Exposing filters doesn't really fit that need. It seems that unless I can prevent the node body from being indexed, this won't work.
Since we're only searching titles and taxonomy terms, both of which aren't very long, I may resort to a custom brute-force search directly of those database elements. This is one of those cases where I can very easily provide the client's needs with a custom module but they prefer contrib, but getting it out of contrib seems problematic at best. I've already told them that if their site gets big or busy, they may have no choice but to switch to ApacheSolr.
But I don't know much about LuceneAPI. Can that do a search that avoids the node body?
I have only read about Search
I have only read about Search Lucene API, not used it, so I don't know if it searches the node body or not (or if that is a configurable option). It sounds as though it's pretty easy to install, and it has has an API for extending it, so it might be a good starting place even if you do have to custom code something.
http://drupal.org/project/luceneapi