Search multiple projects using a 3rd-party search engine

[3lliot]

I publish quite a few projects (10 currently), which are related, but not always closely.

I would like users to be able to search across all projects, but I don't want to publish them all in the same place (under a 'master' project), because I think it would confuse navigation, plus the index would be huge, publishing would get complicated, etc. etc.

So, I was thinking about implementing an open-source search engine (such as http://lucene.apache.org/solr/) on the web server.

Has anyone done anything similar, and if so, do you know if I'd be able to provide the search function, and publish the results, within the Flare skin (e.g. in a dedicated project)? I'd rather not have to push people out to a separate site to search.

I presume that it would be difficult/impossible to build the search box into the skin, but I'm happy with having a top-level search page that can be linked to from the toolbar, as long as it is fairly easy to build into a Flare project.

Any tips much appreciated...

[Judith]

I want to do the same thing - how did you get on?

Judith

[RamonS]

I used swish-e a while ago and wrapped the search and the results in PHP scripts. I made it look like Google, but the presentation can be anything you want. You can get swish-e here: http://swish-e.org/ . It is an older tool, but also indexes most PDFs, which is why I used it. Swish-e generates an index file that is used for searching. You can create the index file each time you upload new content or run the file creation as a scheduled task and 'forget' about it. I found it to be working quite well and rather fast, but you need to build a front end for it.
Alternatively, you can use Google search and limit it to your site only, for an example look here: http://www.google.com/support/forum/p/W ... 7781&hl=en . For that to work you need to allow the Google crawler on your site.

[iand]

Hi, I am also looking at something similar but do not have PHP experience so may not be able to use RamonS's suggestion. Anyone else implemented search in this way?
Elasticsearch or Solr seem options. 3lliot what did you go with in the end?