• Fast
• Stable
• Flexible
• Authors reply to emails for help

Use Apache Lucy and be happy!

Thanks dmitri for your valuable response. I just got to know that Lucy only provides sub-set of features Lucene provides. Would like to know if you know any critical features Lucy is lacking. I think my requirement is not complex. I need to index file system periodically.. need full-text search including HTML, DOC, XLS, PDF etc types. Thanks again.

I've never used Lucene, so I cannot compare the two. I use Lucy to index PDF, HTML, DOC, and a several other document types. Converting them into text indexable by Lucy has to be done separately.

I've graduated from reindexing once every few hours using cron job to using Linux::Inotify2 to provide practically instant updates to the index. Surely impressed my $boss...

As long as you're looking into solutions, also consider Search::ElasticSearch. While it's not recommendable as your primary data store, it seems to be the solution currently en vogue for searching documents.

I've looked at Apache Tika for doing the text extraction from various documents but Tika doesn't find any module specifically talking to Tika.

My concern with Apache Lucy this is that there is no update on CPAN after Dec 2014.

Not necessarily a bad thing - indicates stability and/or caution on new releases. It also probably puts it in the top 20% or so of most recently released dists (guessing).

Anyway, the repository shows updates within the last week, so it is clearly being worked on.

Thanks. Indeed, GitHub source repo has recent activities.

What dmitri said. I've used KinoSearch/Lucy for almost as long and can testify that the devs are the best and I know that if you find a real bug, it will be addressed quickly because it happened to me. I've never used Search::Elasticsearch but it looks like a good thing to try as well and possibly easier to work with than Lucy (its API seems a bit higher level) but I can't imagine it's as fast.

Thanks for your feedback. have done some trials on Search::Elasticserach too. Module is working fine and looks good but was facing difficulties in making json for attachment (HTML,PDF, DOC file stored in a file system) mapper through the module (not found enough documentation). Probably I am missing something? I am using elastic 1.7 but upcoming version 2 is getting a lot of changes mainly how they index filesystem (river is deprecaited in 2.0) so probably Perl module will also get some updates also documentation.

Since this got revived I'll chime in. I'm still using KinoSearch, Lucy's dad, on 5.8 no less, because I'm stuck in upgrade Hell and don't have a cc/gcc new enough to compile Lucy yet. It's still working great, in production at hundreds of customer sites in front of tens of thousands of users, as it has for 6 or 7 years.

Thanks. Will do so.