Using Solr with CouchDB

[Patrick Petermair]

Hi!

I'm currently trying to implement a full text search for CouchDB using
Solr. I went through the tutorial and also some of the examples
(slashdot rss feed import, hsql import,..) within the downloadable
distribution.

Since CouchDB works with REST + plaintext JSON and Solr is looking for
sql queries / xmls (as far as I could gather from the examples), I'm
wondering if I'm using the right tools for the job. Has anyone already
implemented a search for CouchDB with Solr? Any tutorials, links or
sample configs that could help me?

Thanks,
Patrick

[Brendan Grainger]

Hi Patrick,

I don't know much about couch, but if you to return json from solr (which I think couch would understand) you can do that with wt=json in the query string when querying solr. See here for more details: http://wiki.apache.org/solr/SolJSON

HTH a little
Brendan

Hi!
I'm currently trying to implement a full text search for CouchDB using Solr. I went through the tutorial and also some of the examples (slashdot rss feed import, hsql import,..) within the downloadable distribution.
Since CouchDB works with REST + plaintext JSON and Solr is looking for sql queries / xmls (as far as I could gather from the examples), I'm wondering if I'm using the right tools for the job. Has anyone already implemented a search for CouchDB with Solr? Any tutorials, links or sample configs that could help me?
Thanks,
Patrick

[Patrick Petermair]

Hey Brendan!

Thanks for your response.

Post by Brendan Grainger
I don't know much about couch, but if you to return json from solr
(which I think couch would understand) you can do that with wt=json
http://wiki.apache.org/solr/SolJSON

Actually I'm looking for the other way around. I'm trying to get Solr to
index my CouchDB. CouchDB works with a REST API and returns plaintext JSON.
So I'm looking to get JSON into Solr and not out of :)

On the CouchDB wiki I've found a reference to a project "CouchDB Solr2"
which seemed to do exactly what I'm trying to do (full text indexing and
searching with CouchDB), but it is no longer maintained as of January
2009 and cannot be found anymore on github. Maybe it's because there is
now a simple way to do it in Solr and I just haven't found it yet ;)

Patrick

[Markus Jelsma]

Hi,

 

 

Setting up CouchDB-Lucene is quite easy, but you don't want that i guess. You could construct a show function to convert input to Solr accepted XML, should be very straightforward. You just need some program to fetch from CouchDB and push it in Solr.

 

Cheers,
 
-----Original message-----
From: Patrick Petermair <***@openforce.com>
Sent: Wed 28-04-2010 17:45
To: solr-***@lucene.apache.org;
Subject: Re: Using Solr with CouchDB

Hey Brendan!

Thanks for your response.

Post by Brendan Grainger
I don't know much about couch, but if you to return json from solr
(which I think couch would understand) you can do that with wt=json
http://wiki.apache.org/solr/SolJSON

Actually I'm looking for the other way around. I'm trying to get Solr to
index my CouchDB. CouchDB works with a REST API and returns plaintext JSON.
So I'm looking to get JSON into Solr and not out of :)

On the CouchDB wiki I've found a reference to a project "CouchDB Solr2"
which seemed to do exactly what I'm trying to do (full text indexing and
searching with CouchDB), but it is no longer maintained as of January
2009 and cannot be found anymore on github. Maybe it's because there is
now a simple way to do it in Solr and I just haven't found it yet ;)

Patrick

[Patrick Petermair]

Post by Markus Jelsma
Setting up CouchDB-Lucene is quite easy, but you don't want that i
guess.

Yeah, I was thinking about CouchDB-Lucene too (also found it in the
CouchDB wiki). It's not like I HAVE to make it work with Solr. If it
turns out that it's not possible or a pain in the ass, I'll probably go
for the easy way with CouchDB-Lucene.

Patrick

[Markus Jelsma]

Whether you need Solr depends on if you require some features such as highlighting, faceting, more-like-this etc. They will not work with CouchDB-Lucene, nor can you, at this moment, use CoucDB-Lucene behind CouchDB-Lounge although a seperate shard can have a sharded Lucene index, you cannot query them through smartproxyd.

 

You need to know what you want to do with fulltext-search before choosing and join CouchDB's mailinglist if you haven't already.
 
-----Original message-----
From: Patrick Petermair <***@openforce.com>
Sent: Wed 28-04-2010 18:03
To: solr-***@lucene.apache.org;
Subject: Re: Using Solr with CouchDB

Post by Markus Jelsma
Setting up CouchDB-Lucene is quite easy, but you don't want that i
guess.

Yeah, I was thinking about CouchDB-Lucene too (also found it in the
CouchDB wiki). It's not like I HAVE to make it work with Solr. If it
turns out that it's not possible or a pain in the ass, I'll probably go
for the easy way with CouchDB-Lucene.

Patrick

[Uri Boness]

Jumping in late here, but if you're interested, we're currently
implementing a LCF connector for couchdb at JTeam (http://www.jteam.nl)
. We'll make it available on line and try to contribute it back to LCF.
We'll also soon publish a blog post about it as an example of how to
develop custom repository connectors in LCF. (I know it doesn't directly
help you now, but at least it might be something to look forward to).

Cheers,
Uri

Post by Markus Jelsma
Whether you need Solr depends on if you require some features such as highlighting, faceting, more-like-this etc. They will not work with CouchDB-Lucene, nor can you, at this moment, use CoucDB-Lucene behind CouchDB-Lounge although a seperate shard can have a sharded Lucene index, you cannot query them through smartproxyd.
You need to know what you want to do with fulltext-search before choosing and join CouchDB's mailinglist if you haven't already.
-----Original message-----
Sent: Wed 28-04-2010 18:03
Subject: Re: Using Solr with CouchDB

Post by Markus Jelsma
Setting up CouchDB-Lucene is quite easy, but you don't want that i
guess.

Yeah, I was thinking about CouchDB-Lucene too (also found it in the
CouchDB wiki). It's not like I HAVE to make it work with Solr. If it
turns out that it's not possible or a pain in the ass, I'll probably go
for the easy way with CouchDB-Lucene.
Patrick