This is follow up repository for testing elasticsearch's lucene db size vs a plain java lucene db size linked here.

there are two folders:

• es - a clojure impl for loading up the test file to an elasticsearch server (version 0.20.6). it's really straight forward clojure, that gets resolved to REST calls, so nothing fancy here. i wrote it in clojure becase i didnt feel like writing it in bash, and it would output the same exact rest calls. This is very different (and inefficient) from the code I actually use (I use BulkProcessor etc), but the outcome is the same.
• lucene - this one is written in java, to keep it looking like the code we actually use.

Both projects are using lein2 as their buildtool, so get it from here

after cloning this repo:
the default configuration is that ES is running on 127.0.0.1. if its not, edit the file src/es/es.clj.

cd es
lein run -m es.es

this will upload the data to the lucene server, under the index "test".

its really inefficient, and wait a silly 10 seconds between batchs, so it wont hit connection errors. if you get connection errors, you can increase the wait. this takes about 5 minutes to complete.

this one is much simpler and way faster, it actually looks alot like what i have in productions, so any comments on it will be good too. the code sits in lucene/java/test/Indexer.java.

cd lucene
lein javac
lein run -m test.Indexer

Once done, you will have a new folder lucene/luc-index, which contains the optimized lucene index.

you can see the lucene dependency in project.clj, this is maven competible and actually grabs that version of lucene from the public maven repository.