simple in memory full text search solution

Question

I have a small website running on Java with probably a dozen of markdown files. I want to provide a full text search for user to quickly access those markdown files. Since it is small I can afford building up the index in memory each time the web app is started. Any suggestion?

Note

1. I would like to stay away from any database solution, sql or nosql.

2. I prefer the solution is provided as a library rather than builting into an XX framework

Answer 1

Use one of the in-memory databases, either H2 or HSQLDB. Then, for the full text search part, just use Hibernate Search. It will work with either of the two DBs and it will keep you from having to deal with Lucene at all: you can just annotate your entities, and go: all the indexing will happen automatically, and if you want to do things like boost fields, you can do that with a simple annotation.

Answer 2

As a side project I have implemented a simple in memory text search solution for java.

https://github.com/bradforj287/SimpleTextSearch

Key Features:

• Inverted Index
• Cosine Similarity algorithm w/ TFIDF ranking
• MultiThreadded index creation and searching
• Word Stemming (snowball stemmer)
• Strips HTML tags automatically
• Stop words
• String tokenizer (Stanford NLP)

Might want to take a look.

Answer 3

Drop in Apache Lucene, the more-or-less gold standard in full-text search. It is happy to operate in memory.