Implementing full-text search in OpenMRS

This is Shahin here , studying bachelor of software engineering in University of Malaya, Malaysia. I am very interested to contribute to OpenMRS as QA Engineer. I have sound knowledge in multiple programming languages such as Java, PHP, Python and some popular web framework like Spring, Laravel. Recently, I have build-up knowledge in Elastic Search. I think it would be great if I can use that knowledge to implement full-text search functionality in OpenMRS.


OpenMRS uses Apache Lucene to provide indexed search.

You can look at the issues in JIRA and look for related things to work on:"lucene"

Havent found any wiki on the topic, but you could browse the code.

For ex.

Maybe someone else can give some more insight :slight_smile:

@sopnopriyo, glad to heard you’re interested! If you want to work on implementing full-text search, you should call that work development (not QA Engineering).

In OpenMRS Platform 1.11 we started using Lucene for searching concepts. This was a large piece of work, and you should be able to see all the details from the link here:

What we’d like to do is to apply this same approach to patient searching. I see that you came across the ticket, and this is where the work should happen. (That ticket seems to have some very old patches on it, that predate the work on TRUNK-2999, so you can peek at them, but probably don’t use them directly.)

So, at first I suggest you take a look at the work that was done as part of TRUNK-2999, and then ping back again when you have an understanding of it. (Ultimately we’ll want to specify some clear Acceptance Criteria about how we’ll want a lucene-backed patient search to work.)

PS- @sopnopriyo, I moved this message to the general Development category. If you end up wanting to talk about QA, then start another thread in the QA category, but as we’re talking about adding a new feature, this belongs in Dev.

Hi @darius

Thanks for your reply. I will try to understand the code and get back to you on this.

@sopnopriyo, it’s probably worth saying: if you’re not familiar with hibernate this may be a bit tricky; so if you’re running into issues, then please do reach out for help on the IRC channel.

Hi @darius, I am familiar with JDBC template and Hibernate . I am having some issues with setting up the development environment. Once done, I will ping you.

Thanks, Shahin

I recomment using the SDK ( for setting up your dev environment.

Still No luck. Ended up with the following error

I think you are behind a proxy, and you need to configure maven for this. Google should tell you how. (It’s probably documented on some other OpenMRS Talk thread also.)

Hi @darius , Finally I have been able to set up the environment. Besides, I have looked into the trunk you mentioned earlier. Let’s discuss about the task.

Thanks, Shahin

@sopnopriyo, do you have specific questions? (I’m not actually familiar with the code in question, although I know the general idea of what we’re trying to do.)

Can you suggest, based on peeking at the way this was implemented for concepts, how you suggest to implement it for patient search?

Then anybody can reply to specific questions or proposals.

Hi @darius,

Thanks for your reply. The coding part I can manage by myself I guess since there is already some implementations and If I need help I will inform you later for sure .

Let’s discuss about the requirements and overall picture of the task.

Then, I will come up with a plan for implementation and discuss with you.

Thanks, Shahin

@sopnopriyo, sorry for the delay, I was traveling and then at all-day meetings last seek.

I have updated the ticket description on TRUNK-425 to include some Acceptance Criteria describing the work. (I wrote these quickly so they are not perfect!)

Is this good enough to get started? Do you want to discuss further?

@darius, No worries. As a starting point , the supplied information is okay. So, I am going to start with creating index for Patient Entity and will keep you inform time to time :slight_smile:

Thanks, Shahin

One thing to keep in mind: Patient extends Person in our domain model, and really we’d like lucene-based searching with both Patient and Person.

(I haven’t stated this very clearly in the ticket, and I haven’t had time to think through all the implications.)