README

What is this repository for?

Address Index is a Play Framework (2.8.8) application which matches addresses.

The system works via large Elasticsearch (7.17.1) indices build primarily from AddressBase Premium data.

The input can be a complete address (from any source), and the system uses advanced data science techniques to determine the most likely matching AddressBase entries with UPRNs (Unique Property Reference Numbers).

Addresses can be matched one at a time or in batches.

Additional functions exist for postcode searching and partial address string matching for typeaheads.

Plans to deploy the application as a service available to all members of the Public Sector Mapping Agreement have been put on hold for the duration of the Census test. The support team are awaiting a decision on the future of this.

How do I get a sneek peak of the API?

Docker images have been created which give a feel for the API. The Docker images include an Elasticsearch 7.17.1 cluster prebuilt with the required Address Index indices. The indices are a subset of AddressBase (the Exeter subset). Another Docker image contains a version of the API that will work with the Elasticsearch indices. We have recently added a third image to deploy the improved Python UI. The docker-compose.yml file also contains a Kibana and Cerebro service to view the cluster. Unfortunately none of the current images are publicly available at the moment, but we hope to remedy this soon. To get started:

1. Run docker-compose up on https://github.com/ONSdigital/address-index-api/blob/develop/docker-compose.yml

2. The cluster status can be viewed with either Cerebro or Kibana:

    Cerebro: http://localhost:1234
    and then http://es:9200
   
    Kibana: http://localhost:5601
   

3. To run API calls open a browser or API testing app, Postman etc.

    http://localhost:9001/
   

   Endpoints can be found in the Swagger definition /openapi/swagger.json. View as HTML with examples here:

   https://github.com/ONSdigital/address-index-api/blob/develop/api-definitions

   The ai-swagger.json can be copied into the Swagger Editor to view it:

   https://editor.swagger.io/

4. The UI points to the local API and is available on

    http://localhost:5000/
   

How do I get set up for development?

1. Required Installations

   • Java 11
   • sbt 1.5.5
   • Scala 2.13.6
   • Elasticsearch 7.17.1
   • An IDE such as IntelliJ is recommended

2. Create Project from GitHub (IntelliJ shown as example)

   • File, New, Project From Version Control, GitHub
   • Git Repository URL - select https://github.com/ONSdigital/address-index-api
   • Parent Directory: any local drive, typically your IdeaProject directory
   • Directory Name: address-index-api or address-index-data
   • Clone

   The references in the build.sbt are used to draw down additional components

3. Run

   • The project consists of an assembly of several subprojects - server, parsers, model and demo-ui

   • The list of sub projects can be seen by running sbt projects from the root of the project.

   • The list contains the project IDs that must be used for all sbt commands which require a Project ID to be supplied, for example:

   • sbt "project address-index-server" run

   • The application.conf of the demo-ui can point to the API on localhost or a deployed copy of the API

   • The application.conf of the server project points to an elastic search endpoint, this can be local or a server

   To run or test the demo-ui and server together on your local machine:

   Open two command windows running sbt as shown above, one for the API and one for UI

   Use run 9001 to have the API on port 9001 and run 9000 to have the UI on 9000 (i.e. from the root of the address-index-api project run the following commands : sbt "project address-index-server" "run 9001" sbt "project address-index-demo-ui" "run 9000" ) If the UI's application.conf is set to look at localhost:9001 for the API it will work.

   Note that when working on the UI you can save changes and it will autodeploy. This doesn't work with the API becuase of the CRFSuite executable. You have to exit out of sbt and rerun.

How do I run unit tests

sbt test

will run them all, or you can select a subproject, or use testOnly feature to restrict what is run.

How do I run performance tests

See Running Performance Tests

Related Repos

Address Index Data - Apache Spark job used to create the Elasticsearch index

Address Index Developers - Flask web site for API users

Address Index UI - New Flask UI to replace current demo-ui

What if I just want to use the API

See API Help and Swagger