AusSearch

A collection of Node JS scripts to create both a MongoDB Database and an Elasticsearch index of all Australian addresses using open data provided by https://data.gov.au

Requirements

To run the provided scripts you will need to have about 30GB of free space and a few things installed on your machine

• Docker (including docker-compose which is not installed with docker by default on Linux)
• Node JS with ES6 support
• wget
• unzip

Getting started

1. Start up MongoDB and Elasticsearch

   Run the following command (on Linux you may need to run with sudo)

   docker-compose up -d
   

   Running this will pull the Docker images for MongoDB and Elasticsearch and start instances of them both in detached mode.

   If you are on Linux and get an error message about the vm.max_map_count and fs.file-max, run the following commands

   sudo sysctl -w vm.max_map_count=262144
   sudo sysctl -w fs.file-max=65536
   

   You can then re-run the docker-compose command above.

2. Install the required NodeJS packages

   Run the following command

   npm install
   

   Running this will install the NodeJS packages used by our scripts

3. Downloading the G-NAF data

   Run the following command

   ./download.sh
   

   Running this will pull down a copy of the G-NAF data from the http://data.gov.au website and unzip the resulting zip file into the data directory.

4. Import the data into MongoDB

   Run the following command

   node import.js
   

   Running this will parse the data files downloaded and insert them into collections in MongoDB.

5. Create indexes in MongoDB

   To make the following steps as quick as possible run the following command

   node create-indexes.js
   

   Running this will create indexes in the MongoDB Database that will make reading data a lot faster in the following steps.

6. Simplify the data

   Run the following command

   node simplify.js
   

   Running this will combine and simplify the data imported in step 4

7. Create Elasticsearch index

   Run the following command

   node elasticsearch-create.js
   

   Running this will create the Elasticsearch index that we can insert our documents into.

8. Indexing our data with Elasticsearch

   Run the following command

   node elasticsearch-import.js
   

   Running this will import our data into Elasticsearch.

Searching data

Once all of the steps above have been run it's time to search our data. Searching for an address is as easy as making a GET request to the following URL:

http://localhost:9200/address/singleAddress/_search?q=ADDRESS_GOES_HERE

If all is successful you should receive a list of matching addresses.

References

Incorporates or developed using G-NAF ©PSMA Australia Limited licensed by the Commonwealth of Australia under the Open Geo-coded National Address File (G-NAF) End User Licence Agreement.