ES-Geonames

Create a Geonames index running locally in Elasticsearch container. The index is stored in a data volume, which can be moved elsewhere and quickly stood up.

To download the Geonames.org gazetter, set up Dockerized Elasticsearch, and load the gazetteer into Elasticsearch, clone this repository and run

bash create_index.sh

The geonames_index/ directory can be compressed and moved elsewhere. To start a new Elasticsearch instance using the prepared index, run

docker run -d -p 127.0.0.1:9200:9200 -v /PATH/TO/geoname_index/data/:/usr/share/elasticsearch/data elasticsearch:7.10.1

where /PATH/TO/geonames_index/data/ is the full path to the decompressed index on your host machine. Depending on how you have Docker configured, you may need to run this command with sudo.

This Geonames index is meant to be used with Mordecai, a full-text geoparser.

Windows Installation

This segment concern windows, for some weird reason, using Git-Bash with create_index.sh will map the volume to PROGRAM FILES\GIT\USR\SHARE\ELASTICSEARCH\DATA -> which mess up the docker volume.

I assume you already have dockerized elasticsearch:7.10.1 on your system, If not: docker pull elasticsearch:7.10.1

To setup this repo on windows:

1. (Terminal) mkdir geoname_index
2. (Terminal) docker run -dp 127.0.0.1:9200:9200 -e "discovery.type=single-node" -v %cd%/geoname_index/:/usr/share/elasticsearch/data elasticsearch:7.10.1 
3. Download these stuff to this directory:
    - https://download.geonames.org/export/dump/allCountries.zip
    - https://download.geonames.org/export/dump/admin1CodesASCII.txt
    - https://download.geonames.org/export/dump/admin2Codes.txt
4. unzip allCountries.zip
5. (Git-Bash) create_index_windows.sh

The geonames_index/ directory can be compressed and moved elsewhere. To start a new Elasticsearch instance using the prepared index, run

docker run -d -p 127.0.0.1:9200:9200 -e "discovery.type=single-node" -v /PATH/TO/geoname_index/:/usr/share/elasticsearch/data elasticsearch:7.10.1

Changes

April 2021: The fields in the Elasticsearch index have changed slightly and now include admin1_name and admin2_name fields. See the mapping for the new field names.