elasticsearch-py spatial example

elastic-spatial.py

from datetime import datetime
from elasticsearch import Elasticsearch


class Program():
	es = Elasticsearch()

	def __init__(self):
		pass

	@staticmethod
	def main():
		Program.spatial_test()

	@staticmethod
	def spatial_test():
		index_name = "my-index"
		doc_type_name = "my-doc"
		mapping = {
			doc_type_name: {
				"properties": {
					"account_email": {"type": "string"},
					"expiration_date": {"type": "date", "format": "YYYY-MM-dd"},
					"text": {"type": "string"},
					"location": {
						"type": "geo_point",
						"fielddata": {
							"format": "compressed",
							"precision": "3m"  # to save memory, use lossy compression and lower precision
						}
					}
				}
			}
		}

		# The indices.create method isn't idempotent hence the check for existence
		if not Program.es.indices.exists(index_name):
			Program.es.indices.create(index_name)
		Program.es.indices.put_mapping(index=index_name, doc_type=doc_type_name, body=mapping)

		doc = {
			"account_email": "[email protected]",
			"expiration_date": "2015-07-15",
			"text": "Hey there. Nice to see you nearby dude!",
			"location": {
				"lat": 33.892265,
				"lon": -118.224154
			}
		}

		# This would be idempotent if I specify the id
		# Since I don't this manifests as a POST and a new document would get created each time with
		# an autogenerated id, hence the delete in the for loop below
		result = Program.es.index(index=index_name, doc_type=doc_type_name, body=doc)
		print(result['created'])

		Program.es.indices.refresh(index=index_name)

		query = {
			"query": { "match_all": {}},
			"filter": {
				"geo_distance": {
					"distance": "5km",
					"location": {
						"lat": 33.895956,
						"lon": -118.203982
					}
				}
			}
		}
		result = Program.es.search(index=index_name, body=query)
		print("Got %d hits: " % result['hits']['total'])

		for hit in result['hits']['hits']:
			print("message from %(account_email)s: %(text)s" % hit["_source"])
			Program.es.delete(index=index_name, doc_type=doc_type_name, id=hit['_id'])


if __name__ == "__main__":
	Program.main()