linhfishcr7/django-earthdistance

Using PostgreSQL's EarthDistance extension with django

django-earthdistance

Using PostgreSQL's EarthDistance extension for django 1.11, 2.2 and 3.2 (for older versions see with_djorm_expressions branch)

Earthdistance allows to do fast geolocalized queries without using PostGIS

Usage

Cube and EarthDistance extensions must be enabled in postgreSQL BD, so log in database using pgsql and install extensions:

=> create extension cube;
=> create extension earthdistance;

Filter by rows inside a circunference of radius r

from django.db import models

from django_earthdistance.models import EarthDistanceQuerySet

class MyModel(models.Model):
    latitute = models.FloatField()
    longitude = models.FloatField()

    objects = EarthDistanceQuerySet.as_manager()

# Define fields to query in DistanceExpression initialization
# search with lat=0.2546 and lon=-38.25 and distance 1500 meters
# use param `annotate` to set a custom field for the distance, `_ed_distance` as default

MyModel.objects.in_distance(1500, fields=['latitude', 'longitude'], points=[0.2546, -38.25])

Annotate each row returned by a query with distance between two points

from django_earthdistance.models import EarthDistance, LlToEarth

MyModel.objects.filter(....).annotate(
    distance=EarthDistance([
        LlToEarth([0.2546, -38.25]),
        LlToEarth(['latitude', 'longitude'])
    ]))

Optimizing perfomance with indexes

PostgreSQL allow to use GiST indexes with functions results, a good perfomance improvement is to store [ll_to_earth]{.title-ref} results in an index, [ll_to_earth]{.title-ref} is a function that calculates the position of a point on the surface of the earth (assuming earth is perfectly spherical)

-- Example MyModel table is app_mymodel and points columns are latitude and longitude
CREATE INDEX mymodel_location ON app_mymodel USING gist (ll_to_earth(latitude, longitude));

For django < 1.7

Also, using south is preferred, just add this migration to migrations/ folder and edit it to your needs, index will be created

class Migration(SchemaMigration):

    def forwards(self, orm):
        cursor = connection.cursor()
        cursor.execute("CREATE INDEX mymodel_location ON app_mymodel USING gist (ll_to_earth(latitude, longitude));")


    def backwards(self, orm):
        # Deleting field 'Venue.coords'
        cursor = connection.cursor()
        cursor.execute("DROP INDEX mymodel_location ON app_mymodel;")