Published on

Geolocation tutorial project with Geodjango and GIS data to build using REST api.


Boilerplate with Geodjango and Leaflet - learn while using it and deploy to Heroku

I had to make one service using geolocation for users. I searched quite a bit but haven't found any book and there is not quite that much materials online - hence this tutorial with bunch of examples. An alternative for geodjango would be to use external API provider e.g. Google Maps. You can also use OpenLayers but this is only front-end solution. The benefit of geodjango vs openlayers is that you can use it in your models and perform calculations in your backend. In the geodjango template I made for this tutorial I used VueJS with Leaflet package to display data on front-end. You can see it live or deploy with heroku Buttons:


The topic we will cover here. Click to go to code examples:

Plus potential error fixes

Important: We will be using vuejs with geodjango (connecting django rest framework api with javascript) throughout this tutorial - django-rest-framework-gis. This django package is compatible with both currently used django versions - Django 2 and Django 3. For more compatibility check geodjango documentation.

Before starting

In order to work with GeoDjango you need to install PostGIS extension in your PostgreSQL database. If you want to use the SQLite you will need SQLite with SpatiaLite.

Get geolocation from the IP

There are 3 ways of obtaining geolocation based on IP address: (easy) ⚪ Call an external geocoding webservices API that returns location data (done from frontend or backend) (harder) 🟠 Use geo IP database with mapped IP.

First let's get IP address from the django request. Let's install django-ipware

pip install django-ipware

and use it in class views:

from rest_framework.generics import ListAPIView
from rest_framework.response import Response
from rest_framework import status
from ipware import get_client_ip

class ShowUsersIP(ListPIView):
    Sample API endpoint that shows user's IP
    queryset = Message.objects.all()
    serializer_class = UserSerializer

    def get(self, request):
        client_ip, is_routable = get_client_ip(request)
        print('IP_INFO: ',client_ip, is_routable)
        data = {
            'client_ip': client_ip,
            'is_routable': is_routable
        return Response(data, status=status.HTTP_200_OK)

or function based views:

from django.http.response import JsonResponse
from ipware import get_client_ip

def show_users_ip(request):
    client_ip, is_routable = get_client_ip(request)
    data = {
        'client_ip': client_ip,
        'is_routable': is_routable
    return JsonResponse(data)

(The easy one) ⚪

Now that we know how to get the IP of the user we can track it's location. With we get 50k calls on freemium which is more than enough. In order to obtain the location you need to pass the IP to the request:

import sys
import urllib.request
import json


def get_lat_lng_from_ip(ip_address):
    """Returns array that consists of latitude and longitude"""
    URL = '{}?token={}'.format(ip_address, ACCESS_TOKEN)
        result = urllib.request.urlopen(URL).read()
        result = json.loads(result)
        result = result['loc']
        lat, lng = result.split(',')
        return [lat, lng]
        print("Could not load: ", URL)
        return None

location = get_lat_lng_from_ip('')
print('Latitude: {0}, Longitude: {1}'.format(*location))
# Prints: Latitude: 32.7831, Longitude: -96.8067

The response we get and transform to get lat and lng is this:

  "ip": "",
  "city": "Dallas",
  "region": "Texas",
  "country": "US",
  "loc": "32.7831,-96.8067",
  "org": "AS14907 Wikimedia Foundation Inc.",
  "postal": "75270",
  "timezone": "America/Chicago"

Responses vary between providers. Some of other service providers are: Google, OpenCage

Now let's save the latitude and longitude we got from user's IP to his account ! All examples are shown in the boilerplate code. The User model we have there extends Abstract user.

# User model
class User(AbstractUser):
    coordinates = models.PointField(blank=True, null=True, srid=4326)
# Update view
# ...
from rest_framework_gis.pagination import GeoJsonPagination
from backend.accounts.api.serializers import UpdateLocationSerializer
from backend.utils.coordinates import get_lat_lng_from_ip
from rest_framework.response import Response
from django.contrib.gis.geos import GEOSGeometry
from django.contrib.gis.geos import Point
from rest_framework import status
import json

class UpdateLocation(UpdateAPIView):
    """ Updates location of a user using it's IP taken from request"""
    model = User
    serializer_class = UpdateLocationSerializer
    pagination_class = GeoJsonPagination

    def update(self, request, *args, **kwargs):
        """Here we:
        - get location from the IP
        - change coordinates to a randomly
        close one in order to anonymize users location.

        # Using ipware library
        client_ip, is_routable = get_client_ip(request)
        # Using our function created previously in utils
        latitude, longitude = get_lat_lng_from_ip(client_ip)
        if not latitude:
            return Response({'message': 'IP or location was not found.'}, status=status.HTTP_406_NOT_ACCEPTABLE)
        # Anonimizing users location before saving
        latitude, longitude = self.anonymize_location(latitude, longitude)
        print('Users latitude {} and longitude {}'.format(latitude, longitude))

        # For GeoDjango PointField we define y and x
        # more in docs:
        point = Point(float(longitude), float(latitude), srid=4326)

        # Requires at least 1 user in DB (e.g.  admin)
        user_obj = User.objects.all().first()

        # Using partial update to not create a new user obj
        anonymize_profile = UpdateLocationSerializer(user_obj, data={'coordinates': point}, partial=True)
        if not anonymize_profile.is_valid():
            return Response(anonymize_profile.errors, status=status.HTTP_400_BAD_REQUEST)

        pnt = GEOSGeometry(point)
        geojson = json.loads(pnt.geojson)
        return Response({'coordinates': reversed(geojson['coordinates'])}, status=status.HTTP_201_CREATED)

For your purposes there is a high chance you should use the first solution. There is many providers with free tiers offering many calls with high degree of accuracy.

(The hard one) 🟠

This option is based on downloading databases of IP and locations and querying it with the obtained IP of the user. You can check IP2location page to see how to do it. Install python package for querying the DB and download the BINs with data to query.

(for more loading dbsync data check out this realpython django tutorial)

Searching via Address

Here we talk about geocoding and reverse geocoding. You can imagine that although with IP we could have location data in our DB - here, with addresses it is not going to be possible because we need much higher accuracy and the amount of data that is needed to achieve it is a lot bigger. You will be forced to use external service's API.

What if you don't want to use Google Geocoding API? There is a geopy library that provides easy API access to Nominatim geocoding information that is open source. You just need to specify the name when using it. It can be name of your app.

Let's see how easy it is:

>>> from geopy.geocoders import Nominatim
>>> geolocator = Nominatim(user_agent="mysuperapp")
>>> location = geolocator.geocode("Warsaw Culture Palace")
>>> print(location.address)
Pałac Kultury, 1, Plac Defilad, Śródmieście Północne, Warszawa, województwo mazowieckie, 00-110, Polska
>>> print((location.latitude, location.longitude))
(52.2317641, 21.005799675616117)

In order to implement it we need to follow good REST API practices! How to implement it into django search endpoint?

First it's important to follow a good URL naming practices:

In Django we can implement it like this starting from


from django.urls import path
from backend.accounts.api.views import GetCoordinatesFromAddress

app_name = 'accounts'

urlpatterns = [
    # ...
     path("coordinates", # the rest will be in views: ?address=QUERY_ADDRESS
          GetCoordinatesFromAddress.as_view(), name="get-coordinates"),

and in

from geopy.geocoders import Nominatim
# ...

class GetCoordinatesFromAddress(ListAPIView, GeoFilterSet):
    """ Shows nearby Users"""

    def get(self, *args, **kwargs):
        # We can check on the server side the location of the users, using request
        # point = self.request.user.coordinates
        # ?address=QUERY_ADDRESS
        # QUERY_ADDRESS is the information user passes to the query
        QUERY_ADDRESS = self.request.query_params.get('address', None)

        if QUERY_ADDRESS not in [None, '']:
            # here we can use the geopy library:
            geolocator = Nominatim(user_agent="mysuperapp")
            location = geolocator.geocode(QUERY_ADDRESS)
            return Response({'coordinates': [location.latitude ,location.longitude]}, status=status.HTTP_200_OK)
            return Response({'message': 'No address was passed in the query'}, status=status.HTTP_400_BAD_REQUEST)

Now if we call our app's API with:

We get in response body:

  "coordinates": [52.2319581, 21.0067249]

There is a front end example if you want to [try it][] or [see the code][].

Adding new objects on map

In order to add an object with coordinates on the map we need to be able to store this info in DB. For that we need a model with PointField first:

class Profile(models.Model):

    company_name = models.CharField(blank=True, null=True, max_length=100)
    coordinates = models.PointField(blank=False, null=True, srid=4326)
    is_premium = models.BooleanField(blank=False, null=True, default=False)
    updated_on = models.DateTimeField(auto_now=True)

    def __str__(self):
        return '{}'.format(self.nickname)

Then we make a serializer for saving POST requests:

from rest_framework_gis import serializers as geo_serializers
# ...

class ProfileSerializer(geo_serializers.GeoFeatureModelSerializer):
    """ A class to serialize locations as GeoJSON compatible data """
    is_premium = serializers.BooleanField(validators=[required])

    class Meta:
        model = Profile
        geo_field = "coordinates"
        id_field = False # we don't want to return id field in the response
        fields = ["coordinates",
                'is_premium', # to use on front-end e.g. to color pins
                'updated_on' # to use on front-end e.g. to order Profiles

Then we make the endpoint:

class ProfileApiView(CreateAPIView):
    """ Creates a new email user with geolocation and returns
    10 closest people to the person"""
    model = TemporaryProfile
    serializer_class = ProfileSerializer
    pagination_class = GeoJsonPagination
    permission_classes = [permissions.AllowAny]

    # If you need more data operations e.g. randomizing location or
    # reading locale you can also use perform_create method
    def create(self, request, *args, **kwargs):
        """Request data for this endpoint should have the following body:
            "coordinates" : [int, int],
            "is_premium" : Boolean

        # Because we used GeoFeatureModelSerializer we need to prepare
        # the data here to be passed correctly to the serializer
        # other way would be to pass it in a cormat of request['geometry']['coordinates']
        [latitude, longitude] =['coordinates']['coordinates'] = Point(float(longitude), float(latitude), srid=4326)

        profile = ProfileSerializer(
        if profile_serializer.is_valid():
            return Response(profile_serializer, status=status.HTTP_201_CREATED)
            return Response(file_serializer.errors, status=status.HTTP_400_BAD_REQUEST)

Closest object

We are using here annotate distance to get closest User's profile with geodjango.

from django.contrib.gis.db.models.functions import Distance as ClosestDistance
from django.contrib.gis.geos import Point

# ...

class ClosestUserApiView(ListAPIView):
    model = Profile
    serializer_class = ProfileSerializer
    pagination_class = GeoJsonPagination
    permission_classes = [AllowAny]

    def get_queryset(self):
        queryset = []
        radius = 50
        latitude, longitude = request['coordinates'] # e.g. { coordinates : [21, -2] }
        point = Point(float(longitude), float(latitude), srid=4326)

        # We are passing
        closest_user = Profile.objects.annotate(coordinates=ClosestDistance(
            'coordinates', point)).first()
        return queryset

🚧🏗️👷 It will be here soon! -->

List of objects within radius

Here we list objects and use geodjango to order by distance using order_by() method. You can also use filter django method to narrow down the list according to your needs.

from geopy.geocoders import Nominatim
from rest_framework_gis.filterset import GeoFilterSet
from rest_framework_gis import filters as geofilters
from django.db.models import Q
from django.contrib.gis.measure import Distance
# ...

class ListNearbyUsers(ListAPIView, GeoFilterSet):
    """ Shows nearby Users"""
    model = User
    serializer_class = NearbyUsersSerializer
    pagination_class = GeoJsonPagination
    contains_geom = geofilters.GeometryFilter(name='coordinates', lookup_expr='exists')

    def get_queryset(self, *args, **kwargs):
        QUERY_ADDRESS = self.request.query_params.get('address', None)
        queryset = []

        if QUERY_ADDRESS not in [None, '']:
            queryset = User.objects.all()
            # here we can use the geopy library:
            geolocator = Nominatim(user_agent="")
            location = geolocator.geocode(QUERY_ADDRESS)

            # Let's use the obtained information to create a geodjango Point
            point = Point(float(location.longitude), float(location.latitude), srid=4326)
            # and query for 10 Users objects to find active users within radius
            queryset = queryset.filter(Q(coordinates__distance_lt=(
                point, Distance(km=settings.RADIUS_SEARCH_IN_KM))) & Q(is_active=True)).order_by('coordinates')[0:10]
            return queryset
            return queryset

Calculate distance between two points

You can use geodjango to filter by distance between 2 points. It's a good point fields example.

from django.contrib.gis.geos import GEOSGeometry

# Get lat and lng from e.g. request or DB User model
latitude1, longitude1

# Get data to compare
latitude2, longitude2

# Prepare geometry points
point1 = GEOSGeometry('SRID=4326;POINT({} {})'.format(latitude1, longitude1))
point2 = GEOSGeometry('SRID=4326;POINT({} {})'.format(latitude2, longitude2))

point1.distance(point2) * 100

You can also use Point fields:

point_field1 = Point(float(longitude1), float(latitude1), srid=4326)
point_field2 = Point(float(longitude2), float(latitude2), srid=4326)

point1 = GEOSGeometry(point_field1)
point2 = GEOSGeometry(point_field2)

point1.distance(point2) * 100

Read more about distance in docs

Deploy geodjango to production on heroku

You can use the heroku button on the top of this article to get your own geodjango app. In case you would like to have more custom solution or want to just do the process of deploying yourself you need to have special heroku buildpack for geodjango. You may have found this one but it's deprecated. Just 4 days after starting writing this post heroku withdraws support for it.

Current solution to deploy your geodjango project is:

  1. Add up-to-date buildpack.
  2. Install PostGIS add-on in your DB.

Extras - errors

Some errors you may find on the way:

Geodjango gdal error

django.core.exceptions.ImproperlyConfigured: Could not find the GDAL library (tried "gdal", "GDAL", "gdal3.1.0", "gdal3.0.0", "gdal2.4.0", "gdal2.3.0", "gdal2.2.0", "gdal2.1.0", "gdal2.0.0"). Is GDAL installed? If it is, try setting GDAL_LIBRARY_PATH in your settings.

It can be caused by:

  • Wrong installation of gdal
  • Not using env variables for gdal
  • In this tutorial you won't get this error if you follow deploying according to the instructions or with the heroku button

Geodjango with sqlite cannot be used due to extension incompatibility for postgis.

Install other relational database management system.

Don't forget about using correct django DB engine.

DATABASES['default'] ={ 'ENGINE': 'django.contrib.gis.db.backends.postgis',
        # ...

toml.TomlDecodeError: Invalid date or number

You may have specified python package version with wrong format in Procfile

e.g. with == not = "==v"

AttributeError: 'DatabaseOperations' object has no attribute 'geo_db_type'

You probably have forgotten to put correct django DB engine in

DATABASES['default'] ={ 'ENGINE': 'django.contrib.gis.db.backends.postgis',
        # ...

Thank you for reading!

Did you make any mistakes when using GEODJANGO or you've seen one here? Tell me about your insights. Leave a comment with YOUR opinion. You are most welcome to see more posts of this type just go to home page