django-rest-framework-gis

Geographic add-ons for Django Rest Framework - Mailing List.

Install last stable version from pypi

pip install djangorestframework-gis

Install development version

pip install https://github.com/djangonauts/django-rest-framework-gis/tarball/master

Setup

Add rest_framework_gis in settings.INSTALLED_APPS, after rest_framework:

INSTALLED_APPS = [
    # ...
    'rest_framework',
    'rest_framework_gis',
    # ...
]

Compatibility with DRF, Django and Python

DRF-gis version	DRF version	Django version	Python version
0.11.x	3.1 to 3.5	1.7 to 1.10	2.7 to 3.5
0.10.x	3.1 to 3.3	1.7 to 1.9	2.7 to 3.5
0.9.6	3.1 to 3.2	1.5 to 1.8	2.6 to 3.5
0.9.5	3.1 to 3.2	1.5 to 1.8	2.6 to 3.4
0.9.4	3.1 to 3.2	1.5 to 1.8	2.6 to 3.4
0.9.3	3.1	1.5 to 1.8	2.6 to 3.4
0.9.2	3.1	1.5 to 1.8	2.6 to 3.4
0.9.1	3.1	1.5 to 1.8	2.6 to 3.4
0.9	3.1	1.5 to 1.8	2.6, 2.7, 3.3, 3.4
0.9	3.1	1.5 to 1.8	2.6, 2.7, 3.3, 3.4
0.9	3.1	1.5 to 1.8	2.6, 2.7, 3.3, 3.4
0.8.2	3.0.4 to 3.1.1	1.5 to 1.8	2.6, 2.7, 3.3, 3.4
0.8.1	3.0.4 to 3.1.1	1.5 to 1.8	2.6, 2.7, 3.3, 3.4
0.8	3.0.4	1.5 to 1.7	2.6, 2.7, 3.3, 3.4
0.7	2.4.3	1.5 to 1.7	2.6, 2.7, 3.3, 3.4
0.6	2.4.3	1.5 to 1.7	2.6, 2.7, 3.3, 3.4
0.5	from 2.3.14 to 2.4.2	1.5 to 1.7	2.6, 2.7, 3.3, 3.4
0.4	from 2.3.14 to 2.4.2	1.5 to 1.7	2.6, 2.7, 3.3, 3.4
0.3	from 2.3.14 to 2.4.2	1.5, 1.6	2.6, 2.7
0.2	from 2.2.2 to 2.3.13	1.5, 1.6	2.6, 2.7

Fields

GeometryField

Provides a GeometryField, which is a subclass of Django Rest Framework (from now on DRF) WritableField. This field handles GeoDjango geometry fields, providing custom to_native and from_native methods for GeoJSON input/output.

New in 0.9.3: there is no need to define this field explicitly in your serializer, it's mapped automatically during initialization in rest_framework_gis.apps.AppConfig.ready().

GeometrySerializerMethodField

Provides a GeometrySerializerMethodField, which is a subclass of DRF SerializerMethodField and handles values which are computed with a serializer method and are used as a geo_field. See example below.

Serializers

GeoModelSerializer (DEPRECATED)

Deprecated, will be removed in 1.0: Using this serializer is not needed anymore since 0.9.3 if you add rest_framework_gis in settings.INSTALLED_APPS

Provides a GeoModelSerializer, which is a sublass of DRF ModelSerializer. This serializer updates the field_mapping dictionary to include field mapping of GeoDjango geometry fields to the above GeometryField.

For example, the following model:

class Location(models.Model):
    """
    A model which holds information about a particular location
    """
    address = models.Charfield(max_length=255)
    city = models.CharField(max_length=100)
    state = models.CharField(max_length=100)
    point = models.PointField()

By default, the DRF ModelSerializer will output:

{
    "id": 1,
    "address": "742 Evergreen Terrace",
    "city":  "Springfield",
    "state": "Oregon",
    "point": "POINT(-123.0208 44.0464)"
}

In contrast, the GeoModelSerializer will output:

{
    "id": 1,
    "address": "742 Evergreen Terrace",
    "city":  "Springfield",
    "state": "Oregon",
    "point": {
        "type": "Point",
        "coordinates": [-123.0208, 44.0464],
    }
}

GeoFeatureModelSerializer

GeoFeatureModelSerializer is a subclass of rest_framework.ModelSerializer which will output data in a format that is GeoJSON compatible. Using the above example, the GeoFeatureModelSerializer will output:

 {
    "id": 1,
    "type": "Feature",
    "geometry": {
        "point": {
            "type": "Point",
            "coordinates": [-123.0208, 44.0464],
        },
    },
    "properties": {
        "address": "742 Evergreen Terrace",
        "city":  "Springfield",
        "state": "Oregon"
    }
}

If you are serializing an object list, GeoFeatureModelSerializer will create a FeatureCollection:

{
    "type": "FeatureCollection",
    "features": [
    {
        "id": 1
        "type": "Feature",
        "geometry": {
            "point": {
                "type": "Point",
                "coordinates": [-123.0208, 44.0464],
            }
        },
        "properties": {
            "address": "742 Evergreen Terrace",
            "city":  "Springfield",
            "state": "Oregon",
        }
    }
    {
        "id": 2,
        "type": "Feature",
        "geometry": {
            "point": {
                "type": "Point",
                "coordinates": [-123.0208, 44.0489],
            },
        },
        "properties": {
            "address": "744 Evergreen Terrace",
            "city":  "Springfield",
            "state": "Oregon"
        }
    }
}

Specifying the geometry field: "geo_field"

GeoFeatureModelSerializer requires you to define a geo_field to be serialized as the "geometry". For example:

from rest_framework_gis.serializers import GeoFeatureModelSerializer

class LocationSerializer(GeoFeatureModelSerializer):
    """ A class to serialize locations as GeoJSON compatible data """

    class Meta:
        model = Location
        geo_field = "point"

        # you can also explicitly declare which fields you want to include
        # as with a ModelSerializer.
        fields = ('id', 'address', 'city', 'state')

Using GeometrySerializerMethodField as "geo_field"

geo_field may also be an instance of GeometrySerializerMethodField. In this case you can compute its value during serialization. For example:

from django.contrib.gis.geos import Point
from rest_framework_gis.serializers import GeoFeatureModelSerializer, GeometrySerializerMethodField

class LocationSerializer(GeoFeatureModelSerializer):
    """ A class to serialize locations as GeoJSON compatible data """

    # a field which contains a geometry value and can be used as geo_field
    other_point = GeometrySerializerMethodField()

    def get_other_point(self, obj):
        return Point(obj.point.lat / 2, obj.point.lon / 2)

    class Meta:
        model = Location
        geo_field = 'other_point'

Serializer for geo_field may also return None value, which will translate to null value for geojson geometry field.

Specifying the ID: "id_field"

The primary key of the model (usually the "id" attribute) is automatically used as the id field of each GeoJSON Feature Object.

The default behaviour follows the GeoJSON RFC, but it can be disbaled by setting id_field to False:

from rest_framework_gis.serializers import GeoFeatureModelSerializer

class LocationSerializer(GeoFeatureModelSerializer):

    class Meta:
        model = Location
        geo_field = "point"
        id_field = False
        fields = ('id', 'address', 'city', 'state')

The id_field can also be set to use some other unique field in your model, eg: slug:

from rest_framework_gis.serializers import GeoFeatureModelSerializer

class LocationSerializer(GeoFeatureModelSerializer):

    class Meta:
        model = Location
        geo_field = 'point'
        id_field = 'slug'
        fields = ('slug', 'address', 'city', 'state')

Bounding Box: "auto_bbox" and "bbox_geo_field"

The GeoJSON specification allows a feature to contain a boundingbox of a feature. GeoFeatureModelSerializer allows two different ways to fill this property. The first is using the geo_field to calculate the bounding box of a feature. This only allows read access for a REST client and can be achieved using auto_bbox. Example:

from rest_framework_gis.serializers import GeoFeatureModelSerializer

class LocationSerializer(GeoFeatureModelSerializer):
    class Meta:
        model = Location
        geo_field = 'geometry'
        auto_bbox = True

The second approach uses the bbox_geo_field to specify an addional GeometryField of the model which will be used to calculate the bounding box. This allows boundingboxes differ from the exact extent of a features geometry. Additionally this enables read and write access for the REST client. Bounding boxes send from the client will be saved as Polygons. Example:

from rest_framework_gis.serializers import GeoFeatureModelSerializer

class LocationSerializer(GeoFeatureModelSerializer):

    class Meta:
        model = BoxedLocation
        geo_field = 'geometry'
        bbox_geo_field = 'bbox_geometry'

Custom GeoJSON properties source

In GeoJSON each feature can have a properties member containing the attributes of the feature. By default this field is filled with the attributes from your Django model, excluding the id, geometry and bounding box fields. It's possible to override this behaviour and implement a custom source for the properties member.

The following example shows how to use a PostgreSQL HStore field as a source for the properties member:

# models.py
class Link(models.Model):
    """
    Metadata is stored in a PostgreSQL HStore field, which allows us to
    store arbitrary key-value pairs with a link record.
    """
    metadata = HStoreField(blank=True, null=True, default=dict)
    geo = models.LineStringField()
    objects = models.GeoManager()

# serializers.py
class NetworkGeoSerializer(GeoFeatureModelSerializer):
    class Meta:
        model = models.Link
        geo_field = 'geo'
        auto_bbox = True

    def get_properties(self, instance, fields):
        # This is a PostgreSQL HStore field, which django maps to a dict
        return instance.metadata

    def unformat_geojson(self, feature):
        attrs = {
            self.Meta.geo_field: feature["geometry"],
            "metadata": feature["properties"]
        }

        if self.Meta.bbox_geo_field and "bbox" in feature:
            attrs[self.Meta.bbox_geo_field] = Polygon.from_bbox(feature["bbox"])

        return attrs

When the serializer renders GeoJSON, it calls the method get_properties for each object in the database. This function should return a dictionary containing the attributes for the feature. In the case of a HStore field, this function is easily implemented.

The reverse is also required: mapping a GeoJSON formatted structure to attributes of your model. This task is done by unformat_geojson. It should return a dictionary with your model attributes as keys, and the corresponding values retrieved from the GeoJSON feature data.

Pagination

We provide a GeoJsonPagination class.

GeoJsonPagination

Based on rest_framework.pagination.PageNumberPagination.

Code example:

from rest_framework_gis.pagination import GeoJsonPagination
# --- other omitted imports --- #

class GeojsonLocationList(generics.ListCreateAPIView):
    # -- other omitted view attributes --- #
    pagination_class = GeoJsonPagination

Example result response (cut to one element only instead of 10):

{
    "type": "FeatureCollection",
    "count": 25,
    "next": "http://localhost:8000/geojson/?page=2",
    "previous": null,
    "features": [
        {
            "type": "Feature",
            "geometry": {
                "type": "Point",
                "coordinates": [
                    42.0,
                    50.0
                ]
            },
            "properties": {
                "name": "test"
            }
        }
    ]
}

Filters

note: this feature is compatible with django-filter up to version 0.15. If you need compatibility with version django-filter 1.0 please send a patch (see issue #120).

We provide a GeometryFilter field as well as a GeoFilterSet for usage with django_filter. You simply provide, in the query string, one of the textual types supported by GEOSGeometry. By default, this includes WKT, HEXEWKB, WKB (in a buffer), and GeoJSON.

GeometryFilter

from rest_framework_gis.filterset import GeoFilterSet

class RegionFilter(GeoFilterSet):
    slug = filters.CharFilter(name='slug', lookup_type='istartswith')
    contains_geom = filters.GeometryFilter(name='geom', lookup_type='contains')

    class Meta:
        model = Region

We can then filter in the URL, using GeoJSON, and we will perform a __contains geometry lookup, e.g. /region/?contains_geom={ "type": "Point", "coordinates": [ -123.26436996459961, 44.564178042345375 ] }.

GeoFilterSet

The GeoFilterSet provides a django_filter compatible FilterSet that will automatically create GeometryFilters for GeometryFields.

InBBoxFilter

Provides a InBBoxFilter, which is a subclass of DRF BaseFilterBackend. Filters a queryset to only those instances within a certain bounding box.

views.py:

from rest_framework_gis.filters import InBBoxFilter

class LocationList(ListAPIView):

    queryset = models.Location.objects.all()
    serializer_class = serializers.LocationSerializer
    bbox_filter_field = 'point'
    filter_backends = (InBBoxFilter, )
    bbox_filter_include_overlapping = True # Optional

We can then filter in the URL, using Bounding Box format (min Lon, min Lat, max Lon, max Lat), and we can search for instances within the bounding box, e.g.: /location/?in_bbox=-90,29,-89,35.

By default, InBBoxFilter will only return those instances entirely within the stated bounding box. To include those instances which overlap the bounding box, include bbox_filter_include_overlapping = True in your view.

Note that if you are using other filters, you'll want to include your other filter backend in your view. For example:

filter_backends = (InBBoxFilter, DjangoFilterBackend,)

TMSTileFilter

Provides a TMSTileFilter, which is a subclass of InBBoxFilter. Filters a queryset to only those instances within a bounding box defined by a TMS tile address.

views.py:

from rest_framework_gis.filters import TMSTileFilter

class LocationList(ListAPIView):

    queryset = models.Location.objects.all()
    serializer_class = serializers.LocationSerializer
    bbox_filter_field = 'point'
    filter_backends = (TMSTileFilter, )
    bbox_filter_include_overlapping = True # Optional

We can then filter in the URL, using TMS tile addresses in the zoom/x/y format, eg:. /location/?tile=8/100/200 which is equivalant to filtering on the bbox (-39.37500,-71.07406,-37.96875,-70.61261).

For more information on configuration options see InBBoxFilter.

Note that the tile address start in the upper left, not the lower left origin used by some implementations.

DistanceToPointFilter

Provides a DistanceToPointFilter, which is a subclass of DRF BaseFilterBackend. Filters a queryset to only those instances within a certain distance of a given point.

views.py:

from rest_framework_gis.filters import DistanceToPointFilter

class LocationList(ListAPIView):

    queryset = models.Location.objects.all()
    serializer_class = serializers.LocationSerializer
    distance_filter_field = 'geometry'
    filter_backends = (DistanceToPointFilter, )
    bbox_filter_include_overlapping = True # Optional

We can then filter in the URL, using a distance and a point in (lon, lat) format. The distance can be given in meters or in degrees.

eg:. /location/?dist=4000&point=-122.4862,37.7694&format=json which is equivalant to filtering within 4000 meters of the point (-122.4862, 37.7694).

By default, DistanceToPointFilter will pass the 'distance' in the URL directly to the database for the search. The effect depends on the srid of the database in use. If geo data is indexed in meters (srid 3875, aka 900913), a distance in meters can be passed in directly without conversion. For lat-lon databases such as srid 4326, which is indexed in degrees, the 'distance' will be interpreted as degrees. Set the flag, 'distance_filter_convert_meters' to 'True' in order to convert an input distance in meters to degrees. This conversion is approximate, and the errors at latitudes > 60 degrees are > 25%.

Projects using this package

• Nodeshot: Extensible Django web application for management of community-led georeferenced data

Running the tests

Assuming one has the dependencies installed (restframework and restframework_gis), and one of the Spatial Database server supported by GeoDjango is up and running:

./runtests.py

You might need to tweak the DB settings according to your DB configuration. You can copy the file local_settings.example.py to local_settings.py and change the DATABASES and/or INSTALLED_APPS directives there.

If you want to contribute you need to install the test app in a proper development environment.

These steps should do the trick:

• create a spatial database named "django_restframework_gis"
• create local_settings.py, eg: cp local_settings.example.py local_settings.py
• tweak the DATABASES configuration directive according to your DB settings
• optionally install olwidget with pip install olwidget
• uncomment INSTALLED_APPS (remove olwidget if you did not install it)
• run python manage.py syncdb
• run python manage.py collectstatic
• run python manage.py runserver

Contributing

1. Join the Django REST Framework GIS Mailing List and announce your intentions
2. Follow the PEP8 Style Guide for Python Code
3. Fork this repo
4. Write code
5. Write tests for your code
6. Ensure all tests pass
7. Ensure test coverage is not under 90%
8. Document your changes
9. Send pull request