Geocoding address involving non-ASCII characters in PyQGIS

Question

I have installed the "GeoCoding" Plugin in QGIS and it worked well even though I have used non-ASCII characters to geocode (i.e. Hacettepe Üniversitesi). However, when I tried to do the same in PyQGIS with the following code I got the error:

"UnicodeEncodeError: 'ascii' codec can't encode character"

The problem is that the returned address contains non-ASCII characters. Any ideas to solve this?

from geopy.geocoders import Nominatim

#Next, we'll create our geocoder:
geocoder = Nominatim()

#geocode an address:
location = geocoder.geocode("Hacettepe Üniversitesi")

print location

Another relevant question: I rely on Joel Lawhead's book ('QGIS Python Programming Cookbook') and the import statement relevant to this example in the book is:

from GeoCoding.geopy.geocoders import Nominatim

However, this statement gave the following error

ImportError: No module named geopy.geocoders

Is there a source we could follow to make these import statements without a try-and-guess?

Answer 1

IMHO the issue with non-ASCII characters shall not arise anymore, due to improvements that were done in software since 2017 year. Perhaps the problem was about the PyQGIS's version or OS settings.

Nowadays the geoding of 'Hacettepe Üniversitesi' object works well with both GeoPy (as was mentioned by @ivan) and "GeoCoding" Plugin using PyQGIS.

1. GeoPy in QGIS

   Using this code

   from geopy.geocoders import Nominatim
   
   geocoder = Nominatim(user_agent="MyGeocoder", timeout=10)
   
   location = geocoder.geocode("Hacettepe Üniversitesi", exactly_one=False)
   
   print(location.raw)
   

   I was able to get this result, see the image below

   [Location(Hacettepe Üniversitesi Beytepe Kampüsü, Hacettepe-Beytepe Kampüs Yolu, Üniversiteler Mahallesi, Çankaya, Ankara, İç Anadolu Bölgesi, 06800, Türkiye, (39.867445950000004, 32.73548807525388, 0.0)), Location(Hacettepe Universitesi, Hacettepe-Beytepe Kampüs Yolu, Üniversiteler Mahallesi, Çankaya, Ankara, İç Anadolu Bölgesi, 06800, Türkiye, (39.8688599, 32.7356556, 0.0))]

   

2. GeoCoding in QGIS

   This time using this code

   from qgis import utils
   geocoding_plugin = utils.plugins['GeoCoding']
   geocoding_plugin.geocode()
   

   I achieved this result, see the image below

   

Answer 2

Since QGIS 3.18, no need for third party to geocode from Nominatim or Google Maps (not demonstrated below as you need an API key) as there is a class dedicated for this purpose in PyQGIS

Below, a recipe for Nominatim

import json
from qgis.core import (
    QgsNominatimGeocoder,
    QgsGeocoderContext,
    QgsCoordinateTransformContext,
    QgsGeocoderInterface
)

n = QgsNominatimGeocoder()
context = QgsGeocoderContext(QgsCoordinateTransformContext())

address = '28 rue paul bellamy, 44000 Nantes'
out = n.geocodeString(address, context)

getter_methods_geocoder_result = [
    'additionalAttributes',
    'crs',
    'geometry',
    'isValid',
    'description',
    'error',
    'group',
    'identifier'
]

for f in out:
    for method in getter_methods_geocoder_result:
        print(method, getattr(f, method)())

For reverse geocoding, there is an interface method geocodeFeature in parent class QgsGeocoderInterface of both QgsNominatimGeocoder and QgsGoogleMapsGeocoder. You need to implement it and change flags to set that your geocoder supports both geocoding and reverse geocoding

A "naive" reverse geocoding implementation below

import json
from qgis.core import (
    QgsNominatimGeocoder,
    QgsGeocoderContext,
    QgsCoordinateTransformContext,
    QgsGeocoderInterface
)

class QgsNominatimGeocoderOverloaded(QgsNominatimGeocoder):
    def __init__(self, endpointReverse="https://nominatim.openstreetmap.org/reverse"):
        super().__init__()
        self.endpointReverse = endpointReverse
    def flags(self):
        return QgsGeocoderInterface.Flag.GeocodesStrings & QgsGeocoderInterface.Flag.GeocodesFeatures
    def geocodeFeature(self, feature, context, feedback=None):
        pt = feature.geometry().asPoint()
        lon, lat = pt.x(), pt.y()
        url = f"{self.endpointReverse}?lat={lat}&lon={lon}&format=json"
        request = QgsBlockingNetworkRequest()
        request.get(QNetworkRequest(QUrl(url)))
        reply = request.reply()
        content = reply.content()
        jsonContent = json.loads(content.data().decode())
        return self.jsonToResult(jsonContent)

context1 = QgsGeocoderContext(QgsCoordinateTransformContext())

if len(iface.activeLayer().selectedFeatures()) > 0:
    firstSelected = iface.activeLayer().selectedFeatures()[0]

n1 = QgsNominatimGeocoderOverloaded()

out1 = n1.geocodeFeature(firstSelected, context1)

getter_methods_geocoder_result = [
    'additionalAttributes',
    'crs',
    'geometry',
    'isValid',
    'description',
    'error',
    'group',
    'identifier'
]


for method in getter_methods_geocoder_result:
    print(method, getattr(out1, method)())

To account for rate limit, cache (avoid two calls with same URLs in a short interval to avoid being "banned"), a more complete code should be coded using code similar to geocodeString implementation e.g https://qgis.org/api/qgsnominatimgeocoder_8cpp_source.html#l00072

Answer 3

Here is one way to do it:

• I used Ubuntu with Jupyter notebook, but it should work with any Jupyter environment with Python 3

• pip install geopy

  after install, make that line as comment to avoid reinstall

• from geopy.geocoders import Nominatim

• geocoder = Nominatim(user_agent="http")

  The line above sets user agent to avoid the error you were getting as without agent, it violates this https://operations.osmfoundation.org/policies/nominatim/

• location = geocoder.geocode("Hacettepe Üniversitesi")

• print(location)

Result is

Hacettepe Üniversitesi Beytepe Kampüsü, Hacettepe-Beytepe Kampüs Yolu, Bilkent, Üniversiteler Mahallesi, Çankaya, Ankara, İç Anadolu Bölgesi, 06800, Türkiye

Added link to colab where you can see it in action 
https://colab.research.google.com/drive/1GX29YuMwsehPGL-WcQkHtyh-0w97rpoN?usp=sharing