GeoJson fix property as identifier

For styling and highlights we need a unique field on each geojson feature. This functionality was broken for geojson data with certain properties layouts. This change makes the functionality work for all properties allowed by the geojson spec.

Author: Conengmo

folium/features.py

@@ -25,6 +25,7 @@
     none_min,
     get_obj_in_upper_tree,
     parse_options,
+    dict_get,
 )
 from folium.vector_layers import PolyLine, path_options
 
@@ -521,17 +522,29 @@ def _validate_function(self, func, name):
                              .format(name))
 
     def find_identifier(self):
-        """Find a unique identifier for each feature, create it if needed."""
+        """Find a unique identifier for each feature, create it if needed.
+
+        According to the GeoJSON specs a feature:
+         - MAY have an 'id' field with a string or numerical value.
+         - MUST have a 'properties' field. The content can be any json object
+           or even null.
+
+        """
         features = self.data['features']
         n = len(features)
         feature = features[0]
+        # Each feature has an 'id' field with a unique value.
         if 'id' in feature and len(set(feat['id'] for feat in features)) == n:
             return 'feature.id'
-        for key in feature.get('properties', []):
-            if len(set(feat['properties'][key] for feat in features)) == n:
-                return 'feature.properties.{}'.format(key)
+        # Each feature has a unique string/float/int property.
+        #   'properties' may be missing or None:
+        if isinstance(feature.get('properties', None), dict):
+            field_name = self._search_properties(features, 'properties')
+            if field_name is not None:
+                return field_name
+        # We add an 'id' field with a unique value to the data.
         if self.embed:
-            for i, feature in enumerate(self.data['features']):
+            for i, feature in enumerate(features):
                 feature['id'] = str(i)
             return 'feature.id'
         raise ValueError(
@@ -540,6 +553,30 @@ def find_identifier(self):
             'field to your geojson data or set `embed=True`. '
         )
 
+    @classmethod
+    def _search_properties(cls, features, *keys):
+        """Find a property for which each feature has a unique str/num value."""
+        value_first_feature = dict_get(features[0], *keys)
+        # Recursively look through the dictionary to find a unique value.
+        if isinstance(value_first_feature, dict):
+            for key in value_first_feature:
+                field_name = cls._search_properties(features, *[*keys, key])
+                if field_name is not None:
+                    return field_name
+        # Check that all features have these keys and that the values are unique.
+        unique_values = set()
+        for feat in features:
+            try:
+                value = dict_get(feat, *keys)
+            except TypeError:
+                # not all features have the same properties layout
+                return None
+            if not isinstance(value, (str, float, int)):
+                return None
+            unique_values.add(value)
+        if len(unique_values) == len(features):
+            return '.'.join(['feature', *keys])
+
     def _get_self_bounds(self):
         """
         Computes the bounds of the object itself (not including it's children)

tests/test_features.py

@@ -12,7 +12,7 @@
 from branca.element import Element
 
 import folium
-from folium import Map, Popup
+from folium import Map, Popup, GeoJson
 
 import pytest
 
@@ -213,3 +213,72 @@ def test_geojson_tooltip():
         warnings.simplefilter('always')
         m._repr_html_()
         assert issubclass(w[-1].category, UserWarning), 'GeoJsonTooltip GeometryCollection test failed.'
+
+
+def test_geojson_search_properties():
+    features = [{'properties': None} for _ in range(3)]
+    assert GeoJson._search_properties(features, 'properties') is None
+    features = [{'properties': {'hi': 'there'}},
+                {'properties': {'hi': 'what'}}]
+    assert GeoJson._search_properties(features, 'properties') == 'feature.properties.hi'
+    features = [{'properties': {'hi': {'more': 'some value'}}},
+                {'properties': {'hi': {'more': 'another value'}}}]
+    assert (GeoJson._search_properties(features, 'properties')
+            == 'feature.properties.hi.more')
+    features = [{'properties': {'hi': 'there'}},
+                {'properties': {'hi': 'there'}}]
+    assert GeoJson._search_properties(features, 'properties') is None
+    features = [{'properties': {'hi': 'there'}},
+                {'properties': {'hi': None}}]
+    assert GeoJson._search_properties(features, 'properties') is None
+    features = [{'properties': {'hi': 'there'}},
+                {'properties': 42}]
+    assert GeoJson._search_properties(features, 'properties') is None
+    features = [{'properties': [42, 43]},
+                {'properties': [1, 2]}]
+    assert GeoJson._search_properties(features, 'properties') is None
+
+
+def test_geojson_find_identifier():
+
+    def _create(properties):
+        return {"type": "FeatureCollection", "features": [
+            {"type": "Feature",
+             "properties": properties}
+        ]}
+
+    data_bare = _create(None)
+    data_with_id = _create(None)
+    data_with_id['features'][0]['id'] = 'this-is-an-id'
+    data_with_unique_property = _create({
+        'property-key': 'some-value',
+    })
+    data_with_nested_properties = _create({
+        "summary": {"distance": 343.2},
+        "way_points": [3, 5],
+    })
+    data_with_incompatible_properties = _create({
+        "summary": {"distances": [0, 6], "durations": None},
+        "way_points": [3, 5],
+    })
+
+    geojson = GeoJson(data_with_id)
+    assert geojson.find_identifier() == 'feature.id'
+    geojson = GeoJson(data_bare)
+    assert geojson.find_identifier() == 'feature.id'
+    assert geojson.data['features'][0]['id'] == '0'  # the id got added
+    geojson = GeoJson(data_with_unique_property)
+    assert geojson.find_identifier() == 'feature.properties.property-key'
+    geojson = GeoJson(data_with_nested_properties)
+    assert geojson.find_identifier() == 'feature.properties.summary.distance'
+    geojson = GeoJson(data_with_incompatible_properties)
+    assert geojson.find_identifier() == 'feature.id'
+    assert geojson.data['features'][0]['id'] == '0'  # the id got added
+
+    data_loose_geometry = {"type": "LineString", "coordinates": [
+        [3.961389, 43.583333], [3.968056, 43.580833], [3.974722, 43.578333],
+        [3.986389, 43.575278], [3.998333, 43.5725], [4.163333, 43.530556],
+    ]}
+    geojson = GeoJson(data_loose_geometry)
+    geojson.convert_to_feature_collection()
+    assert geojson.find_identifier() == 'feature.id'  # id got added