feat: reduce geojson maps precision and remove ids (#1335)

Author: davidgamez

functions-python/helpers/locations.py

@@ -46,7 +46,7 @@ def get_country_code(country_name: str) -> Optional[str]:
         if country:
             return country.alpha_2
 
-        # Try searching by name
+        # Try searching with fuzzy matching
         countries = pycountry.countries.search_fuzzy(country_name)
         if countries:
             return countries[0].alpha_2
@@ -231,3 +231,50 @@ def get_geopolygons_covers(stop_point: WKTElement, db_session: Session):
         ).all()
     )
     return geopolygons
+
+
+def round_geojson_coords(geometry, precision=5):
+    """
+    Recursively round all coordinates in a GeoJSON geometry to the given precision.
+    Handles Point, LineString, Polygon, MultiPoint, MultiLineString, MultiPolygon, GeometryCollection.
+    """
+    geom_type = geometry.get("type")
+    if geom_type == "GeometryCollection":
+        return {
+            "type": "GeometryCollection",
+            "geometries": [
+                round_geojson_coords(g, precision)
+                for g in geometry.get("geometries", [])
+            ],
+        }
+    elif "coordinates" in geometry:
+        return {
+            **geometry,
+            "coordinates": round_coords(geometry["coordinates"], precision),
+        }
+    else:
+        return geometry
+
+
+def round_coords(coords, precision):
+    """
+    Recursively round coordinates to the given precision.
+    Handles nested lists of coordinates.
+    Args:
+        coords: A coordinate or list of coordinates (can be nested)
+        precision: Number of decimal places to round to
+    Returns:
+        Rounded coordinates with the same structure as input
+    """
+    if isinstance(coords, (list, tuple)):
+        if coords and isinstance(coords[0], (list, tuple)):
+            return [round_coords(c, precision) for c in coords]
+        else:
+            result = []
+            for c in coords:
+                if isinstance(c, (int, float)):
+                    result.append(round(float(c), precision))
+                else:
+                    result.append(c)
+            return result
+    return coords

functions-python/helpers/tests/test_locations.py

@@ -1,5 +1,5 @@
 """Unit tests for locations helper module."""
-
+import math
 import unittest
 from unittest.mock import MagicMock
 
@@ -16,6 +16,8 @@
     select_highest_level_polygon,
     select_lowest_level_polygon,
     get_country_code_from_polygons,
+    round_geojson_coords,
+    round_coords,
 )
 from unittest.mock import patch
 
@@ -455,3 +457,154 @@ def test_get_country_code_from_polygons_all_none_admin_levels_returns_one_with_c
         ]
         result = get_country_code_from_polygons(polys)
         self.assertIn(result, {"US", "CA"})
+
+    def _coords_equal(self, a, b, abs_tol=1e-5):
+        if isinstance(a, (list, tuple)) and isinstance(b, (list, tuple)):
+            if len(a) != len(b):
+                return False
+            return all(self._coords_equal(x, y, abs_tol=abs_tol) for x, y in zip(a, b))
+        elif isinstance(a, (list, tuple)) or isinstance(b, (list, tuple)):
+            return False
+        else:
+            return math.isclose(a, b, abs_tol=abs_tol)
+
+    def test_round_point(self):
+        geom = {"type": "Point", "coordinates": [1.1234567, 2.9876543]}
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(rounded["coordinates"], [1.12346, 2.98765])
+
+    def test_round_linestring(self):
+        geom = {
+            "type": "LineString",
+            "coordinates": [[1.1234567, 2.9876543], [3.1111111, 4.9999999]],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["coordinates"], [[1.12346, 2.98765], [3.11111, 5.0]]
+        )
+
+    def test_round_polygon(self):
+        geom = {
+            "type": "Polygon",
+            "coordinates": [
+                [[1.1234567, 2.9876543], [3.1111111, 4.9999999], [1.1234567, 2.9876543]]
+            ],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["coordinates"],
+            [[[1.12346, 2.98765], [3.11111, 5.0], [1.12346, 2.98765]]],
+        )
+
+    def test_round_multipoint(self):
+        geom = {
+            "type": "MultiPoint",
+            "coordinates": [[1.1234567, 2.9876543], [3.1111111, 4.9999999]],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["coordinates"], [[1.12346, 2.98765], [3.11111, 5.0]]
+        )
+
+    def test_round_multilinestring(self):
+        geom = {
+            "type": "MultiLineString",
+            "coordinates": [
+                [[1.1234567, 2.9876543], [3.1111111, 4.9999999]],
+                [[5.5555555, 6.6666666], [7.7777777, 8.8888888]],
+            ],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["coordinates"],
+            [
+                [[1.12346, 2.98765], [3.11111, 5.0]],
+                [[5.55556, 6.66667], [7.77778, 8.88889]],
+            ],
+        )
+
+    def test_round_multipolygon(self):
+        geom = {
+            "type": "MultiPolygon",
+            "coordinates": [
+                [
+                    [
+                        [1.1234567, 2.9876543],
+                        [3.1111111, 4.9999999],
+                        [1.1234567, 2.9876543],
+                    ]
+                ],
+                [
+                    [
+                        [5.5555555, 6.6666666],
+                        [7.7777777, 8.8888888],
+                        [5.5555555, 6.6666666],
+                    ]
+                ],
+            ],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["coordinates"],
+            [
+                [[[1.12346, 2.98765], [3.11111, 5.0], [1.12346, 2.98765]]],
+                [[[5.55556, 6.66667], [7.77778, 8.88889], [5.55556, 6.66667]]],
+            ],
+        )
+
+    def test_round_geometrycollection(self):
+        geom = {
+            "type": "GeometryCollection",
+            "geometries": [
+                {"type": "Point", "coordinates": [1.1234567, 2.9876543]},
+                {
+                    "type": "LineString",
+                    "coordinates": [[3.1111111, 4.9999999], [5.5555555, 6.6666666]],
+                },
+            ],
+        }
+        rounded = round_geojson_coords(geom, precision=5)
+        assert self._coords_equal(
+            rounded["geometries"][0]["coordinates"], [1.12346, 2.98765]
+        )
+        assert self._coords_equal(
+            rounded["geometries"][1]["coordinates"],
+            [[3.11111, 5.0], [5.55556, 6.66667]],
+        )
+
+    def test_empty_coords(self):
+        geom = {"type": "Point", "coordinates": []}
+        rounded = round_geojson_coords(geom, precision=5)
+        assert rounded["coordinates"] == []
+
+    def test_non_list_coords(self):
+        geom = {"type": "Point", "coordinates": 1.1234567}
+        rounded = round_geojson_coords(geom, precision=5)
+        assert rounded["coordinates"] == 1.1234567
+
+    def test_round_coords_single_float(self):
+        assert (
+            round_coords(1.1234567, 5) == 1.1234567
+        )  # Non-list input returns unchanged
+
+    def test_round_coords_list_of_floats(self):
+        assert round_coords([1.1234567, 2.9876543], 5) == [1.12346, 2.98765]
+
+    def test_round_coords_tuple_of_floats(self):
+        assert round_coords((1.1234567, 2.9876543), 5) == [1.12346, 2.98765]
+
+    def test_round_coords_nested_lists(self):
+        coords = [[[1.1234567, 2.9876543], [3.1111111, 4.9999999]]]
+        expected = [[[1.12346, 2.98765], [3.11111, 5.0]]]
+        assert round_coords(coords, 5) == expected
+
+    def test_round_coords_empty_list(self):
+        assert round_coords([], 5) == []
+
+    def test_round_coords_non_numeric(self):
+        assert round_coords("not_a_number", 5) == "not_a_number"
+
+    def test_round_coords_mixed_types(self):
+        coords = [1.1234567, "foo", 2.9876543]
+        expected = [1.12346, "foo", 2.98765]
+        assert round_coords(coords, 5) == expected

functions-python/reverse_geolocation/src/reverse_geolocation_processor.py

@@ -36,7 +36,7 @@
 )
 from shared.dataset_service.dataset_service_commons import Status
 
-from shared.helpers.locations import ReverseGeocodingStrategy
+from shared.helpers.locations import ReverseGeocodingStrategy, round_geojson_coords
 from shared.helpers.logger import get_logger
 from shared.helpers.runtime_metrics import track_metrics
 from shared.helpers.utils import (
@@ -169,9 +169,10 @@ def create_geojson_aggregate(
         "features": [
             {
                 "type": "Feature",
-                "geometry": mapping(geo_polygon_count[osm_id]["clipped_geometry"]),
+                "geometry": round_geojson_coords(
+                    mapping(geo_polygon_count[osm_id]["clipped_geometry"])
+                ),
                 "properties": {
-                    "osm_id": osm_id,
                     "country_code": geo_polygon_count[osm_id]["group"].iso_3166_1_code,
                     "subdivision_code": geo_polygon_count[osm_id][
                         "group"

functions-python/reverse_geolocation/tests/test_reverse_geolocation_processor.py

@@ -329,7 +329,6 @@ def test_create_geojson_aggregate(self, mock_getenv, mock_storage_client):
         self.assertEqual(geojson_data["type"], "FeatureCollection")
         self.assertEqual(len(geojson_data["features"]), 1)
         feature = geojson_data["features"][0]
-        self.assertEqual(feature["properties"]["osm_id"], str(geopolygon_1.osm_id))
         self.assertEqual(
             feature["properties"]["country_code"], geopolygon_1.iso_3166_1_code
         )

web-app/src/app/components/PopupTable.tsx

@@ -18,9 +18,6 @@ interface PopupTableProps {
 }
 
 const fieldDescriptions: Record<string, { description?: string }> = {
-  osm_id: {
-    description: 'OpenStreetMap ID of the geographic area.',
-  },
   stops_in_area: {
     description:
       'This is the number of stops in stops.txt that are located within this geographic area.',