Added change to add geopandas data frame to Datahub Space

Signed-off-by: Omkar Mestry <omkar.mestry@here.com>

changes/features/pr.71.md

@@ -0,0 +1 @@
+Added feature to upload data from `Geopandas Dataframe` file to the space.

tests/space/test_space_objects.py

@@ -21,6 +21,7 @@
 from pathlib import Path
 from time import sleep
 
+import geopandas as gpd
 import pytest
 from geojson import GeoJSON
 
@@ -690,3 +691,12 @@ def test_add_features_duplicate(empty_space):
     empty_space.add_features(geojson, features_size=100)
     stats = empty_space.get_statistics()
     assert stats["count"]["value"] == 180
+
+
+@pytest.mark.skipif(not XYZ_TOKEN, reason="No token found.")
+def test_add_features_geopandas(empty_space):
+    geojson_file = Path(__file__).parents[1] / "data" / "countries.geo.json"
+    df = gpd.read_file(geojson_file)
+    empty_space.add_features_geopandas(data=df)
+    stats = empty_space.get_statistics()
+    assert stats["count"]["value"] == 292

xyzspaces/spaces.py

@@ -42,7 +42,7 @@
 from geojson import Feature, GeoJSON
 
 from .apis import HubApi
-from .utils import divide_bbox, grouper, wkt_to_geojson
+from .utils import divide_bbox, flatten_geometry, grouper, wkt_to_geojson
 
 logger = logging.getLogger(__name__)
 
@@ -1179,24 +1179,7 @@ def add_features_kml(
 
         gdf = gpd.read_file(path, driver="KML")
 
-        geometry = gdf.geometry
-        flattened_geometry = []
-
-        flattened_gdf = gpd.GeoDataFrame()
-
-        for geom in geometry:
-            if geom.type in [
-                "GeometryCollection",
-                "MultiPoint",
-                "MultiLineString",
-                "MultiPolygon",
-            ]:
-                for subgeom in geom:
-                    flattened_geometry.append(subgeom)
-            else:
-                flattened_geometry.append(geom)
-
-        flattened_gdf.geometry = flattened_geometry
+        flattened_gdf = flatten_geometry(gdf)
 
         with tempfile.NamedTemporaryFile() as temp:
             flattened_gdf.to_file(temp.name, driver="GeoJSON")
@@ -1228,3 +1211,28 @@ def add_features_geobuf(
             features_size=features_size,
             chunk_size=chunk_size,
         )
+
+    def add_features_geopandas(
+        self,
+        data: gpd.GeoDataFrame,
+        features_size: int = 2000,
+        chunk_size: int = 1,
+    ):
+        """
+        Add features from GeoPandas dataframe to a space.
+
+        :param data: GeoPandas dataframe to be uploaded
+        :param features_size: The number of features to upload at
+            a time.
+        :param chunk_size: Number of chunks for each process to handle. The default value
+            is 1, for a large number of features please use `chunk_size` greater than 1.
+        """
+        flattened_gdf = flatten_geometry(data)
+
+        with tempfile.NamedTemporaryFile() as temp:
+            flattened_gdf.to_file(temp.name, driver="GeoJSON")
+            self.add_features_geojson(
+                path=temp.name,
+                features_size=features_size,
+                chunk_size=chunk_size,
+            )

xyzspaces/utils.py

@@ -28,6 +28,7 @@
 from itertools import zip_longest
 from typing import List, Optional
 
+import geopandas as gpd
 from geojson import Feature, FeatureCollection, Point, Polygon
 from shapely import geometry, wkt
 from turfpy.measurement import bbox, bbox_polygon, distance, length
@@ -234,3 +235,35 @@ def divide_bbox(
         except Exception:
             logger.debug("The intersection geometry is incorrect")
     return final
+
+
+def flatten_geometry(data: gpd.GeoDataFrame) -> gpd.GeoDataFrame:
+    """
+    Flatten the geometries in the given GeoPandas dataframe.
+    Flatten geometry is formed by extracting individual geometries from
+    GeometryCollection, MultiPoint, MultiLineString, MultiPolygon.
+
+    :param data: GeoPandas dataframe to be flatten
+    :return: Flat GeoPandas dataframe
+    """
+    geometry = data.geometry
+
+    flattened_geometry = []
+
+    flattened_gdf = gpd.GeoDataFrame()
+
+    for geom in geometry:
+        if geom.type in [
+            "GeometryCollection",
+            "MultiPoint",
+            "MultiLineString",
+            "MultiPolygon",
+        ]:
+            for subgeom in geom:
+                flattened_geometry.append(subgeom)
+        else:
+            flattened_geometry.append(geom)
+
+    flattened_gdf.geometry = flattened_geometry
+
+    return flattened_gdf