def test_spatial_join_to_spatial_rdd(self):
poi_point_rdd = WktReader.readToGeometryRDD(self.sc, bank_csv_path, 1, False, False)
areas_polygon_rdd = WktReader.readToGeometryRDD(self.sc, areas_csv_path, 1, False, False)
poi_point_rdd.analyze()
areas_polygon_rdd.analyze()
poi_point_rdd.spatialPartitioning(GridType.QUADTREE)
areas_polygon_rdd.spatialPartitioning(poi_point_rdd.getPartitioner())
jvm_sedona_rdd = JoinQueryRaw.spatialJoin(poi_point_rdd, areas_polygon_rdd, JoinParams(considerBoundaryIntersection=True))
sedona_rdd = jvm_sedona_rdd.to_rdd().collect()
assert sedona_rdd.__len__() == 5
def spatialJoin(cls, queryWindowRDD: SpatialRDD, objectRDD: SpatialRDD,
joinParams: JoinParams) -> RDD:
"""
:param queryWindowRDD: SpatialRDD
:param objectRDD: SpatialRDD
:param joinParams: JoinParams
:return:
"""
pair_rdd = JoinQueryRaw.spatialJoin(queryWindowRDD, objectRDD,
joinParams)
return pair_rdd.to_rdd()
def test_spatial_join_to_df(self):
poi_point_rdd = WktReader.readToGeometryRDD(self.sc, bank_csv_path, 1,
False, False)
areas_polygon_rdd = WktReader.readToGeometryRDD(
self.sc, areas_csv_path, 1, False, False)
poi_point_rdd.analyze()
areas_polygon_rdd.analyze()
poi_point_rdd.spatialPartitioning(GridType.QUADTREE)
areas_polygon_rdd.spatialPartitioning(poi_point_rdd.getPartitioner())
jvm_sedona_rdd = JoinQueryRaw.spatialJoin(poi_point_rdd,
areas_polygon_rdd,
JoinParams())
sedona_df = Adapter.toDf(jvm_sedona_rdd, ["area_id", "area_name"],
["poi_id", "poi_name"], self.spark)
assert sedona_df.count() == 5
assert sedona_df.columns == [
"leftgeometry", "area_id", "area_name", "rightgeometry", "poi_id",
"poi_name"
]