def test_empty_constructor_test(self):
object_rdd = PointRDD(sparkContext=self.sc,
InputLocation=point_rdd_input_location,
Offset=point_rdd_offset,
splitter=point_rdd_splitter,
carryInputData=False)
object_rdd_copy = PointRDD()
object_rdd_copy.rawJvmSpatialRDD = object_rdd.rawJvmSpatialRDD
object_rdd_copy.analyze()
def test_empty_constructor(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=input_location,
Offset=offset,
splitter=splitter,
carryInputData=True,
partitions=num_partitions,
newLevel=StorageLevel.MEMORY_ONLY)
spatial_rdd.buildIndex(IndexType.RTREE, False)
spatial_rdd_copy = PointRDD()
spatial_rdd_copy.rawJvmSpatialRDD = spatial_rdd.rawJvmSpatialRDD
spatial_rdd_copy.analyze()
def test_spatial_range_query(self):
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True, StorageLevel.MEMORY_ONLY
)
object_rdd.rawJvmSpatialRDD.persist(StorageLevel.MEMORY_ONLY)
for _ in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd, range_query_window, False, False).count()
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True, StorageLevel.MEMORY_ONLY
)
object_rdd.rawJvmSpatialRDD.persist(StorageLevel.MEMORY_ONLY)
for _ in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd, range_query_window, False, False).count()
def test_spatial_knn_query_using_index(self):
point_rdd = PointRDD(self.sc, input_location, offset, splitter, False)
point_rdd.buildIndex(IndexType.RTREE, False)
for i in range(self.loop_times):
result = KNNQuery.SpatialKnnQuery(point_rdd, self.query_point, self.top_k, False)
assert result.__len__() > -1
def load(cls, sc: SparkContext, path: str) -> SpatialRDD:
jvm = sc._jvm
point_rdd = PointRDD()
srdd = SpatialObjectLoaderAdapter(jvm).load_point_spatial_rdd(
sc._jsc, path)
point_rdd.set_srdd(srdd)
return point_rdd
def test_creating_point_rdd(self):
point_rdd = PointRDD(self.spark._sc, point_path, 4,
FileDataSplitter.WKT, True)
point_rdd.analyze()
cnt = point_rdd.countWithoutDuplicates()
assert cnt == 12872, f"Point RDD should have 12872 but found {cnt}"
def test_point_rdd(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=crs_test_point,
Offset=0,
splitter=splitter,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
raw_spatial_rdd = spatial_rdd.rawSpatialRDD.map(
lambda x: [x.geom, *x.getUserData().split("\t")])
self.spark.createDataFrame(raw_spatial_rdd).show()
schema = StructType([
StructField("geom", GeometryType()),
StructField("name", StringType())
])
spatial_rdd_with_schema = self.spark.createDataFrame(
raw_spatial_rdd, schema, verifySchema=False)
spatial_rdd_with_schema.show()
assert spatial_rdd_with_schema.take(
1)[0][0].wkt == "POINT (32.324142 -88.331492)"
def test_spatial_knn_correctness(self):
point_rdd = PointRDD(self.sc, input_location, offset, splitter, True,
StorageLevel.MEMORY_ONLY, "epsg:4326",
"epsg:3005")
result_no_index = KNNQuery.SpatialKnnQuery(point_rdd, query_point,
top_k, False)
point_rdd.buildIndex(IndexType.RTREE, False)
result_with_index = KNNQuery.SpatialKnnQuery(point_rdd, query_point,
top_k, True)
sorted_result_no_index = sorted(
result_no_index,
key=lambda geo_data: distance_sorting_functions(
geo_data, query_point))
sorted_result_with_index = sorted(
result_with_index,
key=lambda geo_data: distance_sorting_functions(
geo_data, query_point))
difference = 0
for x in range(top_k):
difference += sorted_result_no_index[x].geom.distance(
sorted_result_with_index[x].geom)
assert difference == 0
def test_spatial_join_query_with_polygon_rdd_using_index(self):
query_rdd = PolygonRDD(self.sc, input_location_query_polygon, splitter,
True, num_partitions, StorageLevel.MEMORY_ONLY,
"epsg:4326", "epsg:3005")
spatial_rdd = PointRDD(self.sc, input_location, offset, splitter, True,
num_partitions, StorageLevel.MEMORY_ONLY,
"epsg:4326", "epsg:3005")
query_rdd.analyze()
spatial_rdd.analyze()
spatial_rdd.spatialPartitioning(grid_type)
spatial_rdd.buildIndex(IndexType.RTREE, True)
query_rdd.spatialPartitioning(spatial_rdd.getPartitioner())
result = JoinQuery.SpatialJoinQuery(spatial_rdd, query_rdd, False,
True).collect()
assert result[1][0].getUserData() is not None
for data in result:
if data[1].__len__() != 0:
for right_data in data[1]:
assert right_data.getUserData() is not None
def test_saving_to_disc_index_point(self):
from tests.properties.point_properties import input_location, offset, splitter, num_partitions
point_rdd = PointRDD(self.sc, input_location, offset, splitter, True,
num_partitions, StorageLevel.MEMORY_ONLY)
point_rdd.buildIndex(IndexType.RTREE, False)
point_rdd.indexedRawRDD.saveAsObjectFile(
os.path.join(disc_location, "point_index"))
def test_raw_spatial_rdd_assignment(self):
spatial_rdd = PointRDD(self.sc, input_location, offset, splitter, True,
num_partitions, StorageLevel.MEMORY_ONLY)
spatial_rdd.analyze()
empty_point_rdd = PointRDD()
empty_point_rdd.rawSpatialRDD = spatial_rdd.rawSpatialRDD
empty_point_rdd.analyze()
assert empty_point_rdd.countWithoutDuplicates(
) == spatial_rdd.countWithoutDuplicates()
assert empty_point_rdd.boundaryEnvelope == spatial_rdd.boundaryEnvelope
assert empty_point_rdd.rawSpatialRDD.map(
lambda x: x.geom.area).collect()[0] == 0.0
assert empty_point_rdd.rawSpatialRDD.take(9)[4].getUserData(
) == "testattribute0\ttestattribute1\ttestattribute2"
def test_spatial_knn_query(self):
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True, StorageLevel.MEMORY_ONLY
)
object_rdd.rawJvmSpatialRDD.persist(StorageLevel.MEMORY_ONLY)
for _ in range(each_query_loop_times):
result = KNNQuery.SpatialKnnQuery(object_rdd, knn_query_point, 1000, False)
def test_spatial_range_query(self):
spatial_rdd = PointRDD(self.sc, input_location, offset, splitter,
False)
for i in range(self.loop_times):
result_size = RangeQuery.\
SpatialRangeQuery(spatial_rdd, self.query_envelope, False, False)\
.count()
assert result_size == 2830
def test_saving_to_disc_spatial_rdd_point(self):
from tests.properties.point_properties import input_location, offset, splitter, num_partitions
point_rdd = PointRDD(self.sc, input_location, offset, splitter, True,
num_partitions, StorageLevel.MEMORY_ONLY)
point_rdd.rawJvmSpatialRDD.saveAsObjectFile(
os.path.join(disc_location, "point"))
def test_spatial_knn_query(self):
point_rdd = PointRDD(self.sc, input_location, offset, splitter, False)
for i in range(self.loop_times):
result = KNNQuery.SpatialKnnQuery(point_rdd, self.query_point,
self.top_k, False)
assert result.__len__() > -1
assert result[0].getUserData() is not None
def test_crs_transformation_spatial_range_query_using_index(self):
object_rdd = PointRDD(self.sc, point_rdd_input_location, point_rdd_offset,
point_rdd_splitter, True, StorageLevel.MEMORY_ONLY, "epsg:4326", "epsg:3005")
object_rdd.buildIndex(point_rdd_index_type, False)
object_rdd.indexedRawRDD.persist(StorageLevel.MEMORY_ONLY)
for _ in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd, range_query_window, False, True).count()
assert result_size > -1
def test_knn_query(self):
object_rdd = PointRDD(sparkContext=self.sc,
InputLocation=point_rdd_input_location,
Offset=point_rdd_offset,
splitter=point_rdd_splitter,
carryInputData=False)
for i in range(each_query_loop_times):
result = KNNQuery.SpatialKnnQuery(object_rdd, knn_query_point,
1000, False)
def create_spatial_rdd(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=input_file_location,
Offset=offset,
splitter=splitter,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
return spatial_rdd
def test_spatial_knn_query_using_index(self):
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True, StorageLevel.MEMORY_ONLY
)
object_rdd.buildIndex(point_rdd_index_type, False)
object_rdd.indexedRawRDD.persist(StorageLevel.MEMORY_ONLY)
for _ in range(each_query_loop_times):
result = KNNQuery.SpatialKnnQuery(object_rdd, knn_query_point, 1000, True)
def test_build_index_without_set_grid(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=input_location,
Offset=offset,
splitter=splitter,
carryInputData=True,
partitions=num_partitions,
newLevel=StorageLevel.MEMORY_ONLY)
spatial_rdd.buildIndex(IndexType.RTREE, False)
def test_spatial_knn_query(self):
point_rdd = PointRDD(
self.sc, input_location, offset, splitter, True, StorageLevel.MEMORY_ONLY, "epsg:4326", "epsg:3005"
)
for i in range(loop_times):
result = KNNQuery.SpatialKnnQuery(point_rdd, query_point, top_k, False)
assert result.__len__() > 0
assert result[0].getUserData() is not None
def test_save_as_geo_json_with_data(self, remove_wkb_directory):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=inputLocation,
Offset=offset,
splitter=splitter,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
spatial_rdd.saveAsGeoJSON(test_save_as_wkb_with_data)
result_wkb = PointRDD(sparkContext=self.sc,
InputLocation=test_save_as_wkb_with_data,
splitter=FileDataSplitter.GEOJSON,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
assert result_wkb.rawSpatialRDD.count(
) == spatial_rdd.rawSpatialRDD.count()
def test_range_query_using_index(self):
object_rdd = PointRDD(sparkContext=self.sc,
InputLocation=point_rdd_input_location,
Offset=point_rdd_offset,
splitter=point_rdd_splitter,
carryInputData=False)
object_rdd.buildIndex(point_rdd_index_type, False)
for i in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd,
range_query_window,
False, True).count
def test_spatial_range_query(self):
object_rdd = PointRDD(sparkContext=self.sc,
InputLocation=point_rdd_input_location,
Offset=point_rdd_offset,
splitter=point_rdd_splitter,
carryInputData=False)
for i in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd,
range_query_window,
False, False).count()
logging.info(result_size)
def test_equal_partitioning(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=input_location,
Offset=offset,
splitter=splitter,
carryInputData=False,
partitions=10,
newLevel=StorageLevel.MEMORY_ONLY)
spatial_rdd.analyze()
spatial_rdd.spatialPartitioning(GridType.QUADTREE)
assert spatial_rdd.countWithoutDuplicates(
) == spatial_rdd.countWithoutDuplicatesSPRDD()
def test_crs_transformed_spatial_range_query(self):
object_rdd = PointRDD(sparkContext=self.sc,
InputLocation=point_rdd_input_location,
Offset=point_rdd_offset,
splitter=point_rdd_splitter,
carryInputData=False,
newLevel=StorageLevel.DISK_ONLY,
sourceEpsgCRSCode="epsg:4326",
targetEpsgCode="epsg:3005")
for i in range(each_query_loop_times):
result_size = RangeQuery.SpatialRangeQuery(object_rdd,
range_query_window,
False, False)
def test_get_crs_transformation(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=crs_test_point,
Offset=0,
splitter=splitter,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
assert not spatial_rdd.getCRStransformation()
spatial_rdd.CRSTransform("epsg:4326", "epsg:3857")
assert spatial_rdd.getCRStransformation()
def test_spatial_range_query_using_index(self):
spatial_rdd = PointRDD(self.sc, input_location, offset, splitter, False)
spatial_rdd.buildIndex(IndexType.RTREE, False)
for i in range(self.loop_times):
result_size = RangeQuery.\
SpatialRangeQuery(spatial_rdd, self.query_envelope, False, False)\
.count()
assert result_size == 2830
assert RangeQuery.SpatialRangeQuery(
spatial_rdd, self.query_envelope, False, False).take(10)[1].\
getUserData() is not None
def test_indexed_rdd_assignment(self):
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True)
query_window_rdd = CircleRDD(object_rdd, 0.1)
object_rdd.analyze()
object_rdd.spatialPartitioning(GridType.QUADTREE)
object_rdd.buildIndex(IndexType.QUADTREE, True)
query_window_rdd.spatialPartitioning(object_rdd.getPartitioner())
object_rdd.buildIndex(IndexType.RTREE, False)
object_rdd.indexedRDD.persist(StorageLevel.MEMORY_ONLY)
query_window_rdd.jvmSpatialPartitionedRDD.persist(StorageLevel.MEMORY_ONLY)
query_window_rdd.jvmSpatialPartitionedRDD.count()
object_rdd.indexedRDD.count()
import time
start = time.time()
for _ in range(each_query_loop_times):
result_size = JoinQuery.DistanceJoinQuery(object_rdd, query_window_rdd, True, True).count()
diff = time.time() - start
object_rdd = PointRDD(
self.sc, point_rdd_input_location, point_rdd_offset, point_rdd_splitter, True)
query_window_rdd = CircleRDD(object_rdd, 0.1)
object_rdd.analyze()
object_rdd.spatialPartitioning(GridType.QUADTREE)
object_rdd.buildIndex(IndexType.QUADTREE, True)
query_window_rdd.spatialPartitioning(object_rdd.getPartitioner())
object_rdd.buildIndex(IndexType.RTREE, False)
start1 = time.time()
for _ in range(each_query_loop_times):
result_size = JoinQuery.DistanceJoinQuery(object_rdd, query_window_rdd, True, True).count()
def test_crs_transform(self):
spatial_rdd = PointRDD(sparkContext=self.sc,
InputLocation=crs_test_point,
Offset=0,
splitter=splitter,
carryInputData=True,
partitions=numPartitions,
newLevel=StorageLevel.MEMORY_ONLY)
spatial_rdd.CRSTransform("epsg:4326", "epsg:3857")
assert spatial_rdd.rawSpatialRDD.collect(
)[0].geom.wkt == "POINT (-9833016.710450118 3805934.914254189)"