def serialize(cls, obj: Point, binary_buffer: BinaryBuffer):
if isinstance(obj, Point):
add_shape_geometry_metadata(GeomEnum.point.value, binary_buffer)
binary_buffer.put_double(obj.x)
binary_buffer.put_double(obj.y)
binary_buffer.put_int(-2130640127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def add_offsets_to_polygon(geom: Polygon, binary_buffer: BinaryBuffer, initial_offset: int) -> int:
offset = initial_offset
num_rings = get_number_of_rings(geom)
binary_buffer.put_int(offset)
offset += geom.exterior.coords.__len__()
for _ in range(num_rings - 1):
binary_buffer.put_int(offset)
offset = offset + geom.interiors[_].coords.__len__()
return offset
def serialize(cls, obj: MultiPolygon, binary_buffer: BinaryBuffer):
if isinstance(obj, MultiPolygon):
num_polygons = len(obj.geoms)
num_points = sum([
get_number_of_polygon_points(polygon) for polygon in obj.geoms
])
num_rings = sum(
[get_number_of_rings(polygon) for polygon in obj.geoms])
add_shape_geometry_metadata(GeomEnum.polygon.value, binary_buffer)
binary_buffer.add_empty_bytes("double", 4)
binary_buffer.put_int(num_rings)
binary_buffer.put_int(num_points)
offset = 0
for geom in obj.geoms:
offset = add_offsets_to_polygon(geom, binary_buffer, offset)
for geom in obj.geoms:
coordinates_exterior = reverse_linear_ring(
geom.exterior, False)
put_coordinates(coordinates_exterior, binary_buffer)
for ring in geom.interiors:
coordinates = reverse_linear_ring(ring)
put_coordinates(coordinates, binary_buffer)
binary_buffer.put_int(-2130640127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def serialize(cls, obj: Polygon, binary_buffer: BinaryBuffer):
if isinstance(obj, Polygon):
add_shape_geometry_metadata(GeomEnum.polygon.value, binary_buffer)
num_rings = get_number_of_rings(obj)
num_points = get_number_of_polygon_points(obj)
binary_buffer.add_empty_bytes("double", 4)
binary_buffer.put_int(num_rings)
binary_buffer.put_int(num_points)
add_offsets_to_polygon(obj, binary_buffer, 0)
coordinates_exterior = reverse_linear_ring(obj.exterior, False)
put_coordinates(coordinates_exterior, binary_buffer)
for ring in obj.interiors:
coordinates = reverse_linear_ring(ring)
put_coordinates(coordinates, binary_buffer)
binary_buffer.put_int(-2130640127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def serialize(cls, obj: LineString, binary_buffer: BinaryBuffer):
if isinstance(obj, LineString):
add_shape_geometry_metadata(GeomEnum.polyline.value, binary_buffer)
binary_buffer.add_empty_bytes("double", 4)
num_points = obj.coords.__len__()
binary_buffer.put_int(1)
binary_buffer.put_int(num_points)
binary_buffer.put_int(0)
put_coordinates(obj.coords, binary_buffer)
binary_buffer.put_int(-2130640127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def to_bytes(cls, geom: BaseGeometry) -> List[int]:
geom_name = str(geom.__class__.__name__).lower()
try:
appr_parser = cls.parsers[geom_name]
except KeyError:
raise KeyError(f"Parser for geometry {geom_name}")
return appr_parser.serialize(geom, BinaryBuffer())
def to_bytes(cls, geom: BaseGeometry) -> List[int]:
from geo_pyspark.sql.types import GeometryType
geom_name = str(geom.__class__.__name__).lower()
try:
appr_parser = PARSERS[geom_name]
geom.__UDT__ = GeometryType()
except KeyError:
raise KeyError(f"Parser for geometry {geom_name} is not available")
return appr_parser.serialize(geom, BinaryBuffer())
def serialize(cls, obj: MultiPoint, binary_buffer: BinaryBuffer):
if isinstance(obj, MultiPoint):
add_shape_geometry_metadata(GeomEnum.multipoint.value, binary_buffer)
binary_buffer.add_empty_bytes("double", 4)
binary_buffer.put_int(len(obj.geoms))
for point in obj.geoms:
binary_buffer.put_double(point.x)
binary_buffer.put_double(point.y)
binary_buffer.put_byte(-127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def to_bytes(self):
from geo_pyspark.utils.binary_parser import BinaryBuffer
bin_buffer = BinaryBuffer()
bin_buffer.put_double(self.minx)
bin_buffer.put_double(self.maxx)
bin_buffer.put_double(self.miny)
bin_buffer.put_double(self.maxy)
return bin_buffer.byte_array
def serialize(cls, obj: MultiLineString, binary_buffer: BinaryBuffer):
if isinstance(obj, MultiLineString):
add_shape_geometry_metadata(GeomEnum.polyline.value, binary_buffer)
binary_buffer.add_empty_bytes("double", 4)
num_parts = len(obj.geoms)
num_points = sum([len(el.coords) for el in obj.geoms])
binary_buffer.put_int(num_parts)
binary_buffer.put_int(num_points)
offset = 0
for _ in range(num_parts):
binary_buffer.put_int(offset)
offset = offset + obj.geoms[_].coords.__len__()
for geom in obj.geoms:
put_coordinates(geom.coords, binary_buffer)
binary_buffer.put_int(-2130640127)
else:
raise TypeError(f"Need a {cls.name} instance")
return binary_buffer.byte_array
def add_shape_geometry_metadata(geom_type: int, binary_buffer: BinaryBuffer):
binary_buffer.put_byte(ShapeEnum.shape.value)
binary_buffer.put_byte(geom_type)
def put_coordinates(coordinates: Iterable[Iterable[numeric]],
binary_buffer: BinaryBuffer):
for coordinate in coordinates:
binary_buffer.put_double(Point(coordinate).x)
binary_buffer.put_double(Point(coordinate).y)