Cythonize ewkb reader and writer

Author: yohanboniface

benchmark/__init__.py

@@ -0,0 +1,19 @@
+from timeit import timeit
+
+from postgis import Point
+
+
+def main():
+    point = Point([1.23456789, 48.765432])
+    ewkb = point.to_ewkb()
+    time = timeit('read(ewkb)',
+                  setup='from postgis.ewkb import read',
+                  number=1000000, globals=locals())
+    print("Read Point", time)
+    time = timeit('point.to_ewkb()',
+                  number=1000000, globals=locals())
+    print("Write Point", time)
+
+
+if __name__ == '__main__':
+    main()

postgis/ewkb.py renamed to postgis/ewkb.pyx

@@ -1,6 +1,7 @@
 import binascii
 from io import BytesIO
 import struct
+from cpython cimport bool
 
 
 class Typed(type):
@@ -20,17 +21,23 @@ def __call__(cls, *args, **kwargs):
         return super().__call__(*args, **kwargs)
 
 
-class Reader:
+cdef class Reader:
 
-    __slots__ = ['stream', 'endianness', 'has_z', 'has_m']
+    cdef object stream
+    cdef bytes endianness
+    cdef public bool has_z
+    cdef public bool has_m
 
-    def __init__(self, stream):
+    def __cinit__(self, stream: object):
         self.stream = stream
 
-    def clone(self):
+    cpdef clone(self):
         return type(self)(self.stream)
 
-    def read(self):
+    cpdef read(self):
+        return self._read()
+
+    cdef _read(self):
         # https://en.wikipedia.org/wiki/Well-known_text#Well-known_binary
         byte_order = self.stream.read(1)
         if byte_order == b'\x00':
@@ -53,22 +60,22 @@ def read(self):
         else:
             return class_.from_ewkb_body(self, srid)
 
-    def read_int(self):
+    cpdef read_int(self):
         return struct.unpack(self.endianness + b'I', self.stream.read(4))[0]
 
-    def read_double(self):
+    cpdef read_double(self):
         return struct.unpack(self.endianness + b'd', self.stream.read(8))[0]
 
-    @classmethod
-    def from_hex(cls, value):
-        return cls(BytesIO(binascii.a2b_hex(value))).read()
+
+def read(value: bytes):
+    return Reader(BytesIO(binascii.a2b_hex(value))).read()
 
 
-class Writer:
+cdef class Writer:
 
-    __slots__ = ['stream']
+    cdef object stream
 
-    def __init__(self, geometry, stream=None):
+    def __cinit__(self, geometry: object, stream: object=None):
         self.stream = stream or BytesIO()
         try:
             type_ = geometry.TYPE
@@ -85,17 +92,17 @@ def __init__(self, geometry, stream=None):
         if geometry.has_srid:
             self.write_int(geometry.srid)
 
-    def write_int(self, value):
+    cpdef write_int(self, value):
         self.stream.write(struct.pack(b'<I', value))
 
-    def write_double(self, value):
+    cpdef write_double(self, value):
         self.stream.write(struct.pack(b'<d', value))
 
-    def clone(self, geometry):
+    cpdef clone(self, geometry):
         return type(self)(geometry, self.stream)
 
-    @classmethod
-    def to_hex(cls, value):
-        writer = cls(value)
-        value.write_ewkb_body(writer)
-        return binascii.b2a_hex(writer.stream.getvalue()).upper()
+
+def write(value):
+    writer = Writer(value)
+    value.write_ewkb_body(writer)
+    return binascii.b2a_hex(writer.stream.getvalue()).upper()

postgis/geometry.py

@@ -5,7 +5,7 @@
     print('psycopg2 not installed')
 
 
-from .ewkb import Reader, Typed, Writer
+from .ewkb import read, Typed, write
 from .geojson import GeoJSON
 
 
@@ -27,10 +27,10 @@ def has_srid(self):
     def from_ewkb(value, cursor=None):
         if not value:
             return None
-        return Reader.from_hex(value)
+        return read(value)
 
     def to_ewkb(self):
-        return Writer.to_hex(self).decode()
+        return write(self).decode()
 
     def write_ewkb(self, writer):
         self.write_ewkb_body(writer.clone(self))

setup.py

@@ -5,6 +5,7 @@
 from codecs import open  # To use a consistent encoding
 from os import path
 import sys
+# from Cython.Build import cythonize
 
 HERE = path.abspath(path.dirname(__file__))
 
@@ -42,6 +43,7 @@ def list_modules(dirname):
     ext_modules = [
         Extension('postgis.' + ext, [path.join('postgis', ext + '.py')])
         for ext in list_modules(path.join(HERE, 'postgis'))]
+    ext_modules.append(Extension('postgis.ewkb', ['postgis/ewkb.pyx']))
 
     cmdclass = {'build_ext': build_ext}