Add support for xxhash

Author: varunkumar

pybloom_live/benchmarks.py

@@ -1,9 +1,13 @@
 #!/usr/bin/env python
 #
 """Test performance of BloomFilter at a set capacity and error rate."""
+import math
 import sys
+import time
+
+import bitarray
+
 from pybloom import BloomFilter
-import bitarray, math, time
 from utils import range_fn
 
 

pybloom_live/pybloom.py

@@ -5,11 +5,15 @@
 Requires the bitarray library: http://pypi.python.org/pypi/bitarray/
 """
 from __future__ import absolute_import
-import math
-import hashlib
+
 import copy
-from pybloom_live.utils import range_fn, is_string_io, running_python_3
-from struct import unpack, pack, calcsize
+import hashlib
+import math
+from struct import calcsize, pack, unpack
+
+import xxhash
+
+from pybloom_live.utils import is_string_io, range_fn, running_python_3
 
 try:
     import bitarray
@@ -34,7 +38,7 @@ def make_hashfuncs(num_slices, num_bits):
     elif total_hash_bits > 128:
         hashfn = hashlib.sha1
     else:
-        hashfn = hashlib.md5
+        hashfn = xxhash.xxh128
 
     fmt = fmt_code * (hashfn().digest_size // chunk_size)
     num_salts, extra = divmod(num_slices, len(fmt))

pybloom_live/test_pybloom.py

@@ -1,35 +1,35 @@
 from __future__ import absolute_import
+
 from pybloom_live.pybloom import (BloomFilter, ScalableBloomFilter,
                                   make_hashfuncs)
-from pybloom_live.utils import running_python_3, range_fn
+from pybloom_live.utils import range_fn, running_python_3
 
 try:
-    import StringIO
     import cStringIO
+    import StringIO
 except ImportError:
     pass
 
 import io
-
-import unittest
 import random
 import tempfile
+import unittest
 
 import pytest
 
 
 class TestMakeHashFuncs(unittest.TestCase):
     def test_make_hashfuncs_returns_hashfn(self):
         make_hashes, hashfn = make_hashfuncs(100, 20)
-        self.assertEquals('openssl_sha512', hashfn.__name__)
+        self.assertEqual('openssl_sha512', hashfn.__name__)
         make_hashes, hashfn = make_hashfuncs(20, 3)
-        self.assertEquals('openssl_sha384', hashfn.__name__)
+        self.assertEqual('openssl_sha384', hashfn.__name__)
         make_hashes, hashfn = make_hashfuncs(15, 2)
-        self.assertEquals('openssl_sha256', hashfn.__name__)
+        self.assertEqual('openssl_sha256', hashfn.__name__)
         make_hashes, hashfn = make_hashfuncs(10, 2)
-        self.assertEquals('openssl_sha1', hashfn.__name__)
+        self.assertEqual('openssl_sha1', hashfn.__name__)
         make_hashes, hashfn = make_hashfuncs(5, 1)
-        self.assertEquals('openssl_md5', hashfn.__name__)
+        self.assertEqual('xxh3_128', hashfn.__name__)
 
 
 class TestUnionIntersection(unittest.TestCase):

requirements.txt

@@ -1 +1,2 @@
-bitarray>=0.3.4
+bitarray>=0.3.4
+xxhash>=3.0.0