moolb

commit ae667ff39f95808fd62c99ebde9d36a5bfb3c797
parent b3a16cb3829e0622de4be86b2addf0295d942ef1
Author: nolash <dev@holbrook.no>
Date:   Sun, 14 Feb 2021 21:28:24 +0100

Add default data, merge filter

Diffstat:
M
CHANGELOG
 | 
3
+++
M
moolb/moolb.py
 | 
20
++++++++++++++++++--
M
setup.py
 | 
2
+-
M
tests/test_basic.py
 | 
37
+++++++++++++++++++++++++++++++++++--

4 files changed, 57 insertions(+), 5 deletions(-)
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,3 +1,6 @@
+- 0.1.1-pending
+	* Add default data import
+	* Add filter merge
 - 0.1.0
 	* Use bytes for serial instead of str
 - 0.0.4
diff --git a/moolb/moolb.py b/moolb/moolb.py
@@ -10,18 +10,34 @@ logg = logging.getLogger()
 
 class Bloom:
 
-    def __init__(self, bits, rounds, hasher=None):
+    def __init__(self, bits, rounds, hasher=None, default_data=None):
         self.bits = bits
         self.bytes = int(bits / 8)
         if self.bytes * 8 != self.bits:
             raise ValueError('Need byte boundary bit value')
         self.rounds = rounds
-        self.filter = [0] * self.bytes
         if hasher == None:
             logg.info('using default hasher (SHA256)')
             hasher = self.__hash
         self.hasher = self.set_hasher(hasher)
 
+        self.filter = None
+        if default_data != None:
+            self.merge(default_data)
+        else:
+            self.filter = [0] * self.bytes
+
+
+    def merge(self, filter_data):
+        datalen = len(filter_data)
+        if datalen != self.bytes:
+            raise ValueError('expected byte array bit size {}, got {}'.format(self.bits, datalen * 8))
+
+        if self.filter == None:
+            self.filter = filter_data
+        else:
+            self.filter = list(map(lambda x, y: x | y, filter_data, self.filter))
+
 
     def set_hasher(self, hasher):
         self.hasher = hasher
diff --git a/setup.py b/setup.py
@@ -6,7 +6,7 @@ f.close()
 
 setup(
         name='moolb',
-        version='0.1.0',
+        version='0.1.1b1',
         description='Simple bloom filter with pluggable hash backend',
         author='Louis Holbrook',
         author_email='dev@holbrook.no',
diff --git a/tests/test_basic.py b/tests/test_basic.py
@@ -18,19 +18,52 @@ def hashround(self, b, s):
 
 class Test(unittest.TestCase):
 
-    def test_default(self):
+    def test_basic(self):
         f = Bloom(8192 * 8, 3)
         f.add(b'1024')
         self.assertTrue(f.check(b'1024'))
         self.assertFalse(f.check(b'1023'))
 
-    
+
+    def test_defaul(self):
+        b = bytearray(8192)
+        b[42] = 13
+        f = Bloom(8192 * 8, 3, default_data=b)
+        self.assertEqual(f.filter[42], 13)
+        
+        b = bytearray(8193)
+        with self.assertRaises(ValueError):
+            f = Bloom(8192 * 8, 3, default_data=b)
+   
+
     def test_plug(self):
         f = Bloom(8192 * 8, 3, hashround)
         f.add(b'1024')
         self.assertTrue(f.check(b'1024'))
         self.assertFalse(f.check(b'1023'))
 
+
+    def test_merge(self):
+        f = Bloom(8 * 8, 3, hashround)
+        b = bytearray(8)
+        b[2] = 2
+        b[6] = 4
+        f.merge(b)
+        self.assertEqual(f.filter[2], 2)
+        self.assertEqual(f.filter[6], 4)
+
+        b = bytearray(8)
+        b[2] = 1
+        b[6] = 8
+        f.merge(b)
+        self.assertEqual(f.filter[2], 3)
+        self.assertEqual(f.filter[6], 12)
+
+        b = bytearray(9)
+        with self.assertRaises(ValueError):
+            f.merge(b)
+
+
 #    def test_dump(self):
 #        f = Bloom(8192 * 8, 3)
 #        f.add(b'1024')