Add in Bloomfilter.popcount() to get the number of bits set

Makefile

@@ -5,6 +5,9 @@ install:
 	@# Support Debian package building with fall-back default
 	python setup.py install --root $${DESTDIR:-/}
 
+test:
+	python setup.py test
+
 
 clean:
 	rm -rf build/

src/cbloomfilter.pxd

@@ -17,6 +17,7 @@ cdef extern from "mmapbitarray.h":
      MBArray * mbarray_And_Ternary(MBArray * dest, MBArray * a, MBArray * b)
      MBArray * mbarray_Or_Ternary(MBArray * dest, MBArray * a, MBArray * b)
      MBArray * mbarray_Xor_Ternary(MBArray * dest, MBArray * a, MBArray * b)
+     int mbarray_Popcount(MBArray *array)
      int mbarray_Update(MBArray * array, char * data, int size)
      int mbarray_FileSize(MBArray * array)
      char * mbarray_CharData(MBArray * array)

src/mmapbitarray.c

@@ -236,6 +236,22 @@ int mbarray_Sync(MBArray * array)
     return 0;
 }
 
+int popcount(DTYPE x)
+{
+    int c = 0;
+    for (; x > 0; x &= x -1) c++;
+    return c;
+}
+
+int mbarray_Popcount(MBArray * array)
+{
+    int pc = 0;
+    int offset = array->preamblesize;
+    for(int i = 0; i < array->size + offset; i++) {
+        pc += popcount(array->vector[offset + i]);
+    }
+    return pc;
+}
 
 int mbarray_ClearAll(MBArray * array)
 {

src/mmapbitarray.h

@@ -38,6 +38,8 @@ MBArray * mbarray_Create_Mmap(BTYPE num_bits, const char * file, const char * he
 
 void mbarray_Destroy(MBArray * array);
 
+int mbarray_Popcount(MBArray * array);
+
 int mbarray_ClearAll(MBArray * array);
 
 int mbarray_Sync(MBArray * array);

src/pybloomfilter.pyx

@@ -1,4 +1,4 @@
-VERSION = (0, 3, 14)
+VERSION = (0, 3, 15)
 AUTHOR = "Michael Axiak"
 
 __VERSION__ = VERSION
@@ -196,6 +196,12 @@ cdef class BloomFilter:
         self._assert_open()
         cbloomfilter.mbarray_Sync(self._bf.array)
 
+
+    def popcount(self):
+        self._assert_open()
+        return cbloomfilter.mbarray_Popcount(self._bf.array)
+
+
     def clear_all(self):
         self._assert_open()
         cbloomfilter.mbarray_ClearAll(self._bf.array)

tests/simpletest.py

@@ -184,6 +184,12 @@ def test_string_nofile(self):
         bf.add("test")
         self.assertEquals("test" in bf, True)
 
+    def test_popcount_nofile(self):
+        bf = pybloomfilter.BloomFilter(100, 0.01)
+        self.assertEquals(bf.popcount(), 0)
+        bf.add("hello")
+        self.assertNotEqual(bf.popcount(), 0)
+
     def test_others_nofile(self):
         bf = pybloomfilter.BloomFilter(100, 0.01)
         for elem in (1.2, 2343L, (1, 2), object(), u'\u2131\u3184'):
@@ -197,6 +203,13 @@ def _test_large_file(self, filename):
         bf.add(1234)
         self.assertEquals(1234 in bf, True)
 
+    @with_test_file
+    def test_popcount(self, filename):
+        bf = pybloomfilter.BloomFilter(1000, 0.01, filename)
+        self.assertEqual(bf.popcount(), 0)
+        bf.add("popcount")
+        self.assertNotEqual(bf.popcount(), 0)
+
     def test_name_does_not_segfault(self):
         bf = pybloomfilter.BloomFilter(100, 0.01)
         self.assertRaises(NotImplementedError, lambda: bf.name)