MultiSimhash

simhash/__init__.py

@@ -168,6 +168,22 @@ def distance(self, another):
         return ans
 
 
+class MultiSimhash(Simhash):
+    def __init__(self, simhashes):
+        multi_f = 0
+        if not isinstance(simhashes, Iterable):
+            raise Exception('Value passed is not a list of simhashes')
+        for i in simhashes:
+            multi_f = multi_f + i.f
+        if multi_f % 8:
+            raise Exception('Simhashes do not the same length (f)')
+        multi_value = self._concatenate_simhashes(simhashes)
+        super(MultiSimhash, self).__init__(value=multi_value, f=multi_f, hashfunc=simhashes[0].hashfunc)
+
+    def _concatenate_simhashes(self, objs):
+        digests = [int_to_bytes(obj.value, obj.f_bytes) for obj in objs]
+        return bytes_to_int(b''.join(digests))
+
 class SimhashIndex(object):
 
     def __init__(self, objs, f=64, k=2, log=None):

tests/test_simhash.py

@@ -4,7 +4,7 @@
 
 from sklearn.feature_extraction.text import TfidfVectorizer
 
-from simhash import Simhash, SimhashIndex
+from simhash import MultiSimhash, Simhash, SimhashIndex
 
 
 class TestSimhash(TestCase):
@@ -146,6 +146,19 @@ def test_get_near_dup(self):
         dups = self.index.get_near_dups(s1)
         self.assertEqual(3, len(dups))
 
+class TestMultiSimhash(TestCase):
+    def test_creation(self):
+        a = Simhash('My name is John')
+        b = 1
+        self.assertRaises(Exception, [a, b])
+
+        a = Simhash(0xaaaaaaaa, f=32)
+        b = Simhash(0xbbbbbbbb, f=32)
+        ms = MultiSimhash([a, b])
+        d = Simhash(0xaaaaaaaabbbbbbbb, f=64)
+        self.assertEqual(ms.value, d.value)
+
+        self.assertEqual(ms.f, d.f)
 
 if __name__ == '__main__':
     main()