Merge pull request #589 from PrimozGodec/fix-simhash

[FIX] Simhash: Fix error when hash function is None
biolab · Oct 20, 2020 · 9c4d6ab · 9c4d6ab
2 parents 18cc6d6 + 21beac3
commit 9c4d6ab
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/orangecontrib/text/vectorization/simhash.py b/orangecontrib/text/vectorization/simhash.py
@@ -27,7 +27,11 @@ def get_shingles(tokens, n):
         return map(lambda x: ''.join(x), nltk.ngrams(tokens, n))
 
     def compute_hash(self, tokens):
-        return Simhash(self.get_shingles(tokens, self.ngram_len), f=self.f, hashfunc=self.hashfunc).value
+        values = self.get_shingles(tokens, self.ngram_len)
+        if self.hashfunc is None:
+            return Simhash(values, f=self.f).value
+        else:
+            return Simhash(values, f=self.f, hashfunc=self.hashfunc).value
 
     def int2binarray(self, num):
         return [int(x) for x in self._bin_format.format(num)]