MNT Add seed parameter to private FeatureHasher transform helper (scikit-learn#14605)

TwsThomas · NicolasHug · commit 9b045d4bde6f · 2019-10-07T09:10:42.000-04:00
diff --git a/sklearn/feature_extraction/_hashing.pyx b/sklearn/feature_extraction/_hashing.pyx
@@ -17,7 +17,8 @@ from ..utils.fixes import sp_version
 np.import_array()
 
 
-def transform(raw_X, Py_ssize_t n_features, dtype, bint alternate_sign=1):
+def transform(raw_X, Py_ssize_t n_features, dtype,
+              bint alternate_sign=1, unsigned int seed=0):
     """Guts of FeatureHasher.transform.
 
     Returns
@@ -65,7 +66,7 @@ def transform(raw_X, Py_ssize_t n_features, dtype, bint alternate_sign=1):
             elif not isinstance(f, bytes):
                 raise TypeError("feature names must be strings")
 
-            h = murmurhash3_bytes_s32(<bytes>f, 0)
+            h = murmurhash3_bytes_s32(<bytes>f, seed)
 
             array.resize_smart(indices, len(indices) + 1)
             indices[len(indices) - 1] = abs(h) % n_features
diff --git a/sklearn/feature_extraction/hashing.py b/sklearn/feature_extraction/hashing.py
@@ -150,7 +150,7 @@ def transform(self, raw_X):
             raw_X = (((f, 1) for f in x) for x in raw_X)
         indices, indptr, values = \
             _hashing_transform(raw_X, self.n_features, self.dtype,
-                               self.alternate_sign)
+                               self.alternate_sign, seed=0)
         n_samples = indptr.shape[0] - 1
 
         if n_samples == 0:
diff --git a/sklearn/feature_extraction/tests/test_feature_hasher.py b/sklearn/feature_extraction/tests/test_feature_hasher.py
@@ -4,6 +4,7 @@
 import pytest
 
 from sklearn.feature_extraction import FeatureHasher
+from sklearn.feature_extraction._hashing import transform as _hashing_transform
 from sklearn.utils.testing import (ignore_warnings,
                                    fails_if_pypy)
 
@@ -45,6 +46,28 @@ def test_feature_hasher_strings():
         assert X.nnz == 6
 
 
+def test_hashing_transform_seed():
+    # check the influence of the seed when computing the hashes
+    raw_X = [["foo", "bar", "baz", "foo".encode("ascii")],
+             ["bar".encode("ascii"), "baz", "quux"]]
+
+    raw_X_ = (((f, 1) for f in x) for x in raw_X)
+    indices, indptr, _ = _hashing_transform(raw_X_, 2 ** 7, str,
+                                            False)
+
+    raw_X_ = (((f, 1) for f in x) for x in raw_X)
+    indices_0, indptr_0, _ = _hashing_transform(raw_X_, 2 ** 7, str,
+                                                False, seed=0)
+    assert_array_equal(indices, indices_0)
+    assert_array_equal(indptr, indptr_0)
+
+    raw_X_ = (((f, 1) for f in x) for x in raw_X)
+    indices_1, _, _ = _hashing_transform(raw_X_, 2 ** 7, str,
+                                         False, seed=1)
+    with pytest.raises(AssertionError):
+        assert_array_equal(indices, indices_1)
+
+
 def test_feature_hasher_pairs():
     raw_X = (iter(d.items()) for d in [{"foo": 1, "bar": 2},
                                        {"baz": 3, "quux": 4, "foo": -1}])
