From 4cb6fd4788d1eea0e53ca9a41170015a5ca17837 Mon Sep 17 00:00:00 2001
From: Mikhail Korobov <kmike84@gmail.com>
Date: Thu, 22 Sep 2016 22:07:22 +0500
Subject: [PATCH 1/2] micro-optimize HashingVectorizer and FeatureHasher

---
 sklearn/feature_extraction/_hashing.pyx | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/sklearn/feature_extraction/_hashing.pyx b/sklearn/feature_extraction/_hashing.pyx
index 201082e94cbf3..fdbf21be752dc 100644
--- a/sklearn/feature_extraction/_hashing.pyx
+++ b/sklearn/feature_extraction/_hashing.pyx
@@ -8,8 +8,6 @@ from libc.stdlib cimport abs
 cimport numpy as np
 import numpy as np
 
-from ..externals.six import string_types
-
 from sklearn.utils.murmurhash cimport murmurhash3_bytes_s32
 
 np.import_array()
@@ -45,7 +43,7 @@ def transform(raw_X, Py_ssize_t n_features, dtype):
 
     for x in raw_X:
         for f, v in x:
-            if isinstance(v, string_types):
+            if isinstance(v, basestring):
                 f = "%s%s%s" % (f, '=', v)
                 value = 1
             else:
@@ -55,13 +53,13 @@ def transform(raw_X, Py_ssize_t n_features, dtype):
                 continue
 
             if isinstance(f, unicode):
-                f = f.encode("utf-8")
+                f = (<unicode>f).encode("utf-8")
             # Need explicit type check because Murmurhash does not propagate
             # all exceptions. Add "except *" there?
             elif not isinstance(f, bytes):
                 raise TypeError("feature names must be strings")
 
-            h = murmurhash3_bytes_s32(f, 0)
+            h = murmurhash3_bytes_s32(<bytes>f, 0)
 
             array.resize_smart(indices, len(indices) + 1)
             indices[len(indices) - 1] = abs(h) % n_features

From 9de996d478b9e37ca9803b2d76502f0abe07f318 Mon Sep 17 00:00:00 2001
From: Mikhail Korobov <kmike84@gmail.com>
Date: Thu, 22 Sep 2016 22:40:10 +0500
Subject: [PATCH 2/2] fix backwards compatibility for Cython < 0.20

---
 sklearn/feature_extraction/_hashing.pyx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sklearn/feature_extraction/_hashing.pyx b/sklearn/feature_extraction/_hashing.pyx
index fdbf21be752dc..39c2b10378132 100644
--- a/sklearn/feature_extraction/_hashing.pyx
+++ b/sklearn/feature_extraction/_hashing.pyx
@@ -43,7 +43,7 @@ def transform(raw_X, Py_ssize_t n_features, dtype):
 
     for x in raw_X:
         for f, v in x:
-            if isinstance(v, basestring):
+            if isinstance(v, (str, unicode)):
                 f = "%s%s%s" % (f, '=', v)
                 value = 1
             else: