Use long long on Python 3

rth · rth · commit 34f53c4edec1 · 2017-09-01T12:47:57.000+02:00
diff --git a/sklearn/feature_extraction/text.py b/sklearn/feature_extraction/text.py
@@ -19,6 +19,7 @@
 from operator import itemgetter
 import re
 import unicodedata
+import sys
 
 import numpy as np
 import scipy.sparse as sp
@@ -763,8 +764,13 @@ def _count_vocab(self, raw_documents, fixed_vocab):
 
         analyze = self.build_analyzer()
         j_indices = []
-        # indptr can overflow in 32 bit, always use 64 bit
-        indptr = _make_int_array(dtype='l')
+        # indptr can overflow in 32 bit, always use 64 bit when possible
+        if sys.version_info >= (3, 3):
+            # use long long for 64 bit integers on Windows
+            indptr = _make_int_array(dtype='q')
+        else:
+            indptr = _make_int_array(dtype='l')
+
         values = _make_int_array()
         indptr.append(0)
         for doc in raw_documents:
@@ -796,7 +802,7 @@ def _count_vocab(self, raw_documents, fixed_vocab):
                 indices_dtype = np.int_
             else:
                 raise ValueError(('sparse CSR array has {} non-zero '
-                                  'elements and require 64 bit indexing, '
+                                  'elements and requires 64 bit indexing, '
                                   ' which is unsupported with scipy {}. '
                                   'Please upgrade to scipy >=0.14')
                                  .format(indptr[-1], '.'.join(sp_version)))
