scikit-learn
diff --git a/‎sklearn/feature_extraction/tests/test_text.py
Copy file name to clipboardExpand all lines: sklearn/feature_extraction/tests/test_text.py
+13-12Lines changed: 13 additions & 12 deletions b/‎sklearn/feature_extraction/tests/test_text.py
Copy file name to clipboardExpand all lines: sklearn/feature_extraction/tests/test_text.py
+13-12Lines changed: 13 additions & 12 deletions
@@ -29,8 +29,7 @@
 from numpy.testing import assert_array_equal
 from sklearn.utils import IS_PYPY
 from sklearn.utils._testing import (assert_almost_equal,
-                                    assert_warns_message, assert_raise_message,
-                                    assert_no_warnings,
+                                    assert_raise_message,
                                     fails_if_pypy,
                                     assert_allclose_dense_sparse,
                                     skip_if_32bit)
@@ -386,8 +385,8 @@ def test_countvectorizer_uppercase_in_vocab():
                " be matched with any documents")
 
     vectorizer = CountVectorizer(lowercase=True, vocabulary=vocabulary)
-    assert_warns_message(UserWarning, message,
-                         vectorizer.fit_transform, vocabulary)
+    with pytest.warns(UserWarning, match=message):
+        vectorizer.fit_transform(vocabulary)
 
 
 def test_tf_idf_smoothing():
@@ -429,8 +428,8 @@ def test_tfidf_no_smoothing():
     tr = TfidfTransformer(smooth_idf=False, norm='l2')
 
     in_warning_message = 'divide by zero'
-    assert_warns_message(RuntimeWarning, in_warning_message,
-                         tr.fit_transform, X).toarray()
+    with pytest.warns(RuntimeWarning, match=in_warning_message):
+        tr.fit_transform(X).toarray()
 
 
 def test_sublinear_tf():
@@ -1213,27 +1212,29 @@ def _check_stop_words_consistency(estimator):
 
 @fails_if_pypy
 def test_vectorizer_stop_words_inconsistent():
-    lstr = "['and', 'll', 've']"
+    lstr = r"\['and', 'll', 've'\]"
     message = ('Your stop_words may be inconsistent with your '
                'preprocessing. Tokenizing the stop words generated '
                'tokens %s not in stop_words.' % lstr)
     for vec in [CountVectorizer(),
                 TfidfVectorizer(), HashingVectorizer()]:
         vec.set_params(stop_words=["you've", "you", "you'll", 'AND'])
-        assert_warns_message(UserWarning, message, vec.fit_transform,
-                             ['hello world'])
+        with pytest.warns(UserWarning, match=message):
+            vec.fit_transform(['hello world'])
         # reset stop word validation
         del vec._stop_words_id
         assert _check_stop_words_consistency(vec) is False
 
     # Only one warning per stop list
-    assert_no_warnings(vec.fit_transform, ['hello world'])
+    with pytest.warns(None) as record:
+        vec.fit_transform(['hello world'])
+    assert not len(record)
     assert _check_stop_words_consistency(vec) is None
 
     # Test caching of inconsistency assessment
     vec.set_params(stop_words=["you've", "you", "you'll", 'blah', 'AND'])
-    assert_warns_message(UserWarning, message, vec.fit_transform,
-                         ['hello world'])
+    with pytest.warns(UserWarning, match=message):
+        vec.fit_transform(['hello world'])
 
 
 @skip_if_32bit