Fix stowords filtering

orangecontrib/text/preprocess/filter.py

@@ -78,7 +78,7 @@ def supported_languages():
         except LookupError:  # when no NLTK data is available
             pass
 
-        return [file.capitalize() for file in stopwords_listdir]
+        return sorted(file.capitalize() for file in stopwords_listdir)
 
     @wait_nltk_data
     def __init__(self, language='English', word_list=None):
@@ -96,7 +96,8 @@ def language(self, value):
         if not self._language:
             self.stopwords = []
         else:
-            self.stopwords = set(stopwords.words(self.language.lower()))
+            self.stopwords = set(
+                x.strip() for x in stopwords.words(self.language.lower()))
 
     def __str__(self):
         config = ''

orangecontrib/text/tests/test_preprocess.py

@@ -281,6 +281,20 @@ def test_stopwords(self):
         self.assertFalse(filter.check('a'))
         self.assertTrue(filter.check('filter'))
 
+        self.assertListEqual(
+            ["snake", "house"],
+            filter(["a", "snake", "is", "in", "a", "house"]))
+
+    def test_stopwords_slovene(self):
+        filter = preprocess.StopwordsFilter('slovene')
+
+        self.assertFalse(filter.check('in'))
+        self.assertTrue(filter.check('abeceda'))
+
+        self.assertListEqual(
+            ["kača", "hiši"],
+            filter(["kača", "je", "v", "hiši", "in"]))
+
     def test_lexicon(self):
         filter = preprocess.LexiconFilter(['filter'])
         self.assertFalse(filter.check('false'))