Merge pull request #251 from behrtam/change-separator-word-count

word-count: Change separator from whitespace to non alphanumeric

Author: Dog

word-count/example.py

@@ -2,7 +2,6 @@
 
 
 def word_count(text):
-    """Return a Counter object that maps from the words contained in
-    the phrase to their respective counts
-    """
-    return Counter(text.lower().split())
+    replace_nonalpha = lambda c: c.lower() if c.isalnum() else ' '
+    text = ''.join(replace_nonalpha(c) for c in text)
+    return Counter(text.split())

word-count/word_count_test.py

@@ -25,7 +25,7 @@ def test_count_multiple_occurences(self):
 
     def test_preserves_punctuation(self):
         self.assertEqual(
-            {'car': 1, 'carpet': 1, 'as': 1, 'java': 1, ':': 2, 'javascript!!&@$%^&': 1},
+            {'car': 1, 'carpet': 1, 'as': 1, 'java': 1, 'javascript': 1},
             word_count('car : carpet as java : javascript!!&@$%^&')
         )
 
@@ -63,5 +63,11 @@ def test_tabs(self):
                        'want your bad romance')
         )
 
+    def test_non_alphanumeric(self):
+        self.assertEqual(
+            {'hey': 1, 'my': 1, 'spacebar': 1, 'is': 1, 'broken': 1},
+            word_count('hey,my_spacebar_is_broken.')
+        )
+
 if __name__ == '__main__':
     unittest.main()