Merge pull request #13 from LuminosoInsight/casefold-tokens

Case-fold instead of just lowercasing tokens

Former-commit-id: 95fc0c8e9d

setup.py

@@ -33,7 +33,7 @@ if sys.version_info < (3, 4):
 
 setup(
     name="wordfreq",
-    version='1.0b2',
+    version='1.0b3',
     maintainer='Luminoso Technologies, Inc.',
     maintainer_email='info@luminoso.com',
     url='http://github.com/LuminosoInsight/wordfreq/',

tests/test.py

@@ -95,6 +95,12 @@ def test_tokenization():
     # apply.
     eq_(tokenize("can.t", 'en'), ['can', 't'])
 
+
+def test_casefolding():
+    eq_(tokenize('WEISS', 'de'), ['weiss'])
+    eq_(tokenize('weiß', 'de'), ['weiss'])
+
+
 def test_phrase_freq():
     plant = word_frequency("plan.t", 'en')
     assert_greater(plant, 0)
@@ -111,7 +117,7 @@ def test_not_really_random():
     # This not only tests random_ascii_words, it makes sure we didn't end
     # up with 'eos' as a very common Japanese word
     eq_(random_ascii_words(nwords=4, lang='ja', bits_per_word=0),
-        'e e e e')
+        'rt rt rt rt')
 
 
 @raises(ValueError)

wordfreq/__init__.py

@@ -149,7 +149,7 @@ def simple_tokenize(text):
     sequence, but they are if they appear internally. "cats'" is not a token,
     but "cat's" is.
     """
-    return [token.lower() for token in TOKEN_RE.findall(text)]
+    return [token.casefold() for token in TOKEN_RE.findall(text)]
 
 mecab_tokenize = None
 def tokenize(text, lang):