Search code examples
pythonnltktokenizesentence

Strange behaviour with nltk sentence tokenizer and special characters

I get some strange behavior when using the sent_tokenizer for German text.

Example Code:

sent_tokenizer = nltk.data.load('tokenizers/punkt/german.pickle')
for sent in sent_tokenizer.tokenize("Super Qualität. Tolles Teil.")
      print sent

This fails with the error:

Traceback (most recent call last):
for sent in sent_tokenize("Super Qualität. Tolles Teil."):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/__init__.py", line 82, in sent_tokenize
    return tokenizer.tokenize(text)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1270, in tokenize
    return list(self.sentences_from_text(text, realign_boundaries))
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1318, in sentences_from_text
    return [text[s:e] for s, e in self.span_tokenize(text, realign_boundaries)]
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1309, in span_tokenize
    return [(sl.start, sl.stop) for sl in slices]
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1348, in _realign_boundaries
    for sl1, sl2 in _pair_iter(slices):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 354, in _pair_iter
    prev = next(it)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1324, in _slices_from_text
    if self.text_contains_sentbreak(context):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1369, in text_contains_sentbreak
    for t in self._annotate_tokens(self._tokenize_words(text)):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 1504, in _annotate_second_pass
    for t1, t2 in _pair_iter(tokens):
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 354, in _pair_iter
    prev = next(it)
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 621, in _annotate_first_pass
    for aug_tok in tokens:
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nltk/tokenize/punkt.py", line 586, in _tokenize_words
    for line in plaintext.split('\n'):
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 6: ordinal not in range(128)

whereas:

  sent_tokenizer = nltk.data.load('tokenizers/punkt/german.pickle')
  for sent in sent_tokenizer.tokenize("Super Qualität des Produktes. Tolles Teil.")
      print sent

works perfectly

Solution

  • I found the solution on the nltk homepage.

    Caution: when tokenizing a Unicode string, make sure you are not using an encoded version of the string (it may be necessary to decode it first, e.g. with s.decode("utf8").

    So 

    text = "Super Qualität. Tolles Teil."
sent_tokenizer = nltk.data.load('tokenizers/punkt/german.pickle')
for sent in sent_tokenizer.tokenize(text.decode('utf8')):
      print sent

    works like a charm.