diff --git a/somajo/tokenizer.py b/somajo/tokenizer.py index 0a96ffe..18836e4 100644 --- a/somajo/tokenizer.py +++ b/somajo/tokenizer.py @@ -84,7 +84,7 @@ def __init__(self, split_camel_case=False, token_classes=False, extra_info=False "*_*", "._.", ":wink:", ">_<", "*<:-)", ":!:", ":;-))"]) emoticon_list = sorted(emoticon_set, key=len, reverse=True) - self.emoticon = re.compile(r"""(?:(?:[:;]|(?