def __init__(self, preserve_case: Boolean(), reduce_len: Boolean(),
strip_handles: Boolean()):
self.preserve_case = preserve_case
self.reduce_len = reduce_len
self.strip_handles = strip_handles
NltkTokenizer.__init__(self)
_TweetTokenizer.__init__(
self,
preserve_case=preserve_case,
reduce_len=reduce_len,
strip_handles=strip_handles,
)
def __init__(self, preserve_case=False, reduce_len=True, strip_handles=False,
normalize_usernames=True, normalize_urls=True, keep_allupper=True):
TweetTokenizer.__init__(self, preserve_case=preserve_case, reduce_len=reduce_len,
strip_handles=strip_handles)
self.keep_allupper = keep_allupper
self.normalize_urls = normalize_urls
self.normalize_usernames = normalize_usernames
if normalize_usernames:
self.strip_handles = False
if self.preserve_case:
self.keep_allupper = True
def __init__(self, preserve_case=True,
strip_handles=True, reduce_len=True):
TweetTokenizer.__init__(self, preserve_case,
strip_handles, reduce_len)
