class MySplitter(object):
def __init__(self):
self._v_splitter = HTMLWordSplitter()
def __call__(self, text, stopdict, *args, **kwargs):
words = self._v_splitter._split(text)
def lookup(w):
return stopdict.get(w, w)
return filter(None, map(lookup, words))
def __init__(self):
self._v_splitter = HTMLWordSplitter()