def __init__(self, max_words, maxlen, embeded_size, embedding_matrix=None, new=True, path=None): self.max_words = max_words self.maxlen = maxlen self.parser = NewsParser() self.cleaner = Cleaner() self.tokenizer = Tokenizer(self.max_words) self.embeded_size = embeded_size self.embedding_matrix = embedding_matrix if new: self._model = self.create_model() else: self._model = load_model(path)
def test(htmls): for key, html in htmls.items(): NewsParser().parse(html)