__init__.py 文件源码

def clean_text(self, txt):
        """
        function to clean a text on the basis of configurations mentioned in clean config.
        """

        txt = str(txt)

        if self.cc['lower']:
            txt = txt.lower()

        if self.cc['punctuation']:
            txt = "".join([x for x in txt if x not in punctuations])

        if self.cc['whitespace']:
            txt = "".join(txt.split()).strip()

        if self.cc['digit']:
            txt = "".join(x for x in txt if x not in "0987654321")

        return txt