cc_cleanup.py

def cc_stripped(x, extended=False):
        """ strip control characters from string """
        if extended:
                # also strip extended characters
                return "".join([i for i in x if ord(i) in range(32, 126)])
        return "".join([i for i in x if ord(i) in range(32, 127)])


def parse(self, response):
	body = cc_stripped(response.body)
	response = HtmlResponse(response.url,
		encoding='utf-8'
		body=body.decode('utf-8'),
	)