import re unsharer = re.compile(r"(\?share=|#).+$", flags=re.IGNORECASE) def sanitizeUrl(url): global unsharer ret = re.sub(unsharer, "", url) return ret def sanitizeSelector(s): s = s.replace(">a", "> a") if re.match(r'=\w+\]', s) is None: return s ret = re.sub(r"=(\w+)]", r'="\1"]', s) return ret