query_regon.py 文件源码

def get_queryset(self, krs, nip, regon, google, no_regon, no_nip):
        regon = regon or []
        nip = nip or []

        self.processor = html2text.HTML2Text()
        self.processor.ignore_emphasis = True
        self.processor.bypass_tables = True
        self.processor.ignore_links = True

        self.session = requests.Session()
        for keyword in tqdm(google or []):
            if not no_regon:
                result = self.search_google("{} REGON".format(keyword), REGON_PATTERN)
                print("For '{}' found {}".format(keyword, result))
                regon += result
            if not no_nip:
                result = self.search_google("{} NIP".format(keyword), NIP_PATTERN)
                print("For '{}' found {}".format(keyword, result))
                nip += [x.replace('-', '') for x in result if len(x.replace('-', '')) == 10]
        queries = [{'krs': v} for v in set(krs)] if krs else []
        queries += [{'nip': v} for v in set(nip)] if nip else []
        queries += [{'regon': v} for v in set(regon)] if regon else []

        return queries