因为代理实效导致的重试还是使用实效的那个IP about feapder HOT 5 OPEN

class S1688(feapder.AirSpider):
    __custom_setting__ = {
        # "USE_SESSION": True,

        "SPIDER_THREAD_COUNT": 12,
        "PROXY_ENABLE": True,
        # "SPIDER_SLEEP_TIME": [2, 5],
        # "LOG_LEVEL": "INFO",
        "PROXY_EXTRACT_API": "http://v2.api.juliangip.com/dynamic/getips",
    }

    def download_midware(self, request: Request):
        cna = ''.join(random.choices(list('DwftHIHbiXICAQHA8429Gdvc'), k=24))
        request.headers = {
            "Cookie": f"cna={cna};",
            "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/113.0.0.0 Safari/537.36 Edg/113.0.1774.42"
        }
        return request

    def exception_request(self, request: Request, response, e):
        request.proxies_pool.tag_proxy(request.requests_kwargs.get("proxies"), -1)

    def start_requests(self):
        dire_list = load_dire_list()
        for dire in dire_list:
            url = f"https://s.1688.com/selloffer/offer_search.htm?keywords={parse.quote('%s %s 中药材' % (dire['dire_name'], dire['dire_location']), encoding='gbk')}&spm="
            yield feapder.Request(url=url, dire=dire, verify=False)

    def parse(self, request: Request, response: Response):
        dire = request.dire
        try:
            data_str_result = re.search("window.data.offerresultData = successDataCheck\((.*)\)", response.text)
            if not data_str_result: return
            data_str = data_str_result.group(1)
            data = json.loads(data_str)['data']
            offer_list = data.get("offerList", [])
            for drug in offer_list:
                drug_id = drug['id']
                title = drug['information']['subject']
                if dire['dire_name'] not in title: continue
                drug.update(dire)
                yield feapder.Request(url=f"https://detail.1688.com/offer/{drug_id}.html",
                                      callback=self.parse_detail, dire=dire, dire_item=drug)
        except Exception as e:
            request.proxies_pool.tag_proxy(request.requests_kwargs.get("proxies"), -1)
            raise Exception(f"链接:{request.url}被触发风控，无法正常获取数据，尝试重试！")

    def parse_detail(self, request: Request, response: Response):
        dire_item = request.dire
        drug_info = request.dire_item
        data_match_str_result = re.search("window.__INIT_DATA=(.*)", response.text)
        if not data_match_str_result: return
        data_match_str = data_match_str_result.group(1)
        data = json.loads(data_match_str)
        item = Item()
        item.item_name = "s1688"
        item.table_name = item.item_name
        item.update({"tempModel": data['globalData']['tempModel']})
        item.update({"skuInfoMap": data['globalData']['skuModel']['skuInfoMap']})
        item.update({"skuModel": data['globalData']['skuModel']})
        item.update({"orderParam": data['globalData']['orderParamModel']['orderParam']})

        attr_param_arr = []
        module = next((x for x in data['modules'] if x['name'] == '@ali/tdmod-od-pc-layout-detail-tab-container'), None)
        if module is not None:
            children = module.get('children', [])
            if len(children) > 0:
                attr_param_arr = [i for i in children if i['name'] == '@ali/tdmod-od-pc-attribute-new']

        if not attr_param_arr: return
        attr_param = attr_param_arr[0]
        item.update({"attrList": data['data'][attr_param['uuid']]['data']})
        item['search_key'] = dire_item['dire_name']
        item['dire_spec'] = dire_item['dire_spec']
        item['ID'] = dire_item['ID']
        item.update(drug_info)

        log.info(f"s1688 {item['ID']}-{item['tempModel']['offerTitle']}-{item['search_key']}-{item['dire_spec']}")

        yield item