shop_live_scraper/libs/dy_barrage_i...

import requests
import time


from log.print_log import PrintLog
from libs.Xg04 import X_Gorgon
from libs.douyin_pb2 import DouYin
from libs.douyin_user_pb2 import DouYinUser
from libs.proxy import Proxy
from libs.en0414 import DouYinApi


class DyBarrageInfo:
    @staticmethod
    def get_data(room_id):
        room_id = str(room_id)
        proxy = Proxy.get()

        proxies = {
            "http": "http://" + proxy,
            "https": "http://" + proxy
        }

        douApi = DouYinApi('', proxies)
        result = douApi.register_device()

        device_id, iid, udid, openudid, cookie = result['device_id'], result['iid'], result['uuid'], result['openudid'], \
                                                 result['cookie']
        douApi.init_device_ids(device_id, iid, udid, openudid)
        url = 'http://webcast3-normal-c-lf.amemv.com/webcast/room/' + room_id + '/_fetch_message_polling/?'

        query_param = 'os_api=23&' \
                     'device_type=HSF-FL00&' \
                     'ssmix=a&' \
                     'manifest_version_code=7301&' \
                     'dpi=480&' \
                     'app_name=aweme&' \
                     'version_name=7.3.0&' \
                     'app_type=normal&' \
                     'ac=wifi&' \
                     'host_abi=armeabi&' \
                     'update_version_code=7302&' \
                     'channel=douyin_tengxun_wzl&' \
                     '_rticket=1629772802255&' \
                     'device_platform=android&' \
                     'iid=' + iid + '&' \
                     'version_code=730&' \
                     'device_id=' + device_id + '&' \
                     'resolution=1080*2340&' \
                     'os_version=6.0.1&' \
                     'language=zh&' \
                     'device_brand=HUAWEI&' \
                     'mcc_mnc=46000&' \
                     'ts=' + str(int(time.time())) + '&' \
                     'aid=1128'

        body = 'live_id=1&' \
               'parse_cnt=0&' \
               'recv_cnt=0&' \
               'cursor=0&' \
               'last_rtt=0&' \
               'identity=audience&' \
               'resp_content_type=protobuf&' \
               'user_id=3117857528618044'

        url = url + query_param

        x_gorgon = X_Gorgon(query_param, body)

        headers = {
            'Cookie': cookie,
            'user-agent': 'okhttp/3.10.0.1',
            "X-Gorgon": x_gorgon.get('X-Gorgon'),
            "X-Khronos": x_gorgon.get('X-Khronos'),
            "Connection": 'close',
            "X-SS-REQ-TICKET": str(int(time.time() * 1000)),
            "x-ss-stub": '6AC103439C12D2FED188040DFB6D7799',
            "Content-Type": 'application/x-www-form-urlencoded',
        }

        retry = 0
        response_json = {'room_id': room_id, 'data': {}}

        while True:
            if retry > 10:
                break

            retry += 1

        try:
            response = requests.post(url, data=body, headers=headers, verify=False, proxies=proxies)
            if(response.status_code == 200) and (response.content is not None) and (response.content != ''):
                douyin = DouYin()
                douyin.ParseFromString(response.content)

                for k, c in list(enumerate(douyin.messages)):
                    if c.method == 'WebcastChatMessage':
                        douyin_user = DouYinUser()
                        douyin_user.ParseFromString(c.payload)
                        user_info = douyin_user.user
                        response_json['data'][k] = {
                            "message_id": c.messageId,
                            "content": douyin_user.content,
                            "uid": user_info.uid,
                            "gender": user_info.gender,
                            "nickname": user_info.nickname,
                            "short_id": user_info.short_id,
                            "publisher_dy_id": user_info.dy_id,
                            "publisher_sec_uid": user_info.sec_uid
                        }

            else:
                PrintLog.print(
                    time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + '爬取http连接失败！' + str(
                        response.status_code) + '\n'
                    + Proxy.proxy_info + '\n'
                    + room_id + '\n'
                )
        except requests.exceptions.ProxyError as e:
            PrintLog.print(
                time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + '代理过期！' + str(e) + '\n'
                + room_id + '\n'
                + Proxy.proxy_info
            )
            Proxy.del_proxy(proxy)
            pass
        except requests.exceptions.ConnectTimeout as e:
            PrintLog.print(
                time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + ' ConnectTimeout！' + str(e) + '\n'
                + room_id + '\n'
                + Proxy.proxy_info
            )
            Proxy.del_proxy(proxy)
            pass
        except Exception as e:
            PrintLog.print(
                time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + '请求抛出异常！' + str(e) + '\n'
                + room_id + '\n'
                + Proxy.proxy_info
            )
            pass

        if not bool(response_json['data']):
            return False

        return response_json