shop_live_scraper/douyin_live_...

from rds_model.rds_room_lottery_request_list import RdsRoomLotteryRequestList
import time
import json
import sys
import threading
import random
import urllib
import requests

from rds_model.db_redis import DbRedis
from log.print_log import PrintLog
from libs.Xg04 import X_Gorgon
from libs.proxy import Proxy

start_time = time.time()

def get_random(i, random_type=1):
    if random_type == 1:
        return str(random.randint(1 * 10 ** (i - 1), 1 * 10 ** i - 1))
    elif random_type == 8:
        seed = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
        sa = []
        for i in range(i):
            sa.append(random.choice(seed))
        salt = ''.join(sa)
        return salt
    else:
        seed = "1234567890abcde"
        sa = []
        for i in range(i):
            sa.append(random.choice(seed))
        salt = ''.join(sa)
        return salt

def get_random_brand_type():
    brand_type = get_random(3, random_type=8) + '-' + get_random(2, random_type=8) + '00'
    return brand_type

def get_random_trace():
    random_one = 'c' + get_random(15)
    random_two = 'ae6' + get_random(9) + '0468'
    trace_id = '00-' + str(random_one) + str(random_two) + '-' + str(random_one) + '-01'
    return trace_id

def get_mc():
    def a():
        seed = "1234567890ABCDEF"
        sa = []
        for i in range(2):
            sa.append(random.choice(seed))
        salt = ''.join(sa)
        return salt

    k = ''
    for i in range(6):
        k += a() + ':'
    return k[:-1]

def get_trace():
    trace_list = [
        '00-70f99f2209e0b045dd14266ee1da0468-70f99f2209e0b045-01',
        '00-ce7faf4409b7fcc0ae6135fdd4250468-ce7faf4409b7fcc0-01',
        '00-ce7faf3b09b7fcc0ae6042f1d8100468-ce7faf3b09b7fcc0-01',
        '00-cdd79d2309b7fcc0ae6625a4cb190468-cdd79d2309b7fcc0-01',
        '00-cefde9f009b7fcc0ae6750e1349e0468-cefde9f009b7fcc0-01',
        '00-ced2e6ef09b7fcc0ae67dd7bfe000468-ced2e6ef09b7fcc0-01',
        '00-cefbfeb509b7fcc0ae659396a6ea0468-cefbfeb509b7fcc0-01',
        '00-cefaa25409b7fcc0ae657726a3c30468-cefaa25409b7fcc0-01',
        '00-6eb478cf09b7fcc0ae6e437c9e160468-6eb478cf09b7fcc0-01',
        '00-6eb4752709b7fcc0ae69f22235260468-6eb4752709b7fcc0-01',
        '00-6eb4469109b7fcc0ae6ba6f250a90468-6eb4469109b7fcc0-01',
        '00-6eb43ea209b7fcc0ae66a34128bd0468-6eb43ea209b7fcc0-01',
    ]

    return random.choice(trace_list)

def get_live_lottery_data(room_id, result):
    domain = 'webcast5-normal-ipv6-lf.amemv.com'
    url = 'https://' + domain + '/webcast/lottery/melon/check_user_right/?'

    rticket = str(int(time.time() * 1000))
    mc = get_mc
    udid = '8604' + get_random(11)
    trace_id = get_trace()
    # trace_id = get_random_trace()
    # openudid = '3b22' + str(udid.uuid4())[-12:]
        

    device_id, iid, udid, openudid, cookie, V1, V2, device_type, device_brand = result[0], result[1], result[3], result[2], result[4], result[8], result[9], result[10], result[11]
    
    query = {
        "room_id" : str(room_id),
        "webcast_sdk_version" : "1690",
        "webcast_language" : "zh",
        "webcast_locale" :"zh_CN",
        "webcast_gps_access" : "1",
        "os_api" : "23",
        "device_type" : device_type,
        "ssmix" : "a",
        "manifest_version_code" : "120801",
        "dpi" : "640",
        "app_name" : "aweme",
        "version_name" : "12.8.0",
        "ts" : int(time.time()),
        "cpu_support64" : "true",
        "storage_type" : "0",
        "app_type" : "normal",
        "appTheme" : "dark",
        "ac" : "wifi",
        "host_abi" : "armeabi-v7a",
        "update_version_code" : "12809900",
        "channel" : "wandoujia_douyinnew_1128",
        "_rticket" : rticket,
        # "_rticket" : "1629688012123",
        "device_platform" : "android",
        "iid" : str(iid),
        "version_code" : "120800",
        "mac_address" : mc,
        #"mac_address" : "FC%3ADB%3AB3%3A56%3ABD%3AFD",
        "cdid" : "6c96979e-c729-419c-9516-3a85a7338d0c",
        "openudid" : str(openudid),
        "device_id" : str(device_id),
        "resolution" : "1440*2560",
        "os_version" : "6.0.1",
        "language" : "zh",
        "device_brand":device_brand,
        "aid" : "1128"
    }

    query_params = urllib.parse.urlencode(query)
    url = url + query_params

    body = ''

    xGorgon = X_Gorgon(query_params, body)

    userAgent = userAgent = f'com.ss.android.ugc.aweme/1208000 (Linux; U; Android 5.1.1; zh_CN; {device_type}; Build/LMY47V; Cronet/58.0.2991.0)'

    headers = {
        'Host': domain,
        'Connection': 'keep-alive',
        'Cache-Control': 'max-age=0',
        'Upgrade-Insecure-Requests': '1',
        'User-Agent': userAgent,
        # 'accept-encoding': 'gzip, deflate',
        # "x-SS-REQ-TICKET": rticket,
        "x-gorgon": xGorgon.get('X-Gorgon'),
        "x-khronos": xGorgon.get('X-Khronos'),
        'passport-sdk-version' : '17',
        'sdk-version' : '2',
        'x-ss-dp' : '1128',
        'x-tt-trace-id' : trace_id
    }

    retry = 0
    response_json = None

    while True:
        if retry > 3:
            break

        retry += 1

        proxy = Proxy.dailiyun_get()

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

        try:
            response = requests.get(
                url,
                headers=headers,
                proxies=proxies,
                timeout=8
            )
            if (response.status_code == 200) and (response.text is not None) and (response.text != ''):
                response_json = response.json()

                if (response_json.get('data') is not None):
                    print(
                        time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
                        + ' 数据获取成功！' 
                        + '\n'
                        + room_id 
                        + '\n'
                    )

                    break
                else:
                    print(
                        time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
                        + ' 数据获取失败！' 
                        + '\n'
                        + room_id 
                        + '\n'
                        + response.text
                        + Proxy.proxy_info
                    )
                
            else:
                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 + 'trace_id:' + trace_id
                    + '\n'
                    + '爬取结果：' + str(response)
                    + '\n'
                )
                time.sleep(1)
        except requests.exceptions.ProxyError as e:
            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.dailiyun_del_proxy(proxy)
            pass
        except requests.exceptions.ConnectTimeout as e:
            print(
                time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
                + ' ConnectTimeout！' 
                + str(e) 
                + '\n'
                + room_id 
                + '\n'
                + Proxy.proxy_info
            )
            Proxy.dailiyun_del_proxy(proxy)
            pass
        except Exception as e:
            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

    return response_json

def scrape():
    rds_list = RdsRoomLotteryRequestList()

    while(True):

        room_info = rds.get_request_param()
        if room_info is None:
            return None

        room_info = json.loads(room_info)
        room_id = room_info.get('room_id')
        room_id = str(room_id)

        print(
            time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
            + ' ' 
            + str(room_id)
        )

        key = 'DOUYIN_SCRAPE_DID_IID_TTREQ_1221'
        rdid = DbRedis.connect().rpop(key)
        if rdid:
            result = rdid.split('@@@')
        else:
            return None

        DbRedis.connect().lpush(key, rdid)

        try:
            response_json = get_live_lottery_data(room_id, result)
        
            if response_json is None:
                # rds_list.record_score(0)
                # rds_list.push_request_id(room_id)
                print(
                    time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
                    + ' 数据获取失败！响应数据为空！' 
                    + '\n'
                    + room_id 
                    + '\n'
                )
                # sys.exit(0)

            data = json.dumps({
                "data": response_json.get('data'),
                "extra": {
                    'room_id': room_id
                }
            })

            # rds_list.record_score(1)
            rds_list.push_data_list(data)
        except Exception as e:
            # rds_list.record_score(0)
            rds_list.push_request_id(room_id)
            print(
                time.strftime("%H:%M:%S", time.localtime()) 
                + ' ' 
                + room_id 
                + '数据异常：' 
                + str(e)
            )

    # sys.exit(0)

if __name__ == "__main__":
    print("主方法开始执行")
    # 并行线程数
    threading_count = int(sys.argv[1])

    rds = RdsRoomLotteryRequestList()

    print(
        time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
        + ' ' 
        + ' 开始执行，更新直播队列长度：' 
        + str(rds.get_len())
    )

    while True:
        sys.stdout.flush()
        
        # 减去主线程
        active_count = threading.active_count() - 1
       
        increment = threading_count - active_count
        
        if increment > 0:
            sys.stdout.flush()
        # scrape()  
            task = threading.Thread(target=scrape, args=())
            task.start()  # 准备就绪, 等待cpu执行
            # increment = increment - 1
           
        current_time = time.time()

        if current_time - start_time > 3600:
            print(
                time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) 
                + ' 主方法执行终止'
            )
            sys.exit(0)

        time.sleep(1)