first commit

.gitignore

@@ -0,0 +1,8 @@
+.idea/
+.vscode/
+__pycache__/
+venv/
+
+*.jpg
+*.jpeg
+*.png

chrome/130-0008/130.0.6723.92.manifest

@@ -0,0 +1,8 @@
+<assembly
+  xmlns='urn:schemas-microsoft-com:asm.v1' manifestVersion='1.0'>
+  <assemblyIdentity
+      name='130.0.6723.92'
+      version='130.0.6723.92'
+      type='win32'/>
+  <file name='chrome_elf.dll'/>
+</assembly>

chrome/130-0008/Extensions/external_extensions.json

@@ -0,0 +1,5 @@
+// This json file will contain a list of extensions that will be included
+// in the installer.
+
+{
+}

chrome/130-0008/MEIPreload/manifest.json

@@ -0,0 +1,8 @@
+{
+  "name": "MEI Preload", 
+  "icons": {}, 
+  "version": "1.0.7.1652906823", 
+  "manifest_version": 2, 
+  "update_url": "https://clients2.google.com/service/update2/crx", 
+  "description": "Contains preloaded data for Media Engagement"
+}

chrome/130-0008/PrivacySandboxAttestationsPreloaded/manifest.json

@@ -0,0 +1,6 @@
+{
+  "manifest_version": 2,
+  "name": "Privacy Sandbox Attestations",
+  "version": "2024.8.23.0",
+  "pre_installed": true
+}

chrome/130-0008/PrivacySandboxAttestationsPreloaded/privacy-sandbox-attestations.dat

@@ -0,0 +1,258 @@
+
+
https://2k.comhttps://33across.comhttps://360yield.comhttps://3lift.comhttps://ad-score.com
+https://ad.gthttps://adentifi.comhttps://adform.nethttps://adingo.jphttps://admatrix.jphttps://admixer.nethttps://adnami.iohttps://adnxs.comhttps://adsafeprotected.comhttps://adsrvr.orghttps://adthrive.comhttps://advividnetwork.comNhttps://aggregation-service-site-dot-clz200258-datateam-italy.ew.r.appspot.comhttps://anonymised.iohttps://appier.nethttps://artistunited.comhttps://avads.nethttps://ayads.iohttps://bidswitch.nethttps://bidtheatre.nethttps://bing.comhttps://blendee.comhttps://bounceexchange.comhttps://btloader.comhttps://bypass.jphttps://casalemedia.comhttps://cdn-net.comhttps://clickonometrics.plhttps://connected-stories.comhttps://crcldu.comhttps://creativecdn.comhttps://criteo.comhttps://ctnsnet.comhttps://dabbs.nethttps://daum.nethttps://display.iohttps://dotdashmeredith.comhttps://dotomi.comhttps://doubleclick.nethttps://dynalyst.jphttps://edkt.iohttps://effinity.frhttps://ezoic.comhttps://fanbyte.comhttps://flashtalking.comhttps://fout.jphttps://funplus.comhttps://fwmrm.nethttps://gama.globohttps://ghtinc.comhttps://gmossp-sp.jphttps://google-analytics.comhttps://gsspat.jphttps://gumgum.comhttps://guoshipartners.comhttps://html-load.comhttps://im-apps.nethttps://impact-ad.jphttps://imrworldwide.comhttps://indexww.comhttps://inmobi.comhttps://innovid.comhttps://jivox.comhttps://kelkoogroup.nethttps://kidoz.nethttps://ladsp.comhttps://lucead.comhttps://mail.ruhttps://media.nethttps://mediaintelligence.dehttps://mediamath.comhttps://mediavine.comhttps://microad.jphttps://naver.comhttps://nhnace.comhttps://onetag-sys.comhttps://openx.nethttps://optable.cohttps://outbrain.com+https://privacy-sandbox-demos-ad-server.dev'https://privacy-sandbox-demos-dsp-a.dev'https://privacy-sandbox-demos-dsp-b.dev%https://privacy-sandbox-demos-dsp.dev'https://privacy-sandbox-demos-ssp-a.dev'https://privacy-sandbox-demos-ssp-b.dev%https://privacy-sandbox-demos-ssp.dev https://privacy-sandbox-test.com0https://privacy-sandcastle-dev-ad-server.web.app-https://privacy-sandcastle-dev-dsp-a1.web.app-https://privacy-sandcastle-dev-dsp-b1.web.app*https://privacy-sandcastle-dev-dsp.web.app,https://privacy-sandcastle-dev-ssp-a.web.app,https://privacy-sandcastle-dev-ssp-b.web.app*https://privacy-sandcastle-dev-ssp.web.apphttps://pub.networkhttps://pubmatic.comhttps://pubtm.comhttps://quantserve.comhttps://relevant-digital.comhttps://sascdn.comhttps://shinystat.comhttps://singular.nethttps://sportradarserving.comhttps://t13.iohttps://teads.tvhttps://theryn.iohttps://tncid.apphttps://toponad.comhttps://tpmark.nethttps://tribalfusion.comhttps://triptease.iohttps://uinterbox.comhttps://uol.com.br
+https://vg.nohttps://vpadn.comhttps://washingtonpost.comhttps://yahoo.co.jphttps://yahoo.comhttps://yandex.ruhttps://yelp.com
+https://connatix.com
+


+https://open-bid.com
+

+https://getyourguide.com
+

+https://snapchat.com
+

+https://atomex.net
+
+https://grxchange.gr
+

+https://boost-web.com
+


+https://adswizz.com
+

+https://undertone.com
+
"
+https://appsflyersdk.com
+
+https://trip.com
+

+https://gokwik.co
+%
+https://creative-serving.com
+

+https://pinterest.com
+

+https://finn.no
+

+https://socdm.com
+

!
+https://ebayadservices.com
+

+https://adroll.com
+#
+https://youronlinechoices.eu
+

+https://eloan.co.jp
+

+https://ad-stir.com
+

+https://semafor.com
+


+https://bidtellect.com
+

+https://metro.co.uk
+

+https://unrulymedia.com
+

+https://seedtag.com
+
!
+https://audience360.com.au
+

+https://torneos.gg
+

+https://getcapi.co
+

+https://r2b2.io
+
6
+/https://ptb-msmt-static-5jyy5ulagq-uc.a.run.app
+

+https://quora.com
+

+https://jkforum.net
+
+https://appconsent.io
+
%
+https://audienceproject.com
+

+https://storygize.net
+

+https://primecaster.net
+

+https://yieldlab.net
+

+https://logly.co.jp
+

+https://disqus.com
+

+https://acxiom.com
+
%
+https://wepowerconnections.com
+
$
+https://lab-dotmetrics.ninja
+
+https://2trk.info
+

+https://ebis.ne.jp
+

+https://presage.io
+

+https://aqfer.com
+
&
+https://googleadservices.com
+
+https://weborama.fr
+
+https://onet.pl
+
 
+https://worldhistory.org
+

+https://iobeya.com
+
 
+https://apex-football.com
+


+https://nexxen.tech
+
"
+https://rubiconproject.com
+

+https://aniview.com
+

+https://usemax.de
+

+https://facebook.com
+
(
+ https://smadexprivacysandbox.com
+!
+https://dailymotion.com
+

+https://shinobi.jp
+

+https://linkedin.com
+
+https://appsflyer.com
+"
+https://media6degrees.com
+

+https://cpx.to
+

+https://tya-dev.com
+
'
+ https://healthproductsforyou.com
+

+https://admission.net
+

+https://appscience.inc
+
+https://s-f.tech
+

+https://lwadm.com
+
1
+(https://paa-reporting-advertising.amazon
+
+https://convertunits.com
+


+https://bluems.com
+
"
+https://authorizedvault.com
+


+https://verve.com
+

+https://beaconmax.com
+

+https://cazamba.com
+

+https://globo.com
+

+https://kargo.com
+

+https://atirun.com
+

 
+https://tailtarget.com
+

+https://dreammail.jp
+

+https://samplicio.us
+

+
+https://wp.pl
+
"
+https://kompaspublishing.nl
+

#
+https://marutishanbhag.com
+
 
+https://postrelease.com
+

+https://akpytela.cz
+

+https://paapi.ai
+
+https://coupang.com
+
%
+https://googlesyndication.com
+
+https://adscale.de
+
 
+https://tangooserver.com
+

+https://taboola.com
+

+https://yieldmo.com
+

+https://gunosy.com
+

+https://docomo.ne.jp
+

+https://thesun.co.uk
+

+https://vidazoo.com
+

+https://pmdragonfly.com
+

+https://stackadapt.com
+
"
+https://explorefledge.com
+?
+6https://protected-audience-api-advertiser.onrender.com
+
#
+https://adsmeasurement.com
+

+https://deepintent.com
+

+https://permutive.app
+


+https://superfine.org
+

+https://fandom.com
+
!
+https://sharethrough.com
+
!
+https://weborama-tech.ru
+"
+https://amazon-adsystem.com
+

+https://pontiac.media
+
 
+https://dailymail.co.uk
+

+https://i-mobile.co.jp
+

+https://tiktok.com
+

+https://moshimo.com
+
7
+/https://shared-storage-demo-publisher-a.web.app
+
+https://retargetly.com
+

+https://a-mo.net
+

+https://demand.supply
+

+https://sephora.com
+


+https://momento.dev
+

+https://mobon.net
+

"
+https://audiencemanager.de
+

+https://trkkn.com
+

+https://sitescout.com

chrome/130-0008/sig.txt

@@ -0,0 +1 @@
+acf29cbb6f7846fead276a05ae05a1c8

chrome/130-0008/vk_swiftshader_icd.json

@@ -0,0 +1 @@
+{"file_format_version": "1.0.0", "ICD": {"library_path": ".\\vk_swiftshader.dll", "api_version": "1.0.5"}}

const.py

@@ -0,0 +1,27 @@
+import os
+BASE_PATH = os.path.dirname(__file__)
+SAFE = '!'
+MIMETYPE = {
+    'jpg': 'image/jpeg',
+    'jpeg': 'image/jpeg',
+    'png': 'image/png',
+    # 'gif': 'image/gif',
+    'mp4': 'video/mp4',
+}
+
+PIC_TYPE = ('image/jpeg', 'image/png')
+
+
+ELEMENT = {
+}
+
+# ucloud-us3
+PUBLIC_KEY = '4ZF9nrEbA93FDL29606cKFGYEb59Zu9ip'         # 账户公钥
+PRIVATE_KEY = 'FD9DgHZVe4Q9GevRb8QcwBD6S8Iesv48ZBsWImwEGPZr'        # 账户私钥
+BUCKET = 'fanzhixitong'             # 空间名称
+UPLOAD_SUFFIX = '.jpn-tky.ufileos.com'
+DOWNLOAD_SUFFIX = '.jpn-tky.ufileos.com'
+
+# DOVE PROXY
+DOVE_USER = 'fetdog'
+DOVE_PWD = 'aAa123456aAa'

exceptions.py

@@ -0,0 +1,6 @@
+class AuthException(Exception):
+    pass
+
+
+class OperationFailed(Exception):
+    pass

logger.py

@@ -0,0 +1,29 @@
+import os
+import logging
+
+log_path = os.path.join(os.path.dirname(__file__), 'log')
+simple_fmt = logging.Formatter('[%(asctime)s-%(levelname)s] %(message)s')
+
+
+def logger_settings(name, file_log=True, level=0):
+    logger = logging.getLogger(name)
+    if file_log:
+        file_log = logging.FileHandler(os.path.join(log_path, f'{name}.log'), encoding='utf-8')
+        file_log.setFormatter(simple_fmt)
+        file_log.setLevel(level)
+        logger.addHandler(file_log)
+        logger.setLevel(level)
+    return logger
+
+
+def record_full_log(logger, err):
+    trace = err.__traceback__
+    while trace:
+        file = trace.tb_frame.f_code.co_filename
+        line = trace.tb_lineno
+        trace = trace.tb_next
+        logger.error(f'    file:{file}, line:{line}')
+    logger.error(f'    error: {str(err)}')
+
+
+error_logger = logger_settings('error')

main.py

@@ -0,0 +1,67 @@
+from concurrent.futures.thread import ThreadPoolExecutor
+from spider.task import *
+from logger import error_logger, record_full_log
+
+TASK_TYPE = {
+    0: post,
+}
+
+HOST = "http://118.193.40.152:8091"
+
+
+def get_task(platform):
+    # if lock._block.locked():
+    #     return
+    response = requests.get(f'{HOST}/services/task', json={'platform': platform})
+    result = response.json()
+    if result['code'] == 0:
+        return result
+
+
+def task_callback(task_type, queue_id, data, status=1, err_msg=''):
+    response = requests.post(
+        f'{HOST}/services/task/callback',
+        json={
+            'task_type': task_type,
+            'queue_id': queue_id,
+            'status': status,
+            'data': data,
+            'err_msg': err_msg,
+        }
+    )
+    result = response.json()
+    if result['code'] != 0:
+        raise RuntimeError(f"任务回调失败:{result['msg']}")
+
+
+def execute_task(queue_id, task_type, **kwargs):
+    try:
+        result = TASK_TYPE.get(task_type)(**kwargs)
+        task_callback(task_type, queue_id, data=result)
+    except AuthException as e:
+        record_full_log(error_logger, e)
+        task_callback(task_type, queue_id, data={}, status=4, err_msg=str(e))
+    except Exception as e:
+        record_full_log(error_logger, e)
+        task_callback(task_type, queue_id, data={}, status=0, err_msg=str(e))
+
+
+def main():
+    with ThreadPoolExecutor(max_workers=1) as t:
+        while True:
+            try:
+                task = get_task(0)
+                if task is None:
+                    time.sleep(10)
+                    continue
+
+                task['data']['queue_id'] = task['queue_id']
+                task['data']['task_type'] = task['task_type']
+                t.submit(execute_task, **task['data'])
+            except Exception as e:
+                error_logger.error(f'Main Error: {e}')
+                time.sleep(10)
+
+
+if __name__ == '__main__':
+    main()

uclouds3.py

@@ -0,0 +1,55 @@
+from io import BytesIO
+
+from ufile import filemanager, config, httprequest
+from const import PUBLIC_KEY, PRIVATE_KEY, UPLOAD_SUFFIX, DOWNLOAD_SUFFIX
+
+
+def _download_file_save_to_bio(url, header):
+    try:
+        response = httprequest.requests.get(url, headers=header, stream=True)
+    except httprequest.RequestException as e:
+        return None, httprequest.ResponseInfo(None, e)
+    if response.status_code in [200, 206]:
+        bio = BytesIO()
+        for block in response.iter_content(config.BLOCKSIZE):
+            bio.write(block)
+    else:
+        return
+    return bio
+
+
+class BioFileManager(filemanager.FileManager):
+    def download_bytes_file(self, bucket, key, isprivate=True, expires=None, content_range=None, header=None):
+        """
+        :param bucket: string类型, UFile空间名称
+        :param key: string类型， 下载文件在空间中的名称
+        :param isprivate: boolean类型，如果为私有空间则为True
+        :param expires: integer类型，私有文件链接有效时间
+        :param content_range: tuple类型，元素为两个整型
+        :param header: dict类型，http 请求header，键值对类型分别为string，比如{'User-Agent': 'Google Chrome'}
+        :return: ret: 如果http状态码为[200, 204, 206]之一则返回None，否则如果服务器返回json信息则返回dict类型，键值对类型分别为string, unicode string类型，否则返回空的dict
+        :return:  ResponseInfo: 响应的具体信息，UCloud UFile 服务器返回信息或者网络链接异常
+        """
+        if header is None:
+            header = dict()
+        else:
+            filemanager._check_dict(header)
+
+        if expires is None:
+            expires = config.get_default('expires')
+
+        if 'User-Agent' not in header:
+            header['User-Agent'] = config.get_default('user_agent')
+
+        if isinstance(content_range, tuple) and len(content_range) == 2:
+            header['Range'] = 'bytes=' + '-'.join(map(lambda x: str(x), content_range))
+
+        if not isprivate:
+            url = self.public_download_url(bucket, key)
+        else:
+            url = self.private_download_url(bucket, key, expires, header, True)
+
+        return _download_file_save_to_bio(url, header)
+
+
+client = BioFileManager(PUBLIC_KEY, PRIVATE_KEY, UPLOAD_SUFFIX, DOWNLOAD_SUFFIX)