def download_video(self, aweme_id):
query_params = self.common_params
query_params['aweme_id'] = aweme_id
params = {**query_params, **self.common_params}
url = URL.video_detail_url()
real_url = gen_url(self.token, url, params)
post_data = {
"aweme_id": aweme_id
}
cookies = self.__COOKIES
cookies['install_id'] = str(self.__device["install_id"])
resp = fetch(real_url,
data=post_data,
cookies=cookies,
headers=self.__HEADERS,
timeout=3).json()
try:
play_addr_raw = resp['aweme_detail']['video']['play_addr']['url_list']
play_addr = play_addr_raw[0]
print(play_addr)
content = fetch(play_addr).content
except:
self.logger.warning("提取视频信息失败...")
content = None
return content
def gen_real_url(token, raw_url, query):
if isinstance(query, dict):
query = params2str(query)
url = raw_url + "?" + query
resp = fetch(URL.api_sign(token), json={"url": url}, method="post")
if not resp:
print("你的当日 token 次数已经用完，请明天再来尝试吧...")
raise BaseException("you have run out of token.Please try tomorrow")
resp_json = resp.json()
real_url = resp_json['url']
return real_url
def post(user_id):
device = gen_device(TOKEN)
common_params = gen_common_params(device)
count = 0
max_cursor = 0
while True:
query_params = {"count": 21 if not count else count, "user_id": user_id, "max_cursor": max_cursor}
search_params = {**common_params, **query_params}
real_url = gen_real_url(TOKEN, URL.post_url(), search_params)
cookies = COMMON_COOKIES
cookies['install_id'] = str(device["install_id"])
# download video
resp_json = fetch(real_url,
verify=False,
cookies=cookies,
headers=COMMON_HEADERS,
timeout=3).json()
# print(json.dumps(resp_json))
results = []
for video_info in resp_json.get("aweme_list"):
aweme_id = video_info.get("aweme_id")
play_url = aweme_id_video_url(aweme_id)
if not play_url:
print("Failed grab <{}> video play url".format(aweme_id))
continue
video_info["play_url"] = play_url
results.append(data_to_video(video_info))
yield results
max_cursor = resp_json.get("max_cursor")
count = 12
if resp_json.get("has_more") != 1:
print("%s post video spider done!" % user_id)
break
def gen_device(token):
resp = fetch(URL.api_device(token), timeout=10).json()
return resp