|
@@ -0,0 +1,58 @@
|
|
|
+# -*- coding: utf-8 -*-
|
|
|
+# @Author: lierqiang
|
|
|
+# @Time: 2023/4/21
|
|
|
+import argparse
|
|
|
+import os
|
|
|
+import sys
|
|
|
+
|
|
|
+sys.path.append(os.getcwd())
|
|
|
+from common.common import Common
|
|
|
+from douyin.douyin_recommend.dy_recommend_scheduling import DyRecommendScheduling
|
|
|
+from common.public import task_fun
|
|
|
+
|
|
|
+
|
|
|
+def main(log_type, crawler, task, oss_endpoint, env):
|
|
|
+ task = task_fun(task)
|
|
|
+ try:
|
|
|
+ Common.logger(log_type, crawler).info(f'开始抓取 {crawler}视频 定向榜\n')
|
|
|
+ DyRecommendScheduling.get_recommend_videos(log_type=log_type,
|
|
|
+ crawler=crawler,
|
|
|
+ task=task,
|
|
|
+ oss_endpoint=oss_endpoint,
|
|
|
+ env=env)
|
|
|
+ Common.del_logs(log_type, crawler)
|
|
|
+ Common.logger(log_type, crawler).info('抓取任务结束\n')
|
|
|
+ except Exception as e:
|
|
|
+ Common.logger(log_type, crawler).info(f"{crawler}视频异常,触发报警:{e}\n")
|
|
|
+ # Feishu.bot(log_type, crawler, f"{e}")
|
|
|
+
|
|
|
+
|
|
|
+if __name__ == "__main__":
|
|
|
+ parser = argparse.ArgumentParser() ## 新建参数解释器对象
|
|
|
+ parser.add_argument('--log_type', default='recommend') ## 添加参数,注明参数类型
|
|
|
+ parser.add_argument('--crawler', default='douyin') ## 添加参数
|
|
|
+ parser.add_argument('--strategy', default='推荐') ## 添加参数
|
|
|
+ parser.add_argument('--task') ## 添加参数
|
|
|
+ parser.add_argument('--oss_endpoint', default='outer') ## 添加参数
|
|
|
+ parser.add_argument('--env', default='dev') ## 添加参数
|
|
|
+ # parser.add_argument('--machine') ## 添加参数
|
|
|
+ args = parser.parse_args() ### 参数赋值,也可以通过终端赋值
|
|
|
+ task = {
|
|
|
+ 'task_dict': {'task_id': '17', 'task_name': '西瓜测试4.21', 'source': 'douyin', 'start_time': '1682010720000',
|
|
|
+ 'interval': '24', 'mode': 'author',
|
|
|
+ 'rule': {'duration': {'min': 40, 'max': 0}, 'play_cnt': {'min': 4000, 'max': 0},
|
|
|
+ 'period': {'min': 10, 'max': 0}, 'fans_cnt': {'min': 0, 'max': 0},
|
|
|
+ 'videos_cnt': {'min': 0, 'max': 0}, 'like_cnt': {'min': 0, 'max': 0},
|
|
|
+ 'width': {'min': 0, 'max': 0}, 'height': {'min': 0, 'max': 0}},
|
|
|
+ 'spider_name': 'run_dy_author_scheduling', 'machine': 'aliyun', 'status': '0',
|
|
|
+ 'create_time': '1682048632396', 'update_time': '1682048632396', 'operator': ''},
|
|
|
+ 'rule_dict': {'duration': {'min': 0, 'max': 0}, 'play_cnt': {'min': 0, 'max': 0},
|
|
|
+ 'period': {'min': 0, 'max': 0}, 'fans_cnt': {'min': 0, 'max': 0},
|
|
|
+ 'videos_cnt': {'min': 0, 'max': 0},
|
|
|
+ 'like_cnt': {'min': 0, 'max': 0}, 'width': {'min': 0, 'max': 0},
|
|
|
+ 'height': {'min': 0, 'max': 0}}}
|
|
|
+ main(log_type=args.log_type,
|
|
|
+ crawler=args.crawler,
|
|
|
+ task=task,
|
|
|
+ oss_endpoint=args.oss_endpoint,
|
|
|
+ env=args.env)
|