import pymysql import json from tqdm import tqdm spider_connection = pymysql.connect( host="rm-bp1159bu17li9hi94.mysql.rds.aliyuncs.com", port=3306, user="crawler", passwd="crawler123456@", db="piaoquan-crawler", charset="utf8mb4" ) def find_videos(trace_id): """ 查询已经发布的视频 :return: """ select_sql = f""" select recall_video_id1, recall_video_id2, recall_video_id3 from long_articles_video where trace_id = '{trace_id}'; """ cursor = spider_connection.cursor() cursor.execute(select_sql) data = cursor.fetchall() vid_tuple = data[0] return vid_tuple def insert_into_queue(video_id, publish_time_stamp, trace_id): update_sql = f""" INSERT INTO get_off_videos (video_id, publish_time, video_status, trace_id) values (%s, %s, %s, %s); """ cursor = spider_connection.cursor() cursor.execute( update_sql, (video_id, time_stamp, 1, trace_id) ) spider_connection.commit() if __name__ == '__main__': with open("tt_id.txt", encoding="utf-8") as f: data_line = f.readlines() for line in tqdm(data_line): line_ = line.split("\t") trace_id = line_[0] time_stamp = int(int(line_[1].strip()) / 1000) vid_tuple = find_videos(trace_id) for video_id in vid_tuple: try: insert_into_queue(video_id, time_stamp, trace_id) except Exception as e: print(e)