| 123456789101112131415161718192021222324252627282930313233 |
- # -*- coding: utf-8 -*-
- # @Author: wangkun
- # @Time: 2023/3/27
- from common.scheduling_db import MysqlHelper
- # from scheduling_db import MysqlHelper
- # 过滤词库
- def filter_word(log_type, crawler, source, env):
- """
- 过滤词库
- :param log_type: 日志
- :param crawler: 哪款爬虫，如：xiaoniangao
- :param source: 哪款爬虫，如：小年糕
- :param env: 环境
- :return: word_list
- """
- select_sql = f""" select * from crawler_filter_word where source="{source}" """
- words = MysqlHelper.get_values(log_type, crawler, select_sql, env, action='')
- if len(words) == 0:
- return None
- word_list = []
- for word in words:
- # ss = word['filter_word']
- # print(type(ss))
- # print(ss)
- word_list.append(word['filter_word'])
- return word_list
- if __name__ == "__main__":
- print(filter_word('public', 'xiaoniangao', '小年糕', 'prod'))
|
