settings.py 1.3 KB

1234567891011121314151617181920212223242526272829303132333435
  1. from common.databases import mongo_table, redis_client
  2. '''Mongo'''
  3. MGO_DATABASE = 'shujuziyuan'
  4. '''判重库'''
  5. MGO_REPETITION = mongo_table(db=MGO_DATABASE, name='repetition_url')
  6. '''去重域名表'''
  7. MGO_REMOVAL_DOMAIN = mongo_table(db=MGO_DATABASE, name='removal_duplicate_domains')
  8. '''新发现的域名'''
  9. MGO_DOMAIN = mongo_table(db=MGO_DATABASE, name='new_domains')
  10. '''搜索【组织|单位】'''
  11. MGO_SEED_ORGS = mongo_table(db=MGO_DATABASE, name='seed_organizations')
  12. '''搜索关键词'''
  13. MGO_SEED_KEYWORDS = mongo_table(db=MGO_DATABASE, name='seed_keywords')
  14. '''种子urls'''
  15. MGO_SEED_URLS = mongo_table(db=MGO_DATABASE, name='seed_urls')
  16. '''redis'''
  17. REDIS = redis_client()
  18. REDIS_KEY = 'retrieve_urls'
  19. '''关键词'''
  20. REQUIREMENT_PHRASE = [
  21. '竞谈', '发包', '比价', '开标', '邀标', '采购', '招标', '中标', '废标', '成交', '单一', '询价',
  22. '项目结果', '邀请', '磋商', '流标', '谈判', '竞争', '遴选', '比选', '招募', '评标', '资格预审',
  23. '议价', '中选', '答疑', '合同', '竞价', '变更', '更正', '预告', '集采', '抽取', '抽签',
  24. '中止公告', '终止公告', '竞卖', '竞买', '论证', '拟建', '审批', '环评'
  25. ]
  26. '''需要剔除或者过滤的关键字'''
  27. SENSITIVE_WORDS = [
  28. '通知', '邮箱', '登录'
  29. ]
  30. '''特殊编码'''
  31. SPECIAL_ENCODINGS = [
  32. 'Windows-1254'
  33. ]