settings.py 1.8 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849
  1. from common.databases import mongo_table, redis_client
  2. '''Mongo'''
  3. MGO_DATABASE = 'shujuziyuan'
  4. '''去重表'''
  5. MGO_REMOVAL_DUPLICATE = mongo_table(db=MGO_DATABASE, name='removal_duplicate')
  6. '''数据挖掘结果表'''
  7. MGO_DOMAIN = mongo_table(db=MGO_DATABASE, name='new_domains')
  8. '''查询结果表'''
  9. MGO_QUERY = mongo_table(db=MGO_DATABASE, name='data_query')
  10. ''''数据采集记录表'''
  11. MGO_RECORDS = mongo_table(db=MGO_DATABASE, name='excavate_records')
  12. '''组织|单位表'''
  13. MGO_ORGS = mongo_table(db=MGO_DATABASE, name='retrieve_orgs')
  14. '''搜索词'''
  15. MGO_KEYWORDS = mongo_table(db=MGO_DATABASE, name='retrieve_keywords')
  16. '''种子列表'''
  17. MGO_URLS = mongo_table(db=MGO_DATABASE, name='retrieve_urls')
  18. '''竞品列表'''
  19. MGO_COMPETING_GOODS = mongo_table(db=MGO_DATABASE, name='retrieve_competing_goods')
  20. '''网站栏目表'''
  21. MGO_WEBSITE = mongo_table(db=MGO_DATABASE, name='website')
  22. '''luaconfig'''
  23. MGO_LUA_SPIDERS = mongo_table(db='editor', name='luaconfig')
  24. '''redis'''
  25. REDIS = redis_client()
  26. '''redis键名前缀'''
  27. REDIS_QUERY_KEYWORD = 'query_keyword'
  28. REDIS_QUERY_ORGS = 'query_org'
  29. REDIS_EXCAVATE = 'data_excavate'
  30. '''过滤词'''
  31. FILTER_WORDS = [
  32. '竞谈', '发包', '比价', '开标', '邀标', '采购', '招标', '中标', '废标', '成交', '单一', '询价',
  33. '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',
  34. '议价', '中选', '答疑', '合同', '竞价', '变更', '更正', '预告', '集采', '抽取', '抽签',
  35. '中止公告', '终止公告', '竞卖', '竞买', '论证', '拟建', '审批', '环评'
  36. ]
  37. '''搜索引擎屏蔽的网址'''
  38. ENGINE_FEATURE_RETRIEVES = [
  39. 'microsoft.com',
  40. 'cn.bing.com',
  41. 'beian.miit.gov.cn',
  42. 'beian.gov.cn/portal/registerSystemInfo',
  43. 'baike.baidu.com'
  44. ]