from common.databases import mongo_table, redis_client '''Mongo''' MGO_DATABASE = 'shujuziyuan' '''判重库''' MGO_REPETITION = mongo_table(db=MGO_DATABASE, name='repetition_url') '''去重域名表''' MGO_REMOVAL_DOMAIN = mongo_table(db=MGO_DATABASE, name='removal_duplicate_domains') '''新发现的域名''' MGO_DOMAIN = mongo_table(db=MGO_DATABASE, name='new_domains') '''搜索【组织|单位】''' MGO_SEED_ORGS = mongo_table(db=MGO_DATABASE, name='seed_organizations') '''搜索关键词''' MGO_SEED_KEYWORDS = mongo_table(db=MGO_DATABASE, name='seed_keywords') '''种子urls''' MGO_SEED_URLS = mongo_table(db=MGO_DATABASE, name='seed_urls') '''redis''' REDIS = redis_client() REDIS_KEY = 'retrieve_urls' '''关键词''' REQUIREMENT_PHRASE = [ '竞谈', '发包', '比价', '开标', '邀标', '采购', '招标', '中标', '废标', '成交', '单一', '询价', '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', '议价', '中选', '答疑', '合同', '竞价', '变更', '更正', '预告', '集采', '抽取', '抽签', '中止公告', '终止公告', '竞卖', '竞买', '论证', '拟建', '审批', '环评' ] '''需要剔除或者过滤的关键字''' SENSITIVE_WORDS = [ '通知', '邮箱', '登录' ] '''特殊编码''' SPECIAL_ENCODINGS = [ 'Windows-1254' ]