topic_spider/find_source/settings.py

from common.databases import mongo_table, redis_client


'''Mongo'''
MGO_DATABASE = 'shujuziyuan'
'''垃圾表'''
MGO_REMOVAL_DUPLICATE = mongo_table(db=MGO_DATABASE, name='removal_duplicate')
'''新发现的域名'''
MGO_DOMAIN = mongo_table(db=MGO_DATABASE, name='new_domains')
'''检索单位组织|关键词'''
MGO_SEARCH = mongo_table(db=MGO_DATABASE, name='retrieve_search')
'''【组织|单位】'''
MGO_ORGS = mongo_table(db=MGO_DATABASE, name='retrieve_orgs')
'''关键词'''
MGO_KEYWORDS = mongo_table(db=MGO_DATABASE, name='retrieve_keywords')
'''种子url'''
MGO_URLS = mongo_table(db=MGO_DATABASE, name='retrieve_urls')
'''竞品url'''
MGO_COMPETING_GOODS = mongo_table(db=MGO_DATABASE, name='retrieve_competing_goods')
'''luaconfig'''
MGO_LUA_SPIDERS = mongo_table(db='editor', name='luaconfig')
'''redis'''
REDIS = redis_client()
REDIS_KEY = 'retrieves'
'''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'
]
'''剔除或者过滤的内容'''
SENSITIVE_WORDS = [
    '通知', '邮箱', '登录', '注册', '亿元', '地址', '招聘', '试驾', '信用卡',
]
'''搜索引擎过滤特征'''
ENGINE_FEATURE_RETRIEVES = [
    'microsoft.com',
    'cn.bing.com',
    'beian.miit.gov.cn',
    'beian.gov.cn/portal/registerSystemInfo',
    'baike.baidu.com'
]
'''特殊编码'''
SPECIAL_ENCODINGS = [
    'Windows-1254'
]