|
|
@@ -7,8 +7,10 @@ MGO_DATABASE = 'shujuziyuan'
|
|
|
MGO_REPETITION = mongo_table(db=MGO_DATABASE, name='repetition_url')
|
|
|
'''结果'''
|
|
|
MGO_RECORDS = mongo_table(db=MGO_DATABASE, name='records')
|
|
|
-'''搜索'''
|
|
|
-MGO_VISIT = mongo_table(db=MGO_DATABASE, name='visit')
|
|
|
+'''搜索|组织|单位'''
|
|
|
+MGO_VISIT_ORGANIZATION = mongo_table(db=MGO_DATABASE, name='visit_organization')
|
|
|
+'''搜索|关键词'''
|
|
|
+MGO_VISIT_KEYWORDS = mongo_table(db=MGO_DATABASE, name='visit_keywords')
|
|
|
'''redis'''
|
|
|
REDIS = redis_client()
|
|
|
REDIS_KEY = 'retrieve_urls'
|
|
|
@@ -19,7 +21,11 @@ REQUIREMENT_PHRASE = [
|
|
|
'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'
|
|
|
]
|
|
|
-'''没用信息|广告|垃圾消息|敏感词'''
|
|
|
+'''无用词组|广告|涉黄信息|涉黑信息|垃圾'''
|
|
|
SENSITIVE_WORDS = [
|
|
|
- '通知'
|
|
|
+ '通知', '邮箱', '登录'
|
|
|
+]
|
|
|
+'''特殊编码'''
|
|
|
+SPECIAL_ENCODINGS = [
|
|
|
+ 'Windows-1254'
|
|
|
]
|
