123456789101112131415161718192021222324252627282930313233343536 |
- '''
- Created on 2019年12月18日
- @author: User
- '''
- import pymongo
- import re
- import pandas as pd
- def getDatafromMongo():
- client = pymongo.MongoClient("mongodb://bxkc_read:BidiReadOnly2017@47.98.60.3:17017/bxkc")
- db = client["bxkc"]
- collection_enterprise = db['zhaobiao_extraction']
- cursor = collection_enterprise.find({"page_time":{"$gte":'2019-03-01'}}).limit(5000)
-
- list_docid = []
- list_projectName = []
- list_title = []
-
-
- print("done")
- _index = 0
- for row in cursor:
- print(row)
- _index += 1
- list_docid.append(row["docId"])
- list_projectName.append(row["project_name"])
- list_title.append(row["document_title"])
- #print(_index,row["project_name"])
- print("done")
- df = pd.DataFrame({"list_docid":list_docid,"list_projectName":list_projectName,"list_title":list_title})
- df.to_excel("repairName.xls",columns=["list_docid","list_projectName","list_title"])
- if __name__=="__main__":
- getDatafromMongo()
|