repairName.py 1010 B

123456789101112131415161718192021222324252627282930313233343536
  1. '''
  2. Created on 2019年12月18日
  3. @author: User
  4. '''
  5. import pymongo
  6. import re
  7. import pandas as pd
  8. def getDatafromMongo():
  9. client = pymongo.MongoClient("mongodb://bxkc_read:BidiReadOnly2017@47.98.60.3:17017/bxkc")
  10. db = client["bxkc"]
  11. collection_enterprise = db['zhaobiao_extraction']
  12. cursor = collection_enterprise.find({"page_time":{"$gte":'2019-03-01'}}).limit(5000)
  13. list_docid = []
  14. list_projectName = []
  15. list_title = []
  16. print("done")
  17. _index = 0
  18. for row in cursor:
  19. print(row)
  20. _index += 1
  21. list_docid.append(row["docId"])
  22. list_projectName.append(row["project_name"])
  23. list_title.append(row["document_title"])
  24. #print(_index,row["project_name"])
  25. print("done")
  26. df = pd.DataFrame({"list_docid":list_docid,"list_projectName":list_projectName,"list_title":list_title})
  27. df.to_excel("repairName.xls",columns=["list_docid","list_projectName","list_title"])
  28. if __name__=="__main__":
  29. getDatafromMongo()