12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576 |
- from bson import ObjectId
- from pymongo import MongoClient
- from BasicMethods.area_quality import au
- db = MongoClient('192.168.3.206', 27080, unicode_decode_error_handler="ignore").data_quality
- coll_user = db["bidding_20230707"]
- # coll_user_inserd = db["user_insder"]
- def pankong(key, item):
- if key in item and item[key]:
- value = item[key]
- else:
- value = ""
- return value
- #正确数量
- correct_count=0
- count = 0
- #find里面 单条数据用法:{"_id":ObjectId("64a8bb45990ffa1883accd78")}
- for item in coll_user.find().batch_size(1000):
- count += 1
- if count % 1000==0:
- print(count)
- title = pankong("title", item)
- #根据title抽取出来的省份
- expect_area1 = au.export_area(title)
- # print(expect_area1)
- # 根据buyer抽取出来的省份
- buyer = pankong("buyer", item)
- expect_area = au.export_area(buyer)
- # 根据buyeraddr抽取出来的省份
- buyeraddr = pankong("buyeraddr", item)
- expect_area4 = au.export_area(buyeraddr)
- # print(expect_area)
- #根据s_winner抽取出来的省份
- s_winner=pankong("s_winner",item)
- expect_area2 = au.export_area(s_winner)
- # 根据winneradder抽取出来的省份
- winneraddr = pankong("winneraddr", item)
- expect_area3 = au.export_area(winneraddr)
- # 根据agency抽取出来的省份
- agency = pankong("agency", item)
- expect_area5 = au.export_area(agency)
- # 根据agencyaddr抽取出来的省份
- agencyaddr = pankong("agencyaddr", item)
- expect_area6 = au.export_area(agencyaddr)
- #数据组抽取出来的省份,需要验证的字段
- tmp_area = pankong("area", item)
- #抽取出来的地区,通过函数在转换一遍
- actual_area = au.export_area(tmp_area)
- flag_buyer = 0
- flag_title = 0
- flag_s_winner = 0
- flag_winneraddr = 0
- flag_buyeraddr = 0
- flag_agency = 0
- flag_agencyaddr = 0
- if expect_area and actual_area and expect_area == actual_area:
- flag_buyer = 1
- if expect_area1 and actual_area and expect_area1 == actual_area:
- flag_title = 1
- if expect_area2 and actual_area and expect_area2 == actual_area:
- flag_s_winner = 1
- if expect_area3 and actual_area and expect_area3 == actual_area:
- flag_winneraddr = 1
- if expect_area4 and actual_area and expect_area4 == actual_area:
- flag_buyeraddr = 1
- if expect_area5 and actual_area and expect_area5 == actual_area:
- flag_agency = 1
- if expect_area6 and actual_area and expect_area6 == actual_area:
- flag_agencyaddr = 1
- if flag_buyer == 1 or flag_title == 1 or flag_s_winner == 1 or flag_winneraddr == 1 or flag_buyeraddr == 1 or flag_agency ==1 or flag_agencyaddr == 1:
- correct_count+=1
- coll_user.update_one({"_id":item["_id"]},{"$set":{"flag_buyer":flag_buyer,"flag_title":flag_title,"flag_s_winner":flag_s_winner,"flag_winneraddr":flag_winneraddr,"flag_buyeraddr":flag_buyeraddr,"flag_agency":flag_agency,"flag_agencyaddr":flag_agencyaddr}})
- # coll_user.update({"_id":item["_id"]},{"$set":{"flag":flag_title}})
- print(correct_count)
|