DataMining/export/cleanAgency.py

#coding:utf8

import pandas as pd
import re

def clean():
    filename = "20230227都大于100.xlsx"
    df = pd.read_excel(filename)

    _count = 0
    df_data = {"name":[],
               "zbn":[],
               "dln":[],
               "sn":[]}
    for name,zbn,dln,sn in zip(df["nicknames"],df["zhao_biao_number"],df["dai_li_number"],df["same_number"]):
        if dln>10000 and re.search("招标|咨询",name) is not None:
            _count += 1
            print(_count,name,zbn,dln,sn)
            df_data["name"].append(name)
            df_data["zbn"].append(zbn)
            df_data["dln"].append(dln)
            df_data["sn"].append(sn)
    df = pd.DataFrame(df_data)
    df.to_excel("daili_check.xlsx")





if __name__ == '__main__':
    clean()