total_df.to_csv(path2+"total-"+data_fname,index=False,sep='\t') print('complete the %s geocoding' %data_fname) print('--------------------------------------') # temp_procedure=math.floor(i/(int(len(data_geoInfo)*percent))*100) # if temp_procedure%5==0: # print("finished "+ str(temp_procedure)+"% total data") ''' RUN THE GEOCODING ''' ## 获取地理位置信息 Baidu=geoBaidu.geoBaidu(ak='DPlowD7PIEfaVtpxLKGkXg8yDCCBanVO') year=input('please input the year for geocoding') cat_list=datafile_cate(path) file_list=cat_list[year] #(prov_df,city_df,county_df)=df_dist_get(year,path_dist) for data_fname in file_list: print('strating the %s gecoding' %data_fname) data_raw=pd.read_csv(path+data_fname,sep='\t',encoding='utf-8') # data_raw.reset_index(inplace=True) data_raw['i']=data_raw.index data_raw['new_ID']=data_raw['i'].apply(lambda x : new_ID(x,data_fname)) data_raw.drop(labels=['i'],axis=1,inplace=True) get_gis_info(data_fname,data_raw,path2)
total_df.to_csv(path2 + "total-" + data_fname, index=False) print('complete the %s geocoding' % data_fname) print('--------------------------------------') # temp_procedure=math.floor(i/(int(len(data_geoInfo)*percent))*100) # if temp_procedure%5==0: # print("finished "+ str(temp_procedure)+"% total data") if __name__ == '__main__': ''' RUN THE GEOCODING ''' ## 获取地理位置信息 Baidu = geoBaidu.geoBaidu(ak='DPlowD7PIEfaVtpxLKGkXg8yDCCBanVO') year = input('please input the year for geocoding') cat_list = datafile_cate(path) file_list = cat_list[year] (prov_df, city_df, county_df) = df_dist_get(year, path_dist) for data_fname in file_list: print('strating the %s gecoding' % data_fname) data_raw = pd.read_csv(path + data_fname, sep='\t', encoding='GBK') data_raw['i'] = data_raw.index data_raw['new_ID'] = data_raw['i'].apply( lambda x: new_ID(x, data_fname)) data_geoInfo = geo_data(data_raw[old_col].copy(), prov_df, city_df, county_df)