python_pandas_csv_cleaning_andramarkov.py

import pandas as pd 
import datetime

#nama file
source = 'marketingresign_is_shipper_false.csv'

try:
	mulai = datetime.datetime.now()
	#baca csv, csv saya separatornya comma, ubah menjadi sep='\t' jika separator file anda tab
	data = pd.read_csv(source, sep=',') 
	#list marketing yg sudah resign
	mktResign = ['RUDY LAI','HENDRA','KEVIN', 'EVIANI', 'MARDI', 'KUSWANDY', 'LIEO HANDINO', 'IBNU', 'FAHRIOZY', 'DEVID', 'ANDES', 'JIMMY' , 'TYO', 'FANNY', 'WINCARNIS', 'YAPTO', 'IBNU', 'ROBERTO SIMON' ]

	#ubah header kolom "Nama Marketing" agar tdk ada spasi 
	data.rename(columns={'Nama Marketing':'Nama_Marketing'}, inplace=True)

	#opsional: drop kolom yang tdk perlu
	kolom = [5, 7, 8, 9, 11, 13, 14]
	data.drop(data.columns[kolom],axis=1,inplace=True)

	try:
		#pisahkan nama file dan ekstensi
		nmfile, extfile = source.split('.')
		
		#kita filter datanya pakai pd.Series.isin
		#IN -> data = data[data.Nama_Marketing.isin(mktResign)]
		#NOT IN -> data = data[~data.Nama_Marketing.isin(mktResign)]
		data = data[data.Nama_Marketing.isin(mktResign)]
		
		#simpan ke file csv baru
		tgl = datetime.datetime.now().strftime('%Y_%m_%d')
		data.to_csv(nmfile + '_cleaned_at_' + tgl + '.csv', index=False, header=True)
		
		
	except Exception as e:
		print ('Penyimpanan data gagal')
		print ('Pesan: ' + str(e) )
	
	else:		
		udah = datetime.datetime.now()
		howlong = udah - mulai
		print ('Penyimpanan data berhasil')
		print ('Elapsed Time: ' + str(howlong))
	

except Exception as e:	
	print ('Terjadi kesalahan dalam pemrosesan file')
	print ('Pesan: ' + str(e) )