added extract code to handle when the cleaned data is available

[happygirlzt]

I made some changes on data_process.py. If we directly download the cleaned_data provided by the authors, we only need to split the data. The added part is for handling this situation.