Tally up code counts per day

icon.py

import os
import pandas

def main():
    
    data_directory = './replay-correction'
    output_fname = './icon_code_count.csv'
    
    dates = []
    codes = []
    
    for (base, dirs, files) in os.walk(data_directory):
        for filename in files:
            with open(os.path.join(base,filename),'rb') as f:
                
                for line in f:
                    date_str = line[44:56]
                    code_str = line[56:60]
                    
                    dates.append(date_str)
                    codes.append(code_str)
                    
    dates = pandas.Series(dates)
    codes = pandas.Series(codes)
    
    data = pandas.DataFrame({'date_str':dates,'code':codes})
    data['date_dt'] = pandas.to_datetime(data['date_str'],format='%Y%m%d%H%M')
    
    daily_counts = data.groupby([data['date_dt'].dt.to_period("d"),data["code"]]).count()
    daily_counts.drop(daily_counts.columns[1],1,inplace=True)
    
    daily_counts.columns = ['counts']
    
    daily_counts.to_csv(output_fname)
    
    return (dates, codes, data, daily_counts)


if __name__=="__main__":
    main()