In [1]:

    

import pandas as pd
import re


    

In [84]:

    

mon = ['jan','feb','mar','apr','may','jun','jul','aug','sep','oct','nov','dec']


    

In [97]:

    

for m in mon:
    filename = 'raw/'+m
    month = []
    with open(filename) as fi:
        for i in fi.readlines():
            month.append(i)

    month = map(lambda y: (filter(lambda x: x!='',re.split('\n|\(|\)',y))[0],
                             int( filter(lambda x: x!='',re.split('\n|\(|\)',y))[1].split(', ')[1])),
                month)
    month = filter(lambda x: x[0][0][0]=='u',month)
    month = map(lambda x: (x[0].split(', '),x[1]),month)
    df_j=pd.DataFrame(columns=['line','day_of_week','hour','day','month','count'],
                      index=range(len(month)))
    for i in range(len(month)):
        df_j.iloc[i,0]= month[i][0][0]
        df_j.iloc[i,1]= month[i][0][1]
        df_j.iloc[i,2]= month[i][0][2]
        df_j.iloc[i,3]= month[i][0][3]
        df_j.iloc[i,4]= month[i][0][4]
        df_j.iloc[i,5]= month[i][1]
    for i in range(1,6):
        df_j.iloc[:,i]= pd.to_numeric(df_j.iloc[:,i])
    df_j.line = map(lambda x: x.split("'")[1],df_j.line)
    output= 'csv/'+m+'.csv'
    df_j.to_csv(output)