nikolajbaer

import re
from operator import itemgetter
from collections import Counter

def clean_data(txt):
    txt = txt.lower()
    txt = txt.replace('mr.','mr').replace('mrs.','mrs').replace('ms.','ms')
    sentences = re.findall(r"[^!\.\?]+[!\.\?]",txt.lower())
    sentences = [s.replace('\n',' ').replace('\r',' ') for s in sentences] 
    sentences = [re.sub(r'[^a-z ]','',x) for x in sentences]

nikolajbaer

bin
.Python
lib
include
pip-selfcheck.json