challenging-america-word-ga.../utils.py

48 lines
1.1 KiB
Python
Raw Normal View History

2022-04-11 00:20:54 +02:00
import pandas as pd
import regex as re
from csv import QUOTE_NONE
ENCODING = "utf-8"
REP = re.compile(r"[{}\[\]\&%^$*#\(\)@\t\n0123456789]+")
REM = re.compile(r"'s|[\-]\\n|\-\\n|\p{P}")
def read_csv(fname):
return pd.read_csv(
fname,
sep="\t",
on_bad_lines='skip',
header=None,
quoting=QUOTE_NONE,
encoding=ENCODING
)
def clean_text(text):
res = str(text).lower().strip()
res = res.replace("", "'")
res = REM.sub("", res)
res = REP.sub(" ", res)
2022-05-08 23:31:17 +02:00
res = res.replace("'t", " not")
2022-04-11 00:20:54 +02:00
res = res.replace("'s", " is")
res = res.replace("'ll", " will")
res = res.replace("won't", "will not")
res = res.replace("isn't", "is not")
res = res.replace("aren't", "are not")
res = res.replace("'ve'", "have")
return res.replace("'m", " am")
2022-05-08 23:31:17 +02:00
def get_words_from_line(line, specials = True):
line = line.rstrip()
if specials:
yield '<s>'
for m in re.finditer(r'[\p{L}0-9\*]+|\p{P}+', line):
yield m.group(0).lower()
if specials:
yield '</s>'
def get_word_lines_from_data(d):
for line in d:
yield get_words_from_line(line)
2022-04-11 00:20:54 +02:00