From ddd5bfe49f1b9c3dfb3b1b448bece9fc636cb88f Mon Sep 17 00:00:00 2001 From: Andrew Date: Mon, 4 Mar 2019 20:03:49 +0400 Subject: Day 1 Task 4 --- day1/task4/task4.py | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) create mode 100644 day1/task4/task4.py (limited to 'day1/task4/task4.py') diff --git a/day1/task4/task4.py b/day1/task4/task4.py new file mode 100644 index 0000000..9ed9f2d --- /dev/null +++ b/day1/task4/task4.py @@ -0,0 +1,25 @@ +import string +import re +import os + +with open('task4_data.html') as f: + data = f.read() + +# Не знаю зачем, но в задании так написано +os.chdir('..') +os.mkdir('htmls') +os.chdir('htmls') + +cleared_data = ''.join(filter(lambda char: char in string.printable, data)) +with open('ascii_cleared.html', 'w', encoding='utf-8') as f: + f.write(cleared_data) + +body_cleared = re.sub( + r'(.|\n)*?', + lambda match: match.string[match.start(): match.end()].replace('\n', ''), + data +) + +with open('body_cleared.html', 'w', encoding='utf-8') as f: + f.write(body_cleared) + -- cgit v1.2.3