diff options
Diffstat (limited to 'day1/task4/task4.py')
| -rw-r--r-- | day1/task4/task4.py | 60 |
1 files changed, 60 insertions, 0 deletions
diff --git a/day1/task4/task4.py b/day1/task4/task4.py new file mode 100644 index 0000000..2583cb4 --- /dev/null +++ b/day1/task4/task4.py @@ -0,0 +1,60 @@ +import string +import re +import os + + +html_path = os.path.join(os.getcwd(), 'task4_data.html') + +folder_path = '' + +# Не знаю зачем, но в задании так написано +try: + os.mkdir('../htmls') + folder_path = '../htmls' +except PermissionError: + print('Директория на уровень выше не доступна для записи.') + print('Создаем папку в текущей директории...') + try: + os.mkdir('htmls') + folder_path = 'htmls' + except PermissionError: + print('Текущая директория не доступна для записи. Завершаем работу...') + quit() + +try: + html_file = open(html_path) + os.chdir(folder_path) + ascii_file = open('ascii_cleared.html', 'w', encoding='utf-8') + body_file = open('body_cleared.html', 'w', encoding='utf-8') + +except FileNotFoundError as e: + exception_path = str(e).strip('[Errno 2] No such file or directory:')[1:-1] + print(f'"{exception_path}" не существует') + +except PermissionError as e: + exception_path = str(e).strip('[Errno 13] Permission denied:')[1:-1] + print(f'Файл "{exception_path}" не доступен для чтения/записи') + +except IsADirectoryError as e: + exception_path = str(e).strip('[Errno 21] Is a directory:')[1:-1] + print(f'"{exception_path}" является директорией') + +else: + data = html_file.read() + + cleared_data = ''.join(filter(lambda char: char in string.printable, data)) + ascii_file.write(cleared_data) + + ################### + + body_cleared = re.sub( + r'<BODY>(.|\n)*?</BODY>', + lambda match: match.string[match.start(): match.end()].replace('\n', ''), + data + ) + + body_file.write(body_cleared) + + html_file.close() + ascii_file.close() + body_file.close() |