Youtube Transcript Cleaner.py

import re
import sys
# Read and Write in python
regex_pattern1 = r'<p.*>'   # Do not delete 'r'.
regex_pattern2 = r'</p>'
regex_pattern3 = r'\n+'
regex_pattern4 = r'<body>|</body>|<timedtext.*>|</timedtext>'
regex_pattern5 = r'^This XML.*.'
regex_pattern6 = r'\n'
regex_pattern7 = r'\.'

with open('text.txt', 'r') as rf:
    file_contents = rf.read();
    # print(file_contents)
    match = re.sub(regex_pattern1, "", file_contents)
    match = re.sub(regex_pattern2, "", match)
    match = re.sub(regex_pattern4, "", match)
    match = re.sub(regex_pattern5, "", match)
    match = re.sub(regex_pattern3, "\n\n", match)
    # match = re.sub(regex_pattern6, "", match)
    # match = re.sub(regex_pattern7, ".\n", match)

    print(match, end='')

    with open('text2.txt', 'w') as wf:
        wf.write(match)