Useful to remove all JS scripts from a web page

clean-js

#!/usr/bin/python3

import datetime
import sys
from hashlib import md5

import requests
import sh
from bs4 import BeautifulSoup

DEFAULT_WEBBROWSER = 'firefox'
DEFAULT_DST = '/tmp/{}.html'.format(
    md5(str(datetime.datetime.now()).encode()).hexdigest()
)

def dispatch_help():
    print(
"""
======= clean-js =======
remove-js and generate .html
usage:
    `clean-js <source-url> [options]`

options:
    -h or --help: dispatch usage
    -o [web-browser] open in a web-browser (default is firefox)
"""
    )

def remove_scripts(string):
    soup = BeautifulSoup(string.lower(), 'html.parser')
    to_extract = soup.findAll('script')

    for item in to_extract:
        item.extract()
    return soup.prettify()


if __name__ == '__main__':
    if (len(sys.argv) < 2 or '-h' in sys.argv or '--help' in sys.argv):
        dispatch_help()
        sys.exit(0)

    sys.argv.pop(0) # pop the program name

    url = sys.argv.pop(0)
    r = requests.get(url)

    if r.status_code not in (200, 201, 202, 203):
        dispatch_help()
        sys.exit(-1)

    content = ''.join(r.text)
    content = remove_scripts(content)

    with open(DEFAULT_DST, 'w+') as dst:
        dst.write(content)


    if '-o' in sys.argv:
        idx = sys.argv.index('-o')
        sys.argv.remove('-o')

        try:
            getattr(sh, sys.argv[idx])(DEFAULT_DST)
        except:
            getattr(sh, DEFAULT_WEBBROWSER)(DEFAULT_DST)