Fix openscap report, externalize all inline scripts and stylesheets

fix-openscap-report.py

#!/usr/bin/env python3

# * finds all inline style and script elements in an html file
#   exports the code into a file (ie. filename.css, filename.js)
#   removes the original tag from the html
# * replaces inline styles and scripts with tags in the header
#   sourcing the new external code


# disable tracebacks
import sys
sys.tracebacklimit = 0

import os
from lxml import html

if len(sys.argv) < 3:
    print("two arguments required, path to html file and output directory", file=sys.stderr)
    raise SystemExit(1)

try:
    os.mkdir(sys.argv[2])
except FileExistsError:
    pass

output_fn = os.path.join(sys.argv[2], os.path.basename(sys.argv[1]))
output_base_fn, _ = os.path.splitext(output_fn)

fp = { 'style': None, 'script': None }

tree = html.parse(sys.argv[1])
root = tree.getroot()

for el in root.xpath('.//style | .//script'):
    if not fp[el.tag]:
        ext = 'css' if (el.tag == 'style') else 'js'
        fp[el.tag] = open(f'{output_base_fn}.{ext}', 'w')
    fp[el.tag].write(el.text_content())
    fp[el.tag].write("\n")
    el.drop_tree()

head = root.find('.//head')

if fp['script']:
    script_tag = html.Element('script')
    script_tag.set('type', 'text/javascript')
    script_tag.set('src', os.path.basename(fp['script'].name))
    head.append(script_tag)

if fp['style']:
    link_tag = html.Element('link')
    link_tag.set('rel', 'stylesheet')
    link_tag.set('href', os.path.basename(fp['style'].name))
    link_tag.set('type', 'text/css')
    link_tag.set('media', 'screen')
    head.append(link_tag)

tree.write(output_fn, encoding='utf-8', method='html')