Web/generate_errors.py

from pprint import pp
from pathlib import Path
from urllib import request
import re

urlroot = 'https://www.w3.org/Protocols/rfc2616'

src = Path('src')
root = src.joinpath('error')
with root.joinpath('root.html').open('r') as root_file:
    root_html = root_file.read()
with root.joinpath('root.css').open('r') as root_file:
    root_css = root_file.read()

error = Path('dist_error')
error.mkdir(exist_ok=True)

for x in error.rglob('*'):
    if x.is_file() or x.is_symlink(): x.unlink()

req = request.Request(
    f'{urlroot}/rfc2616-sec10.html',
    method='GET')

with request.urlopen(req) as res:
    raw = res.read().decode('utf-8')

codes = (re.finditer(r"(?s)<h3><a id='(sec10\.[0-9]\.[0-9]+)'>"
    r'10\.[0-9]\.[0-9]+</a>'
    r'\s*([0-9]+)\s([a-zA-Z\s]+)</h3>\s*<p>(.*?)</p>', raw))

nginx_head = ''
nginx_body = ''

for item in codes:
    sec, code, title, desc = item.groups()
    if int(code) < 400 or int(code) >= 600: continue

    desc = re.sub(r"(<a.*?href=')(.*)('>)",
        lambda x:f'{x[1]}{urlroot}/{x[2]}{x[3]}', desc)

    with error.joinpath(f'{code}.html').open('w') as html:
        content = root_html.format(
            code=code, title=title, desc=desc, style=root_css, sec=sec)
        content = re.sub(r'\s+', ' ', content)
        content = re.sub(r'>\s<', '><', content)
        html.write(content)
    nginx_head += f'error_page {code} @error{code};\n'
    nginx_body += f'location @error{code} {{\n'\
        f'root {error.resolve()};\n'\
        f'try_files /{code}.html /{code}.html;\n'\
        'internal;}\n'

with error.joinpath(f'nginx.conf').open('w') as nginx_file:
    nginx_file.write(f'{nginx_head}\n{nginx_body}\n\n'\
        'proxy_intercept_errors on;')