delete_html_files_and_empty_folders.py

#!/usr/bin/env python
import pathlib
import re

directories = []
for file in pathlib.Path("./output").glob("**/*"):
    if file.is_dir():
        directories.append(file)
    if not file.is_file():
        continue
    with file.open("rb") as fp:
        content = fp.read(100).decode(errors="ignore")
    if re.search(r"\s*<(\!doctype|html|body)", content, re.IGNORECASE):
        file.unlink()
        print("html file deteled:", file)


d: pathlib.Path
for d in reversed(directories):
    try:
        d.rmdir()
        print("empty directory deleted:", d)
    except OSError:
        pass