import re from pathlib import Path def show(name: str, path: str) -> None: t = Path(path).read_text(encoding="utf-8", errors="replace") print("====", name, "len", len(t)) urls = sorted(set(re.findall(r"https?://[^\"'\s>]+", t))) print("absolute urls", len(urls)) for u in urls: print(" ", u[:200]) hrefs = sorted(set(re.findall(r'href="([^"]+)"', t))) srcs = sorted(set(re.findall(r'src="([^"]+)"', t))) print("hrefs:") for h in hrefs: if "css" in h or "style" in h or "font" in h or "http" in h or "wo." in h: print(" ", h[:180]) print("srcs:") for h in srcs: if any(x in h for x in ("js", "http", "png", "jpg", "wo.")): print(" ", h[:180]) for n in ["fonts.googleapis", "gstatic", "83.147", "portal.otantik", "plans"]: print(" contains", n, n in t) show("HUB", "_tmp_hub_login.html") show("NKO", "_tmp_nko_login_now.html")