#!/usr/bin/env python3
"""Cek sitemap TSC setelah perubahan: apakah slug lama sudah hilang dan halaman trash sudah tidak ada."""
import re, urllib.request

UA = {"User-Agent": "Mozilla/5.0"}
DOM = "theskincodeclinic.com"
idx = urllib.request.urlopen(urllib.request.Request(f"https://www.{DOM}/sitemap_index.xml", headers=UA),
                             timeout=60).read().decode("utf-8", "replace")
subs = re.findall(r"<loc>([^<]+)</loc>", idx)
print("sub-sitemap:", len(subs))
urls = []
for s in subs:
    try:
        t = urllib.request.urlopen(urllib.request.Request(s, headers=UA), timeout=60).read().decode("utf-8", "replace")
        u = re.findall(r"<loc>([^<]+)</loc>", t)
        urls += u
        print(f"   {len(u):>4} entri | {s.split('/')[-1]}")
    except Exception as e:
        print(f"   gagal {s[-30:]}: {str(e)[:40]}")

print(f"\ntotal entri: {len(urls)}")
lama = [u for u in urls if re.search(r"klinik[\s-]*dni", u, re.I)]
print(f"masih memuat nama lama: {len(lama)}")
for u in lama[:10]:
    print("   ⚠", u)
trash = [u for u in urls if "tentang-kami" in u]
print(f"halaman tentang-kami di sitemap: {len(trash)}")
baru = [u for u in urls if "the-skin-code-clinic" in u]
print(f"entri dengan nama baru: {len(baru)}")