#!/usr/bin/env python3
"""Pastikan tidak ada lagi tautan tak terlihat di entri yang tadi dibersihkan."""
import base64, json, re, time, urllib.request

WS = "/home/mantapman/.hermes/profiles/atun/workspace"
S = "https://studiosoft.id"
A = base64.b64encode(b"bambangbot:q26r RePo BVy6 pwU2 F3z7 ouyl").decode()
H = {"Authorization": f"Basic {A}", "User-Agent": "Mozilla/5.0"}
KOSONG = re.compile(r'<a\b[^>]*>\s*</a>', re.S)

data = json.load(open(f"{WS}/drafts/ss_tautan_mencurigakan.json"))
entri = sorted({(j, p) for j, p, s, u in data["kosong"]})
print(f"memeriksa ulang {len(entri)} entri")

sisa, gagal = 0, 0
for jenis, pid in entri:
    try:
        isi = json.loads(urllib.request.urlopen(urllib.request.Request(
            f"{S}/wp-json/wp/v2/{jenis}/{pid}?context=edit&_fields=content", headers=H), timeout=150).read())
        isi = isi["content"]["raw"]
    except Exception:
        gagal += 1
        continue
    n = len(KOSONG.findall(isi))
    if n:
        sisa += n
        print(f"   MASIH ADA: #{pid} [{jenis}] {n}")
    time.sleep(0.3)
print(f"\ntautan tak terlihat tersisa: {sisa} · entri gagal diperiksa: {gagal}")

# ringkasan domain yang tadi dititipkan
from collections import Counter
dom = Counter(re.sub(r"^https?://(www\.)?", "", u).split("/")[0] for _, _, _, u in data["kosong"])
print("\ndomain yang menitipkan tautan tak terlihat:")
for d, n in dom.most_common():
    print(f"   {n:>3}× {d}")
