#!/usr/bin/env python3
"""Verifikasi akhir artikel kelima TSLaw (#3653): cek mentah di API + coba URL live."""
import base64, json, re, urllib.request, urllib.error

DOM, USER, PASS = "tslaw.id", "bambangbot", "EFKm EYGx VQCP Y9go ncPr uM2Y"
AUTH = base64.b64encode(f"{USER}:{PASS}".encode()).decode()
UA = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) Chrome/124 Safari/537.36"}
PID = 3653

u = f"https://{DOM}/wp-json/wp/v2/posts/{PID}?context=edit"
r = urllib.request.Request(u, headers={"Authorization": f"Basic {AUTH}", **UA})
p = json.loads(urllib.request.urlopen(r, timeout=90).read())
raw = p["content"]["raw"]
print("status:", p["status"], "| slug:", p["slug"], "| modif:", p["modified"])
print("judul:", p["title"]["raw"])
print("panjang isi:", len(raw), "char")
print("script JSON-LD utuh:", "application/ld+json" in raw)
print("FAQ mainEntity utterances:", raw.count('"@type": "Question"'))
print("H2:", len(re.findall(r"<h2>", raw)), "| H3:", len(re.findall(r"<h3>", raw)))
print("H3 pertama:", re.findall(r"<h3>(.*?)</h3>", raw)[0])
print("H3 entri terakhir:", re.findall(r"<h3>(.*?)</h3>", raw)[6])
print("entri 1 punya tautan beranda:", 'href="https://www.tslaw.id/">beranda TS Law' in raw)
print("tombol WA:", re.search(r'wa\.me/\d+', raw).group(0) if "wa.me" in raw else "TIDAK")
print("jumlah tautan internal unik:", len(set(re.findall(r'href="https://www\.tslaw\.id/([^"]*)"', raw))))
ex = re.sub(r"\s+", " ", p["excerpt"]["raw"])[:200]
print("excerpt:", ex)

# URL live: draf belum terbit, harap tidak bisa diakses publik
try:
    c = urllib.request.urlopen(urllib.request.Request(f"https://www.{DOM}/{p['slug']}/", headers=UA), timeout=45)
    print("URL live:", c.status, "(terbit)")
except urllib.error.HTTPError as e:
    print("URL live:", e.code, "(wajar kalau masih draft)")
