#!/usr/bin/env python3
"""Verifikasi akhir artikel #9332 setelah dijadwalkan (isi, meta, CJK, tombol, tanggal)."""
import base64, json, re, urllib.request

AUTH = base64.b64encode(b"bambangbot:q26r RePo BVy6 pwU2 F3z7 ouyl").decode()
H = {"Authorization": f"Basic {AUTH}", "User-Agent": "Mozilla/5.0"}
u = ("https://studiosoft.id/wp-json/wp/v2/posts/9332?context=edit"
     "&_fields=id,status,date,date_gmt,slug,title,content,meta,categories,link")
with urllib.request.urlopen(urllib.request.Request(u, headers=H), timeout=180) as x:
    p = json.loads(x.read())

raw = p["content"]["raw"]
bersih = re.sub(r"<script.*?</script>", " ", raw, flags=re.S | re.I)
teks = re.sub(r"\s+", " ", re.sub(r"<[^>]+>", " ", bersih)).strip()
m = p.get("meta") or {}
cjk = re.findall(r"[\u3000-\u9fff\uac00-\ud7af]", raw)
print(f"id        : {p['id']}")
print(f"status    : {p['status']}")
print(f"date/date_gmt: {p['date']} / {p['date_gmt']}")
print(f"slug      : {p['slug']}  kategori={p['categories']}")
print(f"judul     : {re.sub('<[^>]+>', '', p['title']['rendered'])}")
print(f"panjang   : {len(teks.split())} kata (tanpa tag)")
print(f"h2/h3     : {raw.count('<h2>')}/{raw.count('<h3>')}")
print(f"meta      : fokus='{m.get('_yoast_wpseo_focuskw')}' judul='{m.get('_yoast_wpseo_title')}' "
      f"desc={len(m.get('_yoast_wpseo_metadesc') or '')} char")
print(f"tombol WA : {'ya' if 'wa.me/6281935000624' in raw else 'TIDAK'} · "
      f"'{'Konsultasi Gratis' if 'Konsultasi Gratis' in raw else 'TIDAK'}'")
print(f"FAQPage   : {'ya' if 'FAQPage' in raw else 'TIDAK'}")
print(f"em dash   : {raw.count(chr(8212))} · kutip melengkung: {len(re.findall(r'[“”‘’]', raw))} · CJK: {len(cjk)}")
POLA_INT = r'href="(https://studiosoft\.id/[^"]+)"'
print("internal  : " + str(re.findall(POLA_INT, raw)))
