#!/usr/bin/env python3
"""Ambil post 2940 (kandidat perbaikan 2026-10-05) via REST untuk diperiksa."""
import base64, json, sys, urllib.request
sys.path.insert(0, "/home/mantapman/.hermes/profiles/atun/workspace")
from cfg_tslaw import CFG

AUTH = base64.b64encode(f"{CFG['user']}:{CFG['pass']}".encode()).decode()
HDR = {"Authorization": f"Basic {AUTH}", "User-Agent": "Mozilla/5.0"}
OUT = "/home/mantapman/.hermes/profiles/atun/workspace/tslaw_pemulihan/perbaikan_20261005"
import os
os.makedirs(OUT, exist_ok=True)

for PID in (2940, 2823):
    req = urllib.request.Request(
        f"https://tslaw.id/wp-json/wp/v2/posts/{PID}?context=edit", headers=HDR)
    with urllib.request.urlopen(req, timeout=90) as r:
        p = json.load(r)
    raw = p["content"]["raw"]
    open(f"{OUT}/post_{PID}_SEBELUM.html", "w").write(raw)
    json.dump(p, open(f"{OUT}/post_{PID}_SEBELUM.json", "w"), ensure_ascii=False, indent=1)
    m = p.get("meta") or {}
    import re
    print(f"=== #{PID} /{p['slug']}/ status={p['status']} modified={p['modified']}")
    print(f"    H1      : {p['title']['raw']}")
    print(f"    seo     : {m.get('_yoast_wpseo_title')!r}")
    print(f"    fokus   : {m.get('_yoast_wpseo_focuskw')!r}")
    print(f"    meta    : {m.get('_yoast_wpseo_metadesc')!r}")
    print(f"    kata    : {len(re.sub(r'<[^>]+>', ' ', raw).split())}")
    print(f"    WA      : {raw.count('wa.me/')}")
    print(f"    pilar   : {raw.count('law-firm-surabaya')}")
    print(f"    FAQ     : {len(re.findall(r'rank-math-faq-question', raw))} tayang | schema Question: {len(re.findall(chr(34)+'@type'+chr(34)+': '+chr(34)+'Question'+chr(34), raw))}")
    print(f"    H2      : {re.findall(r'<h2[^>]*>(.*?)</h2>', raw)}")
    print(f"    H3      : {re.findall(r'<h3[^>]*>(.*?)</h3>', raw)[:12]}")
    print(f"    tipe blok: {sorted(set(re.findall(r'<!-- wp:([a-z0-9/-]+)', raw)))}")
    print()
