#!/usr/bin/env python3
"""Tarik isi mentah kandidat perbaikan 2026-10-10 (HANYA BACA) ke tslaw_pemulihan/perbaikan_20261010/."""
import base64, html as htmlmod, json, os, re, sys, urllib.request

WS = "/home/mantapman/.hermes/profiles/atun/workspace"
sys.path.insert(0, WS)
from cfg_tslaw import CFG

AUTH = base64.b64encode(f"{CFG['user']}:{CFG['pass']}".encode()).decode()
BASE = "https://tslaw.id/wp-json/wp/v2"
HDR = {"Authorization": f"Basic {AUTH}", "User-Agent": "Mozilla/5.0"}
OUT = f"{WS}/tslaw_pemulihan/perbaikan_20261010"
os.makedirs(OUT, exist_ok=True)

for pid in [int(x) for x in sys.argv[1:]]:
    req = urllib.request.Request(f"{BASE}/posts/{pid}?context=edit", headers=HDR)
    with urllib.request.urlopen(req, timeout=120) 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)
    t = htmlmod.unescape(re.sub(r"<(script|style).*?</\1>", " ", raw, flags=re.S))
    print(pid, p["slug"], "kata_tanpa_skema=", len(re.sub(r"<[^>]+>", " ", t).split()))
    print("  judul:", p["title"]["raw"])
    m = p.get("meta") or {}
    for k in ("_yoast_wpseo_focuskw", "_yoast_wpseo_title", "_yoast_wpseo_metadesc"):
        print(f"  {k}: {m.get(k)!r}")
