#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""Periksa akhir draf artikel tslaw.id #3698 sebelum dilaporkan."""
import re
import subprocess
import sys

sys.path.insert(0, "/home/mantapman/.hermes/profiles/atun/workspace")
import tslaw_wp as T  # noqa: E402

CJK_RE = re.compile(r"[\u3000-\u9fff\uac00-\ud7af]")
SPASI_RE = re.compile(r"\s+")

j = T.ambil_post(3698)
raw = j.get("_raw") or ""
print(f"#{j['id']} [{j.get('status')}] {j['title']['rendered']}")
print(f"slug: {j.get('slug')}")
print(f"kategori: {j.get('categories')} | tag: {len(j.get('tags') or [])} | featured: {j.get('featured_media')}")
print(f"panjang HTML: {len(raw)} char")

cek = {
    "<video": raw.count("<video"),
    "VideoObject": raw.count("VideoObject"),
    "FAQPage": raw.count("FAQPage"),
    "ld+json": raw.count("application/ld+json"),
    "<h2>": raw.count("<h2>"),
    "<h3>": raw.count("<h3>"),
    "tautan internal tslaw.id": len(re.findall(r'href="https://www\.tslaw\.id/', raw)),
    "tautan WA": raw.count("wa.me/6281909222026"),
}
for k, v in cek.items():
    print(f"  {k}: {v}")

print(f"  em dash: {raw.count(chr(8212))} | kutip melengkung: {len(re.findall(chr(8220)+'|'+chr(8221), raw))}")
print(f"  CJK: {len(CJK_RE.findall(raw))}")

teks = re.sub(r"<script.*?</script>", " ", raw, flags=re.S)
teks = re.sub(r"<[^>]+>", " ", teks)
print(f"  perkiraan jumlah kata: {len(re.sub(SPASI_RE, ' ', teks).split())}")

print("\n=== media di dalam artikel ===")
for u in set(re.findall(r'https://mantapserver[^\s"\'<>]+', raw)):
    r = subprocess.run(["curl", "-4", "-s", "-o", "/dev/null", "-w", "%{http_code}",
                        "--max-time", "60", "-I", u], capture_output=True, text=True)
    print(f"  {r.stdout.strip()}  {u}")

print("\n=== tautan internal ===")
for u in set(re.findall(r'href="(https://www\.tslaw\.id/[^"]+)"', raw)):
    r = subprocess.run(["curl", "-4", "-s", "-o", "/dev/null", "-w", "%{http_code}",
                        "--max-time", "60", "-L", u], capture_output=True, text=True)
    print(f"  {r.stdout.strip()}  {u}")

print(f"\npratinjau di admin: https://www.tslaw.id/wp-admin/post.php?post=3698&action=edit")
