#!/usr/bin/env python3
"""Pilih kandidat link FREE (belum pernah dipakai di Threads akun manapun) + cek HTTP 200."""
import json, re, subprocess, sys

used = json.load(open('/tmp/threads_used_all.json'))
def norm(u): return u.rstrip('/').lower()
locs = re.findall(r'<loc>(.*?)</loc>', open('/tmp/ss_post_sm.xml', encoding='utf-8', errors='replace').read())

ss_used = {norm(x) for x in used['studiosoft']}
uky_used = {norm(x) for x in used['uky']}
inter = [l for l in locs if norm(l) not in ss_used and norm(l) not in uky_used]
print('kandidat FREE untuk SS+uky:', len(inter))

PREF_SS = ['apa-itu-seo-pengertian-cara-kerja-manfaat', 'strategi-guerilla-marketing-umkm',
           '7-tips-mendapatkan-backlink-berkualitas', '7-teknik-seo-on-page', '12-panduan-seo-vs-geo']
PREF_UKY = ['rahasia-growth-hacking', '10-langkah-mudah-mulai-bisnis-online', '5-alasan-kenapa-umkm-wajib-punya-website',
            'digital-marketing-2026-strategi-studiosoft', 'kelola-social-media-untuk-brand']
pick = {}
for pref in PREF_SS:
    m = [l for l in inter if pref in l]
    if m:
        pick['studiosoft'] = m[0]
        break
for pref in PREF_UKY:
    m = [l for l in inter if pref in l]
    if m:
        pick['uky'] = m[0]
        break

pick['tslaw'] = 'https://www.tslaw.id/klausul-penyelesaian-sengketa-kontrak-umkm/'
pick['tsc'] = 'https://www.theskincodeclinic.com/ciri-ciri-dokter-kecantikan-krian-terpercaya/'
pick['das'] = 'https://dasindonesia.com/standar-kerja-bengkel-mobil-premium-surabaya/'

for k, u in pick.items():
    r = subprocess.run(['curl', '-4', '-sSL', '-o', '/dev/null', '-w', '%{http_code}',
                        '--max-time', '45', u], capture_output=True, text=True)
    code = r.stdout.strip()
    print(f"{k:11s} http={code}  {u}")
json.dump(pick, open('/tmp/pick_links_26sep.json', 'w'), indent=1)
