#!/usr/bin/env python3
"""Lihat sisa teks di sekitar tautan lama (PBN/spam) yang sudah dilepas, cek apakah kalimatnya masih wajar."""
import json, re

WS = "/home/mantapman/.hermes/profiles/atun/workspace"
d = json.load(open(f"{WS}/ss_backup_tautan_mati_lengkap.json"))
TARGET = ("rentpbn", "stikeshb", "papdisurabaya", "localseo.studio", "binari.co.id",
          "repository.", "blog.google", "developers.google.com", "wipo.int", "yoast.com/ai-seo",
          "perplexity.ai", "seoptimer")

for kunci, isi in d.items():
    for t in TARGET:
        for mo in re.finditer(r'<a[^>]+href="([^"]*' + re.escape(t) + r'[^"]*)"[^>]*>(.*?)</a>', isi, re.S):
            jangkar = re.sub(r"\s+", " ", re.sub("<[^>]+>", "", mo.group(2))).strip()
            mulai = max(0, mo.start() - 170)
            konteks = re.sub(r"\s+", " ", re.sub("<[^>]+>", " ", isi[mulai:mo.end() + 90])).strip()
            print(f"■ {kunci} · jangkar: «{jangkar[:70]}»")
            print(f"   …{konteks[:300]}…\n")
