#!/usr/bin/env python3 """Perbaiki tautan mati di StudioSoft. Aturan: - Tautan internal mati → dicari pengganti yang hidup berdasarkan kemiripan kata pada slug; kalau tidak ada, tautannya dilepas (teks jangkarnya tetap, jadi kalimat tidak rusak). - Tautan luar mati → tautannya dilepas, teksnya tetap. - Tautan luar ke domain mencurigakan (PBN/spam) → dilepas juga walau hidup. Jalankan tanpa argumen = uji. Dengan --go = kerjakan (dengan cadangan). """ import base64, json, re, sys, time, urllib.request WS = "/home/mantapman/.hermes/profiles/atun/workspace" DOM = "studiosoft.id" S = f"https://{DOM}" AUTH = base64.b64encode(b"bambangbot:q26r RePo BVy6 pwU2 F3z7 ouyl").decode() H = {"Authorization": f"Basic {AUTH}", "User-Agent": "Mozilla/5.0", "Content-Type": "application/json"} UA = {"User-Agent": "Mozilla/5.0"} ECURIGA = ("rentpbn", "my.id/jasa-seo", "backlink", "pbn", "stikeshb", "repository.", "blogspot") LEWATI = ("/wp-json/", "facebook.com", "instagram.com", "twitter.com", "x.com", "tiktok.com", "youtube.com", "wa.me", "whatsapp.com") PETA = { "https://studiosoft.id/blog": "https://studiosoft.id/artikel/", "https://studiosoft.id/blog/": "https://studiosoft.id/artikel/", "https://studiosoft.id/konsultasi/": "https://studiosoft.id/kontak/", "https://studiosoft.id/google-business-profile-untuk-klinik/": "https://studiosoft.id/google-business-profile-klinik/", "https://studiosoft.id/panduan-seo-untuk-klinik-dan-rumah-sakit/": "https://studiosoft.id/panduan-seo-klinik-rumah-sakit/", "https://studiosoft.id/analitik-digital-bisnis-properti/": "https://studiosoft.id/analitik-properti/", "https://studiosoft.id/digital-marketing-bisnis-keuangan/": "https://studiosoft.id/marketing-keuangan/", "https://studiosoft.id/digital-marketing-bisnis-retail/": "https://studiosoft.id/digital-marketing-retail/", "https://studiosoft.id/digital-marketing-bisnis-properti/": "https://studiosoft.id/digital-marketing-properti/", "https://studiosoft.id/digital-marketing-bisnis-manufaktur/": "https://studiosoft.id/marketing-manufaktur/", "https://studiosoft.id/seo-untuk-bisnis-properti/": "https://studiosoft.id/seo-properti/", "https://studiosoft.id/klinik-kecil-butuh-digital-marketing/": "https://studiosoft.id/jasa-digital-marketing-klinik/", "https://studiosoft.id/strategi-digital-marketing-rumah-sakit/": "https://studiosoft.id/7-strategi-digital-marketing-rumah-sakit-untuk-menambah-pasien-setiap-bulan/", } data = json.load(open(f"{WS}/drafts/ss_tautan_mati.json")) # buang tautan yang tidak relevan diperiksa: URL API, sosial media yang menolak bot, dan balasan 403 # (403 = server menolak pemeriksa otomatis, belum tentu halaman mati) data = {u: v for u, v in data.items() if not any(k in u for k in LEWATI) and v.get("status") != 403} GO = "--go" in sys.argv def rest(p, method="GET", data_=None): body = json.dumps(data_).encode() if data_ is not None else None r = urllib.request.Request(S + p, data=body, headers=H, method=method) for _ in range(3): try: with urllib.request.urlopen(r, timeout=180) as x: return json.loads(x.read()) except Exception: time.sleep(4) raise RuntimeError("gagal ambil " + p) def hidup(u): try: with urllib.request.urlopen(urllib.request.Request(u, headers=UA), timeout=25) as r: return r.status == 200 except Exception: return False def kata(s): return {k for k in re.split(r"[-_/]+", s.lower()) if len(k) > 3} print("=== daftar halaman hidup di situs ===") HIDUP = {} for jenis in ("posts", "pages"): hal = 1 while hal <= 8: d = rest(f"/wp-json/wp/v2/{jenis}?per_page=50&page={hal}&status=publish&_fields=id,slug,link") if not isinstance(d, list) or not d: break for x in d: HIDUP[x["link"]] = x["slug"] if len(d) < 50: break hal += 1 print(f" {len(HIDUP)} halaman hidup") def cari_pengganti(u): """Cari halaman hidup yang paling mirip dengan tautan internal yang mati.""" sasaran = kata(u) skor = [] for link, slug in HIDUP.items(): s = len(sasaran & kata(slug)) if s: skor.append((s, link)) skor.sort(reverse=True) if skor and skor[0][0] >= 3: return skor[0][1] return None print("=== rencana perbaikan ===") per_artikel = {} for u, info in data.items(): for d in info.get("dari", []): per_artikel.setdefault((d.get("jenis", "posts"), d["pid"]), []).append( {"url": u, "jangkar": d["jangkar"], "status": info.get("status")}) print(f" {len(data)} tautan bermasalah · tersebar di {len(per_artikel)} artikel/halaman") for u, info in list(data.items())[:40]: curiga = any(k in u.lower() for k in ECURIGA) print(f" {str(info.get('status'))[:12]:<13} {'⚠ mencurigakan ' if curiga else ''}{u[:92]}") cadangan_isi, ubah = {}, 0 for (jenis, pid), daftar in per_artikel.items(): d = rest(f"/wp-json/wp/v2/{jenis}/{pid}?context=edit&_fields=content,slug,link") isi = d["content"]["raw"] asli = isi rincian = [] for item in daftar: u = item["url"] pola = re.compile(r']+href="' + re.escape(u) + r'"[^>]*>(.*?)', re.S) if not pola.search(isi): continue pengganti = None if "studiosoft.id" in u or u.startswith("/"): pengganti = PETA.get(u) or cari_pengganti(u) if pengganti: isi = pola.sub(lambda mo: f'{mo.group(1)}', isi) rincian.append(f"diganti → {pengganti}") else: isi = pola.sub(lambda mo: mo.group(1), isi) rincian.append(f"dilepas · {u[:70]}") ubah += 1 if isi != asli: cadangan_isi[f"{jenis}:{pid}"] = asli print(f"\n #{pid} [{jenis}] {d['slug'][:50]} — {len(rincian)} tautan") for r in rincian: print(f" {r}") if GO: for kunci, isi in cadangan_isi.items(): jenis, pid = kunci.split(":") rest(f"/wp-json/wp/v2/{jenis}/{pid}", "POST", {"content": isi}) time.sleep(0.8) json.dump(cadangan_isi, open(f"{WS}/ss_backup_tautan_mati.json", "w"), ensure_ascii=False) print(f"\n✅ {ubah} tautan diperbaiki di {len(cadangan_isi)} artikel/halaman · " f"cadangan: ss_backup_tautan_mati.json") else: print(f"\n(uji) {ubah} tautan akan diperbaiki di {len(cadangan_isi)} artikel/halaman. " f"Pakai --go untuk mengerjakan.")