#!/usr/bin/env python3
"""Periksa kenapa teks berubah saat tautan tak terlihat dihapus: bandingkan isi sebelum vs sesudah."""
import difflib, json, re

WS = "/home/mantapman/.hermes/profiles/atun/workspace"
cadangan = json.load(open(f"{WS}/ss_backup_tautan_tak_terlihat.json"))
KOSONG = re.compile(r'<a\b[^>]*>\s*</a>', re.S)


def polos(h):
    return re.sub(r"\s+", " ", re.sub("<[^>]+>", " ", h)).strip()


n = 0
for kunci, isi in cadangan.items():
    baru = KOSONG.sub("", isi)
    t1, t2 = polos(isi), polos(baru)
    if t1 == t2:
        continue
    n += 1
    if n > 3:
        break
    print(f"=== {kunci} ===")
    for i, s in enumerate(difflib.unified_diff(t1.split(), t2.split(), lineterm="", n=2)):
        if i > 14:
            break
        print("   " + s[:150])
    # tunjukkan anchor yang dihapus
    for mo in KOSONG.finditer(isi):
        print(f"   anchor dihapus: {mo.group(0)[:150]}")
    print()
print(f"total entri yang teksnya berbeda: {sum(1 for k, v in cadangan.items() if polos(v) != polos(KOSONG.sub('', v)))}")
