import re,sys,html
sys.path.insert(0,"/home/mantapman/.hermes/profiles/atun/workspace")
from openssl_fetch import fetch
from _ambil_artikel_wp_18okt import strip_chunked  # reuse
urls=[
 "https://studiosoft.id/biaya-digital-agency-surabaya/",
 "https://www.tslaw.id/kontrak-bisnis-yang-aman/",
 "https://www.theskincodeclinic.com/kulit-sensitif-penyebab-dan-cara-merawat/",
 "https://dasindonesia.com/servis-rutin-mobil-alphard-jadwal-lengkap-komponen-yang-dicek-dan-tips-dari-teknisi/",
]
for u in urls:
    st,h,b=fetch(u,timeout=90)
    body=strip_chunked(b).decode("utf-8","replace")
    body=re.sub(r"(?is)<(script|style|nav|header|footer|aside)[^>]*>.*?</\1>"," ",body)
    txt=re.sub(r"(?s)<[^>]+>"," ",body)
    txt=html.unescape(re.sub(r"\s+"," ",txt))
    print("="*70); print(u,"STATUS",st,"len",len(txt))
    print(txt[:2200])
