#!/usr/bin/env python3
"""OCR slide 01 (+02) semua deck carousel feed IG SS yang masih pending (10-31 Okt)."""
import os, re
from rapidocr_onnxruntime import RapidOCR
ocr = RapidOCR()
IMG = "/home/mantapman/.hermes/profiles/atun/workspace/openclaw-img"
NAV = re.compile(r"skip to content|beranda|layanan jasa|whatsapp|hubungi|tentang kami|artikel digital marketing|social media|web development", re.I)

DECKS = [
    ("2026-10-10", "ss_car10_seomalang"),
    ("2026-10-11", "ss_car10_maintenance"),
    ("2026-10-12", "ss_car10_kontenai"),
    ("2026-10-13", "ss_car10_firmahukum"),
    ("2026-10-14", "ss_car10_promptai"),
    ("2026-10-15", "ss_car10_maklon"),
    ("2026-10-16", "ss_car10_aiumkm"),
    ("2026-10-17", "ss_car10_bengkelmobil"),
    ("2026-10-18", "ss_car10_chatbotwa"),
    ("2026-10-19", "ss_car10_creativeagency"),
    ("2026-10-20", "ss_car10_otomasiai"),
    ("2026-10-21", "ss_car10_agencypartner"),
    ("2026-10-22", "ss_car10_konsultanai"),
    ("2026-10-23", "ss_car10_automasimarketing"),
    ("2026-10-24", "ss_car10_brandingagency"),
    ("2026-10-25", "ss_car10_aiagent"),
    ("2026-10-26", "ss_car10_retainer"),
    ("2026-10-27", "ss_car10_brandingcampaign"),
    ("2026-10-28", "ss_car10_kelolaakun"),
    ("2026-10-29", "ss_car10_pilihseo"),
    ("2026-10-30", "ss_car10_biayawebsite"),
    ("2026-10-31", "ss_car10_seo2026"),
]

def baca(p):
    r, _ = ocr(p)
    return " ".join(t[1] for t in r) if r else ""

for tgl, pfx in DECKS:
    f = f"{IMG}/{pfx}_01.jpg"
    if not os.path.exists(f):
        print(f"{tgl} {pfx:28s} NOFILE"); continue
    t = baca(f)
    print(f"{tgl} {pfx:28s} {'JUNK' if NAV.search(t) else 'ok  '} | {t[40:200]}")
