#!/usr/bin/env python3
"""OCR slide 1..10 deck seomalang + scan deck pending lain (slide 01-02) untuk teks navigasi bocor."""
import glob, os, re, sys, json
from rapidocr_onnxruntime import RapidOCR
ocr = RapidOCR()

NAV = re.compile(r"skip to content|beranda|layanan jasa|whatsapp|hubungi|tentang kami|artikel digital marketing", re.I)

def baca(p):
    res, _ = ocr(p)
    if not res:
        return ""
    return " ".join(t[1] for t in res)

def cari(prefix, ns):
    out = []
    for n in ns:
        f = f"/home/mantapman/.hermes/profiles/atun/workspace/openclaw-img/{prefix}_{n:02d}.jpg"
        if not os.path.exists(f):
            out.append((n, "NOFILE", ""))
            continue
        t = baca(f)
        junk = bool(NAV.search(t))
        out.append((n, "JUNK" if junk else "ok", t[:180]))
    return out

mode = sys.argv[1] if len(sys.argv) > 1 else "satu"
if mode == "satu":
    for n, st, t in cari("ss_car10_seomalang", range(1, 11)):
        print(f"--- slide {n:02d} [{st}]\n{t}\n")
