"""CLI : traite un PDF ou un répertoire de PDFs.

Usage :
    python -m pipeline.cli <pdf|dir> [--out output/v2]
"""
import argparse
import glob
import sys
import time
from pathlib import Path

from .extract import extract_dossier
from .persist import save_result


def main():
    p = argparse.ArgumentParser(description="Pipeline OGC v1 (GLM-OCR)")
    p.add_argument("input", help="PDF unique ou répertoire contenant des PDFs")
    p.add_argument("--out", default="output/v2", help="Répertoire de sortie JSON")
    p.add_argument("--quiet", action="store_true")
    args = p.parse_args()

    input_path = Path(args.input)
    if input_path.is_dir():
        pdfs = sorted(input_path.glob("*.pdf"))
    elif input_path.is_file() and input_path.suffix.lower() == ".pdf":
        pdfs = [input_path]
    else:
        # Globbing si chemin avec espaces/motifs
        pdfs = [Path(p) for p in sorted(glob.glob(str(input_path))) if p.lower().endswith(".pdf")]

    if not pdfs:
        print(f"Aucun PDF trouvé pour : {args.input}")
        return 1

    print(f"{len(pdfs)} PDF(s) à traiter → {args.out}")
    t0 = time.time()
    for pdf in pdfs:
        t_pdf = time.time()
        try:
            result = extract_dossier(pdf, verbose=not args.quiet)
            out_path = save_result(result, args.out)
            print(f"  ✓ {pdf.name} → {out_path}  ({time.time()-t_pdf:.1f}s)")
        except Exception as e:
            print(f"  ✗ {pdf.name} : {e}")
            import traceback
            traceback.print_exc()
    print(f"Terminé en {time.time()-t0:.1f}s")
    return 0


if __name__ == "__main__":
    sys.exit(main())