"""Test PaddleOCR sur les 4 dossiers de référence."""
import time
from pathlib import Path
from paddleocr import PaddleOCR
from pipeline.ingest import pdf_to_images

CASES = ["2018 CARC/OGC 7.pdf", "2018 CARC/OGC 27.pdf", "2018 CARC/OGC 55.pdf", "2018 CARC/OGC 86.pdf"]


def main():
    print("--- Chargement PaddleOCR ---")
    t0 = time.time()
    # lang='fr' pour le français
    ocr = PaddleOCR(lang='fr', use_textline_orientation=True)
    print(f"Chargé en {time.time()-t0:.1f}s")

    Path("test_paddle_out").mkdir(exist_ok=True)
    for pdf in CASES:
        name = Path(pdf).stem
        img_path = str(pdf_to_images(pdf)[0])
        print(f"\n=== {name} ===")
        t0 = time.time()
        result = ocr.predict(img_path)
        elapsed = time.time() - t0
        lines = []
        for r in result:
            for txt in r.get('rec_texts', []):
                lines.append(txt)
        print(f"  ({elapsed:.1f}s, {len(lines)} lignes)")
        out = "\n".join(lines)
        print(out[:1500])
        (Path("test_paddle_out") / f"{name}.txt").write_text(out)


if __name__ == "__main__":
    main()