ocr.py

ocr_notebook_map = {
  "easyocr": "easyocr.ipynb",
  "pytesseract": "pytesseract.ipynb",
  "kerasocr": "kerasocr.ipynb",
  "paddle": "paddle.ipynb",
  "tensorflow": "tensorflow.ipynb"
}

def run_ocr(CONFIG):
  if CONFIG.get("skip_ocr_processing", False):
    print("Skipping OCR.")
    return

  notebook_name = ocr_notebook_map.get(CONFIG["ocr_library"])

  if notebook_name:
    %run ocr_module/{notebook_name}
  else:
    print(f"Unknown OCR library: {CONFIG['ocr_library']}")

  file_list = get_file_list(CONFIG["directory"])
  write_results(file_list, CONFIG["output_file"])