Function ppt_to_images

utils/src/utils.py:140–178

(file: str, output_dir: str, warning: bool = False, dpi=72, output_type='png')

@tenacity
def ppt_to_images(file: str, output_dir: str, warning: bool = False, dpi=72, output_type='png'):
    assert pexists(file), f"File {file} does not exist"
    if pexists(output_dir) and warning:
        print(f"ppt2images: {output_dir} already exists")
    os.makedirs(output_dir, exist_ok=True)
    with tempfile.TemporaryDirectory() as temp_dir:
        # Create unique user installation directory for LibreOffice to avoid concurrency issues
        with tempfile.TemporaryDirectory() as user_install_dir:
            command_list = [
                "soffice",
                "--headless",
                "--norestore",
                "--nolockcheck",
                f"-env:UserInstallation=file://{user_install_dir}",
                "--convert-to",
                "pdf",
                file,
                "--outdir",
                temp_dir,
            ]
            # Set environment to ensure UTF-8 encoding
            env = os.environ.copy()
            env['LC_ALL'] = 'en_US.UTF-8'
            env['LANG'] = 'en_US.UTF-8'
            subprocess.run(command_list, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, env=env)

        for f in os.listdir(temp_dir):
            if not f.endswith(".pdf"):
                continue
            temp_pdf = pjoin(temp_dir, f)
            images = convert_from_path(temp_pdf, dpi=72)
            for i, img in enumerate(images):
                if output_type == 'png':
                    img.save(pjoin(output_dir, f"poster.png"), 'PNG')
                else:
                    img.save(pjoin(output_dir, f"poster.jpg"), 'JPEG')
            return

        raise RuntimeError("No PDF file was created in the temporary directory", file)


@tenacity