Function request_gemini_with_video

src/gpt_request.py:124–189

Makes a multimodal request to the Gemini-2.5 model using video + text. Args: prompt (str): The user instruction, e.g., "Please evaluate and suggest improvements for this educational animation." video_path (str): Local path to the video file (MP4 preferred, <20MB recommended

(prompt: str, video_path: str, log_id=None, max_tokens: int = 10000, max_retries: int = 3)

def request_gemini_with_video(prompt: str, video_path: str, log_id=None, max_tokens: int = 10000, max_retries: int = 3):
    """
    Makes a multimodal request to the Gemini-2.5 model using video + text.

    Args:
        prompt (str): The user instruction, e.g., "Please evaluate and suggest improvements for this educational animation."
        video_path (str): Local path to the video file (MP4 preferred, <20MB recommended).
        log_id (str, optional): Tracking ID
        max_tokens (int): Max response token length
        max_retries (int): Max retry attempts

    Returns:
        dict: The Gemini model response
    """
    base_url = cfg("gemini", "base_url")
    api_version = cfg("gemini", "api_version")
    api_key = cfg("gemini", "api_key")
    model_name = cfg("gemini", "model")

    client = openai.AzureOpenAI(
        azure_endpoint=base_url,
        api_version=api_version,
        api_key=api_key,
    )

    if log_id is None:
        log_id = generate_log_id()

    extra_headers = {"X-TT-LOGID": log_id}

    # Load and base64-encode video
    if not os.path.exists(video_path):
        raise FileNotFoundError(f"Video not found: {video_path}")

    with open(video_path, "rb") as f:
        video_bytes = f.read()

    video_base64 = base64.b64encode(video_bytes).decode("utf-8")
    data_url = f"data:video/mp4;base64,{video_base64}"

    retry_count = 0
    while retry_count < max_retries:
        try:
            completion = client.chat.completions.create(
                model=model_name,
                messages=[
                    {
                        "role": "user",
                        "content": [
                            {"type": "text", "text": prompt},
                            {"type": "image_url", "image_url": {"url": data_url, "detail": "high"}, "media_type": "video/mp4"},
                        ],
                    }
                ],
                max_tokens=max_tokens,
                extra_headers=extra_headers,
            )
            return completion