#!/usr/bin/env python3
"""
Generated test file for Vision LLM Test
Framework: litellm
Execution mode: sync
Streaming: yes"""

import os
import sentry_sdk

import time

import litellm
from sentry_sdk.integrations.litellm import LiteLLMIntegration
from sentry_sdk.integrations.openai import OpenAIIntegration


sentry_sdk.init(
    dsn=os.environ.get("SENTRY_DSN"),
    traces_sample_rate=1.0,
    send_default_pii=True,
    stream_gen_ai_spans=True,
    integrations=[LiteLLMIntegration(include_prompts=True)],
    disabled_integrations=[OpenAIIntegration()],
)


def main():
    # Turn 1
    stream = litellm.responses(
        model="openai/gpt-4o-mini",
        instructions="You are a helpful assistant that can analyze images. Be concise.",
        input=[
        {
            "role": "user",
            "content": [
                {"type": "input_text", "text": "What color is this image? Reply with just the color name."},
                {"type": "input_image", "image_url": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mP8z8BQz0AEYBxVSF+FABJADveWkH6oAAAAAElFTkSuQmCC"},
            ]
        },
    ]
,
        stream=True,
    )

    collected_content = []
    for event in stream:
        if hasattr(event, "type") and event.type == "response.output_text.delta":
            collected_content.append(event.delta)

    full_response = "".join(collected_content)
    print(f"Turn 1 Response: {full_response}")
    time.sleep(0.1)  # sleep is necessary for LiteLLM because it needs threaded callbacks to finish

if __name__ == "__main__":
    with sentry_sdk.start_transaction(op="test", name="Vision LLM Test"):
        main()
    sentry_sdk.flush(timeout=5)