app.py

from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
import datetime
import requests
import pytz
import yaml
import time
import os
from tools.final_answer import FinalAnswerTool
from huggingface_hub import InferenceClient, login
from pydub.generators import WhiteNoise
from pydub import AudioSegment
import gradio as gr
from Gradio_UI import GradioUI

# Load Hugging Face token
hf_token = os.getenv("HF_TOKEN")  # Fetch secret from Hugging Face Space
if hf_token:
    login(hf_token)
else:
    print("⚠️ Hugging Face API token is missing! Set HF_TOKEN in Space secrets.")

# ✅ Sound generation tool with structured docstring for smolagents
@tool
def generate_sound(sound_type: str, duration: int) -> str:
    """Generates a simple sound based on the specified type and duration.
    
    Args:
        sound_type: Type of sound to generate (e.g., 'rain', 'white_noise').
        duration: Duration of the sound in seconds.

    Returns:
        Path to the generated audio file.
    """
    try:
        duration_ms = duration * 1000  # Convert to milliseconds
        if sound_type == "rain":
            sound = WhiteNoise().to_audio_segment(duration=duration_ms).low_pass_filter(5000)

        else:
            return f"Unsupported sound type: {sound_type}"

        output_path = f"/tmp/{sound_type}_{duration}s.wav"
        sound.export(output_path, format="wav")
        return output_path

    except Exception as e:
        return f"Error generating sound: {str(e)}"




# ✅ Time zone tool
@tool
def get_current_time_in_timezone(timezone: str) -> str:
    """Fetches the current local time in a specified timezone."""
    try:
        tz = pytz.timezone(timezone)
        local_time = datetime.datetime.now(tz).strftime("%Y-%m-%d %H:%M:%S")
        return f"The current local time in {timezone} is: {local_time}"
    except Exception as e:
        return f"Error fetching time for timezone '{timezone}': {str(e)}"

# ✅ Load the final answer tool
final_answer = FinalAnswerTool()

# ✅ Model configuration
model = HfApiModel(
    max_tokens=2096,
    temperature=0.5,
    model_id='Qwen/Qwen2.5-Coder-32B-Instruct',
    custom_role_conversions=None,
    token=hf_token
)

# ✅ Load external tools
image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)

# ✅ Load prompt templates
with open("prompts.yaml", 'r') as stream:
    prompt_templates = yaml.safe_load(stream)

# ✅ Define the agent (Removed `output_modality`)
agent = CodeAgent(
    model=model,
    tools=[final_answer, generate_sound, image_generation_tool],  # Included image generation tool
    max_steps=6,
    verbosity_level=1,
    grammar=None,
    planning_interval=None,
    name=None,
    description=None,
    prompt_templates=prompt_templates
)

# ✅ Start the UI with processing time display
def launch_with_processing_time():
    def wrapped_launch():
        start_time = time.time()
        print("Processing request...")
        GradioUI(agent).launch()
        end_time = time.time()
        print(f"Processing completed in {end_time - start_time:.2f} seconds")

    wrapped_launch()

launch_with_processing_time()