agentic_implementation/gradio_ag.py

import gradio as gr
import json
from typing import Any, Dict, List, Tuple

from re_act import (
    get_plan_from_llm,
    think,
    act,
    store_name_email_mapping,
    extract_sender_info,
    client,
)
from logger import logger  # Assumes logger is configured
from schemas import PlanStep

# Maintain persistent session results
session_results: Dict[str, Any] = {}

def respond(
    message: str,
    history: List[Tuple[str, str]],
    system_message: str,
    max_tokens: int,
    temperature: float
) -> str:
    logger.info("Gradio agent received message: %s", message)
    full_response = ""

    try:
        # Step 1: Generate plan
        plan = get_plan_from_llm(message)
        logger.debug("Generated plan: %s", plan)
        full_response += "📌 **Plan**:\n"
        for step in plan.plan:
            full_response += f"- {step.action}\n"
        full_response += "\n"

        results = {}

        # Step 2: Execute steps
        for step in plan.plan:
            if step.action == "done":
                full_response += "✅ Plan complete.\n"
                break

            should_run, updated_step, user_prompt = think(step, results, message)

            # Ask user for clarification if needed
            if user_prompt:
                full_response += f"❓ {user_prompt} (Please respond with an email)\n"
                return full_response  # wait for user

            if not should_run:
                full_response += f"⏭️ Skipping `{step.action}`\n"
                continue

            try:
                output = act(updated_step)
                results[updated_step.action] = output
                full_response += f"🔧 Ran `{updated_step.action}` → {output}\n"
            except Exception as e:
                logger.error("Error running action '%s': %s", updated_step.action, e)
                full_response += f"❌ Error running `{updated_step.action}`: {e}\n"
                break

        # Step 3: Summarize results
        try:
            summary_rsp = client.chat.completions.create(
                model="gpt-4o-mini",
                temperature=temperature,
                max_tokens=max_tokens,
                messages=[
                    {"role": "system", "content": "Summarize these results for the user in a friendly way."},
                    {"role": "assistant", "content": json.dumps(results)}
                ],
            )
            summary = summary_rsp.choices[0].message.content
            full_response += "\n📋 **Summary**:\n" + summary
        except Exception as e:
            logger.error("Summary generation failed: %s", e)
            full_response += "\n❌ Failed to generate summary."

    except Exception as e:
        logger.exception("Unhandled error in agent: %s", e)
        full_response += f"\n❌ Unexpected error: {e}"

    return full_response


demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(label="System message", value="You are an email assistant agent."),
        gr.Slider(label="Max tokens", minimum=64, maximum=2048, value=512, step=1),
        gr.Slider(label="Temperature", minimum=0.0, maximum=1.5, value=0.7, step=0.1),
    ],
    title="📬 Email Agent",
    description="Ask me anything related to your email tasks!"
)

if __name__ == "__main__":
    demo.launch()