app.py

import gradio as gr
from huggingface_hub import InferenceClient
from smolagents import DuckDuckGoSearchTool
import re

web_search = DuckDuckGoSearchTool()

SYSTEM_PROMPT = """You are an AI research assistant that can search the web. Follow these steps:

1. FIRST ANALYZE the user's question:
   - If information is missing or ambiguous, ask ONE clarifying question
   - If clear, proceed to search

2. When searching:
   - Generate multiple specific search queries wrapped in <search> tags
   - Focus on factual keywords, one query per line
   Example:
   <search>
   Pont des Arts exact length meters
   History of Pont des Arts bridge
   </search>

3. After receiving results:
   - Analyze information from multiple sources
   - Cross-verify facts
   - If needed, generate follow-up searches
   - Provide final answer with:
     - Clear structure
     - Key facts with sources
     - Concise explanations

Never invent information. Cite sources for all facts. Use neutral, academic tone."""

def process_searches(response):
    searches = re.findall(r'<search>(.*?)</search>', response, re.DOTALL)
    if searches:
        queries = [q.strip() for q in searches[0].split('\n') if q.strip()]
        results = []
        for query in queries:
            search_result = web_search(query)
            results.append(f"🔍 Search results for '{query}':\n{search_result}\n")
        return '\n'.join(results)
    return None

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
    hf_token,
):
    client = InferenceClient(
        provider="hf-inference",
        api_key=hf_token
    )
    
    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})
    
    full_response = ""
    search_cycle = True
    
    while search_cycle:
        search_cycle = False
        
        completion = client.chat.completions.create(
            model="Qwen/QwQ-32B",
            messages=messages,
            max_tokens=10000,
            temperature=temperature,
            top_p=top_p,
            stream=True
        )
        
        response = ""
        for chunk in completion:
            token = chunk.choices[0].delta.content or ""
            response += token
            full_response += token
            yield full_response

        search_results = process_searches(response)
        
    if search_results:
        search_cycle = True
        messages.append({"role": "assistant", "content": response})
        messages.append({
            "role": "user",
            "content": f"SEARCH RESULTS:\n{search_results}\nAnalyze these results..."
        })
        # Add this line to display queries
        full_response += f"\n🔍 SEARCH QUERIES USED:\n{chr(10).join(queries)}\n\n"  # ← NEW
        full_response += "\n🔍 Analyzing search results...\n"
        yield full_response

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value=SYSTEM_PROMPT, label="System Prompt", lines=8),
        gr.Slider(minimum=1000, maximum=15000, value=6000, step=500, label="Max Tokens"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.5, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.85, step=0.05, label="Top-p"),
        gr.Textbox(label="HF API Token", type="password")
    ],
    title="Web Research Agent 🤖",
    description="Advanced AI assistant with web search capabilities",
    examples=[
        ["Compare COVID-19 mortality rates between US and Sweden with sources"],
        ["What's the current consensus on dark matter composition?"],
        ["Latest advancements in fusion energy 2023-2024"]
    ]
)

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