Update app.py

app.py

@@ -1,69 +1,229 @@
-from smolagents import CodeAgent,DuckDuckGoSearchTool, HfApiModel,load_tool,tool
-import datetime
-import requests
-import pytz
-import yaml
+from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, tool
+import re
+from typing import Dict, Any, List
+import json
 from tools.final_answer import FinalAnswerTool
 
-from Gradio_UI import GradioUI
-
-# Below is an example of a tool that does nothing. Amaze us with your creativity !
 @tool
-def my_custom_tool(arg1:str, arg2:int)-> str: #it's import to specify the return type
-    #Keep this format for the description / args / args description but feel free to modify the tool
-    """A tool that does nothing yet 
+def search_youtuber(youtuber_name: str) -> Dict[str, Any]:
+    """
+    Searches for information about a YouTuber using DuckDuckGo.
+    
     Args:
-        arg1: the first argument
-        arg2: the second argument
+        youtuber_name: Name of the YouTuber to search for
+        
+    Returns:
+        Dictionary containing search results about the YouTuber
     """
-    return "What magic will you build ?"
+    search_tool = DuckDuckGoSearchTool()
+    
+    # First search for the YouTuber's channel
+    channel_results = search_tool.run(f"{youtuber_name} youtube channel")
+    
+    # Then search for more specific information
+    stats_results = search_tool.run(f"{youtuber_name} youtube stats subscribers")
+    
+    # Search for recent content/news
+    content_results = search_tool.run(f"{youtuber_name} recent videos content")
+    
+    # Search for background information
+    background_results = search_tool.run(f"{youtuber_name} youtube background history")
+    
+    return {
+        "channel_results": channel_results,
+        "stats_results": stats_results,
+        "content_results": content_results,
+        "background_results": background_results
+    }
 
 @tool
-def get_current_time_in_timezone(timezone: str) -> str:
-    """A tool that fetches the current local time in a specified timezone.
+def extract_youtube_info(search_results: Dict[str, Any]) -> Dict[str, Any]:
+    """
+    Extracts relevant YouTuber information from search results.
+    
     Args:
-        timezone: A string representing a valid timezone (e.g., 'America/New_York').
+        search_results: Dictionary containing search results
+        
+    Returns:
+        Dictionary containing extracted YouTuber information
     """
-    try:
-        # Create timezone object
-        tz = pytz.timezone(timezone)
-        # Get current time in that 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)}"
+    # Extract channel URL
+    channel_url = None
+    for result in search_results.get("channel_results", []):
+        if "youtube.com/channel/" in result.get("url", "") or "youtube.com/c/" in result.get("url", "") or "youtube.com/@" in result.get("url", ""):
+            channel_url = result.get("url")
+            break
+    
+    # Extract subscriber count and other stats
+    subscriber_count = None
+    view_count = None
+    for result in search_results.get("stats_results", []):
+        desc = result.get("desc", "")
+        # Look for subscriber patterns like "10M subscribers" or "10 million subscribers"
+        sub_match = re.search(r"(\d+(?:\.\d+)?[KMB]?)\s+subscribers", desc, re.IGNORECASE)
+        if sub_match:
+            subscriber_count = sub_match.group(1)
+        
+        # Look for view patterns
+        view_match = re.search(r"(\d+(?:\.\d+)?[KMB]?)\s+views", desc, re.IGNORECASE)
+        if view_match:
+            view_count = view_match.group(1)
+    
+    # Extract recent videos/content
+    recent_videos = []
+    for result in search_results.get("content_results", []):
+        if "video" in result.get("desc", "").lower() or "upload" in result.get("desc", "").lower():
+            recent_videos.append({
+                "title": result.get("title"),
+                "url": result.get("url"),
+                "description": result.get("desc")
+            })
+            if len(recent_videos) >= 3:
+                break
+    
+    # Extract background information
+    background_info = []
+    for result in search_results.get("background_results", []):
+        if "started" in result.get("desc", "").lower() or "began" in result.get("desc", "").lower() or "history" in result.get("desc", "").lower():
+            background_info.append({
+                "title": result.get("title"),
+                "url": result.get("url"),
+                "description": result.get("desc")
+            })
+            if len(background_info) >= 2:
+                break
+    
+    return {
+        "channel_url": channel_url,
+        "subscriber_count": subscriber_count,
+        "view_count": view_count,
+        "recent_videos": recent_videos,
+        "background_info": background_info,
+        "raw_results": {
+            "channel": search_results.get("channel_results", [])[:3],
+            "stats": search_results.get("stats_results", [])[:3],
+            "content": search_results.get("content_results", [])[:3],
+            "background": search_results.get("background_results", [])[:3]
+        }
+    }
 
+@tool
+def format_youtuber_profile(youtuber_name: str, info: Dict[str, Any]) -> str:
+    """
+    Formats YouTuber information with proper citations.
+    
+    Args:
+        youtuber_name: Name of the YouTuber
+        info: Dictionary containing YouTuber information
+        
+    Returns:
+        Formatted information with citations
+    """
+    raw_results = info.get("raw_results", {})
+    
+    result = f"# {youtuber_name} - YouTube Creator Profile\n\n"
+    
+    # Channel information
+    result += "## Channel Information\n"
+    if info.get("channel_url"):
+        result += f"- **Channel URL:** [{info.get('channel_url')}]({info.get('channel_url')}) \n"
+    
+    if info.get("subscriber_count"):
+        result += f"- **Subscribers:** {info.get('subscriber_count')} \n"
+    
+    if info.get("view_count"):
+        result += f"- **Views:** {info.get('view_count')} \n"
+    
+    # Add channel source citation
+    if raw_results.get("channel") and len(raw_results.get("channel")) > 0:
+        source = raw_results.get("channel")[0]
+        result += f"\n: [{source.get('title')}]({source.get('url')})\n"
+    
+    # Add stats source citation
+    if raw_results.get("stats") and len(raw_results.get("stats")) > 0:
+        source = raw_results.get("stats")[0]
+        result += f": [{source.get('title')}]({source.get('url')})\n"
+    
+    # Recent content
+    if info.get("recent_videos") and len(info.get("recent_videos")) > 0:
+        result += "\n## Recent Content\n"
+        for i, video in enumerate(info.get("recent_videos")):
+            result += f"- **{video.get('title')}** $CITE_{3+i}\n"
+            result += f"  {video.get('description')}\n"
+            result += f"  [Link]({video.get('url')})\n\n"
+    
+    # Content citations
+    for i, source in enumerate(raw_results.get("content", [])[:len(info.get("recent_videos", []))]):
+        result += f"$CITE_{3+i}: [{source.get('title')}]({source.get('url')})\n"
+    
+    # Background information
+    if info.get("background_info") and len(info.get("background_info")) > 0:
+        result += "\n## Background & History\n"
+        cite_start = 3 + len(info.get("recent_videos", []))
+        for i, bg in enumerate(info.get("background_info")):
+            result += f"- **{bg.get('title')}** $CITE_{cite_start+i}\n"
+            result += f"  {bg.get('description')}\n\n"
+    
+    # Background citations
+    cite_start = 3 + len(info.get("recent_videos", []))
+    for i, source in enumerate(raw_results.get("background", [])[:len(info.get("background_info", []))]):
+        result += f"$CITE_{cite_start+i}: [{source.get('title')}]({source.get('url')})\n"
+    
+    return result
 
+# Create the specialized YouTuber information agent
+search_tool = DuckDuckGoSearchTool()
 final_answer = FinalAnswerTool()
 
-# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
-# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud' 
-
-model = HfApiModel(
-max_tokens=2096,
-temperature=0.5,
-model_id='Qwen/Qwen2.5-Coder-32B-Instruct',# it is possible that this model may be overloaded
-custom_role_conversions=None,
+youtuber_info_agent = CodeAgent(
+    model=HfApiModel(
+        max_tokens=2048,
+        temperature=0.3,
+        model_id='Qwen/Qwen2.5-Coder-32B-Instruct',
+    ),
+    tools=[
+        search_tool,
+        search_youtuber,
+        extract_youtube_info,
+        format_youtuber_profile,
+        final_answer
+    ],
+    max_steps=8,
+    verbosity_level=1,
+    name="YouTuberInfoAgent",
+    description="A specialized agent that finds and provides information about YouTubers using DuckDuckGo search with proper citations."
 )
 
+# Example prompt template
+system_prompt = """You are a specialized YouTuber information agent. Your task is to find and provide detailed information about YouTube creators when given a YouTuber's name.
 
-# Import tool from Hub
-image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)
+ALWAYS include multiple distinct sources in your response, at LEAST 3-4, by $CITE_id at the end of each information part, like , .
 
-with open("prompts.yaml", 'r') as stream:
-    prompt_templates = yaml.safe_load(stream)
-    
-agent = CodeAgent(
-    model=model,
-    tools=[final_answer], ## add your tools here (don't remove final answer)
-    max_steps=6,
-    verbosity_level=1,
-    grammar=None,
-    planning_interval=None,
-    name=None,
-    description=None,
-    prompt_templates=prompt_templates
-)
+Follow these steps:
+1. Search for the YouTuber using the search_youtuber tool
+2. Extract relevant information using the extract_youtube_info tool
+3. Format the information with proper citations using the format_youtuber_profile tool
+4. Present the information to the user in a well-organized format
 
+Always verify the information before presenting it and indicate if certain information couldn't be retrieved.
+"""
+
+# Add the agent to a Gradio UI
+from gradio import Interface
+
+def process_request(youtuber_name):
+    response = youtuber_info_agent.run(
+        system_prompt=system_prompt,
+        user_prompt=f"Please provide information about the YouTuber: {youtuber_name}"
+    )
+    return response
+
+interface = Interface(
+    fn=process_request,
+    inputs="text",
+    outputs="markdown",
+    title="YouTuber Information Agent",
+    description="Enter a YouTuber's name to get information about them with proper citations."
+)
 
-GradioUI(agent).launch()
+interface.launch()