move everything to one app on port 7860 due to hugging face space limitations

Dockerfile

@@ -17,4 +17,8 @@ COPY --chown=user requirements.txt requirements.txt
 RUN pip install --no-cache-dir --upgrade -r requirements.txt
 
 COPY --chown=user . /app
+
+# Expose Gradio port (used by Hugging Face Spaces)
+EXPOSE 7860
+
 CMD ["python", "main.py"]

README.md

@@ -23,7 +23,7 @@ An AI-powered career coaching assistant built with FastAPI, Gradio UI, LangChain
 
 ## 🚀 Features
 
-- **Dual Interface**: REST API (FastAPI) and Web UI (Gradio)
+- **Unified Interface**: Combined FastAPI and Gradio UI on a single port (7860)
 - **AI-Powered Responses**: Utilizing Mixtral-8x7B-Instruct-v0.1 model
 - **Interactive Chat Interface**: Real-time conversation with the AI agent
 - **Multi-tool Integration**: Including webpage visits and time zone conversions
@@ -31,7 +31,7 @@ An AI-powered career coaching assistant built with FastAPI, Gradio UI, LangChain
 
 ## 🛠️ Technical Stack
 
-- **Backend Framework**: FastAPI
+- **Backend Framework**: FastAPI (mounted with Gradio)
 - **UI Framework**: Gradio with SmolaGents
 - **AI Framework**: 
   - LangChain ReAct Agent (Backend) - For structured reasoning and tool usage
@@ -39,7 +39,6 @@ An AI-powered career coaching assistant built with FastAPI, Gradio UI, LangChain
 - **ML Models**: Hugging Face (Mixtral-8x7B-Instruct-v0.1)
 - **Additional Key Libraries**:
   - `uvicorn`: ASGI server
-  - `accelerate`: ML model support
   - `markdownify`: Web content processing
   - `langchain`: AI framework and tools
   - `smolagents`: UI agent framework
@@ -78,16 +77,28 @@ export HUGGINGFACEHUB_API_TOKEN=your_token_here
 python main.py
 ```
 
+The application will be available at:
+- Main UI: http://localhost:7860
+- API Documentation: http://localhost:7860/docs/
+
+## 🌐 Hugging Face Spaces Deployment
+
+This application is specifically designed to work with Hugging Face Spaces:
+- Uses a single port (7860) as required by Spaces
+- Combines FastAPI and Gradio on the same port
+- API documentation is accessible at `/docs/` on the same port
+- All functionality works within Spaces' constraints
+
 ## 📚 API Documentation
 
-Once the server is running, access the API documentation at:
-- Swagger UI: http://localhost:8000/docs
-- ReDoc: http://localhost:8000/redoc
+The API documentation is available at `/docs/` on the same port as the main application (7860). This unified setup ensures compatibility with Hugging Face Spaces while maintaining all functionality.
 
 ## 🔑 Key Endpoints
 
-- `POST /agent/query`: Send queries to the AI agent
-- `GET /`: Redirects to Gradio UI
+All endpoints are available on port 7860:
+- `/`: Main Gradio UI
+- `/docs/`: API Documentation
+- `/agent/query`: Send queries to the AI agent
 
 ## 🔍 How It Works
 
@@ -98,19 +109,12 @@ The application uses a ReAct (Reasoning and Acting) agent pattern, which follows
 4. **Thought**: The agent reasons about the observation
 5. **Action**: The agent either uses another tool or provides a final answer
 
-This pattern allows the agent to:
-- Use tools in a structured way
-- Reason step-by-step about complex problems
-- Provide transparent decision-making
-- Handle multiple tool interactions
-
 ## ⚠️ Important Notes
 
 - The application requires active internet connection for AI model access
 - Hugging Face API token is required for model access
-- The application uses the Mixtral-8x7B-Instruct-v0.1 model for generating responses
-- The UI is built using SmolaGents framework for enhanced agent interactions
-- The backend uses LangChain's ReAct agent for structured reasoning and tool usage
+- All services run on port 7860 to comply with Hugging Face Spaces requirements
+- The UI and API are served from the same port for better integration
 
 ## 🤝 Contributing
 

UI.py

@@ -1,7 +1,5 @@
-import mimetypes
 import os
 import re
-import shutil
 from typing import Optional, List, Dict, Any, Callable
 
 from smolagents.agent_types import AgentAudio, AgentImage, AgentText, handle_agent_output_types
@@ -208,8 +206,8 @@ class AgentUI:
         self.chat_history = []
         return "Started new conversation"
     
-    def launch(self, server_name: str = "0.0.0.0", server_port: int = 7860, **kwargs):
-        """Launch the Gradio interface"""
+    def get_gradio_app(self):
+        """Get the Gradio app for mounting in FastAPI"""
         api_port = int(self.api_url.split(":")[-1])
         
         with gr.Blocks(css="""
@@ -265,12 +263,16 @@ class AgentUI:
                 outputs=[chatbot, menu_output]
             )
             docs_btn.click(
-                fn=lambda: f"<script>window.open('{self.api_url.split(':')[0]}:8000/docs', '_blank')</script>",
+                fn=lambda: f"<script>window.open('/docs', '_blank')</script>",
                 inputs=[],
                 outputs=[menu_output]
             )
         
-        # Launch the interface
+        return interface
+
+    def launch(self, server_name: str = "0.0.0.0", server_port: int = 7860, **kwargs):
+        """Launch the Gradio interface standalone (for development)"""
+        interface = self.get_gradio_app()
         interface.launch(server_name=server_name, server_port=server_port, **kwargs)
 
 __all__ = ["stream_to_gradio", "AgentUI"]

app.py

@@ -4,6 +4,7 @@ from langchain.agents import AgentExecutor, create_react_agent
 from langchain_core.prompts import PromptTemplate
 
 from tools.visit_webpage import visit_webpage
+from tools.final_answer import final_answer
 
 import gradio as gr
 import datetime 
@@ -62,16 +63,6 @@ def get_current_time(timezone: str = "UTC") -> str:
     except Exception as e:
         return f"Error: {str(e)}"
 
-@tool
-def visit_webpage(url: str) -> str:
-    """Visit a webpage and return its content as markdown."""
-    try:
-        response = requests.get(url, timeout=10)
-        response.raise_for_status()
-        return f"Successfully visited {url}. Content length: {len(response.text)} characters"
-    except Exception as e:
-        return f"Error visiting webpage: {str(e)}"
-
 # Load system prompt and template
 with open("prompts.yaml", 'r') as stream:
     prompt_templates = yaml.safe_load(stream)
@@ -80,7 +71,7 @@ with open("prompts.yaml", 'r') as stream:
 prompt = PromptTemplate.from_template(prompt_templates["template"])
 
 # Create the agent
-tools = [get_current_time, visit_webpage]  
+tools = [get_current_time]  
 agent = create_react_agent(
     llm=llm,
     tools=tools,
@@ -104,10 +95,9 @@ class QueryRequest(BaseModel):
 async def root():
     return HTMLResponse("<h2>Welcome! Please use the Gradio UI above.</h2>")
 
-@app.get("/docs")
+@app.get("/docs", include_in_schema=False)
 async def redirect_to_docs():
-    base_url = app.url_path_for('root').replace('/docs', '')
-    return RedirectResponse(url=f"{base_url}:8000/docs")
+    return RedirectResponse(url="/docs/")
 
 @app.post("/agent/query")
 async def query_agent(request: QueryRequest):

docker-compose.yml

@@ -4,7 +4,6 @@ services:
   app:
     build: .
     ports:
-      - "8000:8000"
       - "7860:7860"
     env_file:
       - .env

main.py

@@ -1,36 +1,24 @@
-import threading
-import time
 import uvicorn
 from app import app, agent
 from UI import AgentUI
 
 def main():
     """
-    Run both FastAPI and Gradio UI together
+    Run FastAPI and Gradio UI on the same port
     """
     # Configuration
-    api_port = 8000
-    ui_port = 7860
+    port = 7860
     
-    # Start FastAPI in a background thread
-    def run_fastapi():
-        uvicorn.run(app, host="0.0.0.0", port=api_port)
-    
-    api_thread = threading.Thread(target=run_fastapi)
-    api_thread.daemon = True
-    api_thread.start()
-    
-    # Give FastAPI time to start
-    print(f"Starting FastAPI server on port {api_port}...")
-    time.sleep(1)
-    
-    # Start Gradio UI in the main thread
-    print(f"Starting Gradio UI on port {ui_port}...")
+    # Create and mount Gradio app
     gradio_ui = AgentUI(
         agent=agent,
-        api_url=f"http://localhost:{api_port}"
+        api_url=f"http://localhost:{port}"
     )
-    gradio_ui.launch(server_port=ui_port)
+    app.mount("/", gradio_ui.get_gradio_app())
+    
+    # Start the combined app
+    print(f"Starting combined server on port {port}...")
+    uvicorn.run(app, host="0.0.0.0", port=port)
 
 if __name__ == "__main__":
     main() 

requirements.txt

@@ -4,7 +4,6 @@ requests>=2.31.0
 fastapi>=0.104.1
 uvicorn[standard]>=0.24.0
 gradio>=4.7.1
-accelerate>=0.26.0
 langchain>=0.1.0
 langchain-core>=0.1.0
 langchain-community>=0.0.13
@@ -13,5 +12,4 @@ pydantic>=2.5.2
 pytz>=2023.3
 python-dateutil>=2.8.2
 huggingface-hub>=0.19.4
-python-multipart>=0.0.6
-aiohttp>=3.9.0
+python-multipart>=0.0.6

tools/web_search.py

@@ -1,27 +0,0 @@
-from typing import Any, Optional
-from smolagents.tools import Tool
-import duckduckgo_search
-
-class DuckDuckGoSearchTool(Tool):
-    name = "web_search"
-    description = "Performs a duckduckgo web search based on your query (think a Google search) then returns the top search results."
-    inputs = {'query': {'type': 'string', 'description': 'The search query to perform.'}}
-    output_type = "string"
-
-    def __init__(self, max_results=10, **kwargs):
-        super().__init__()
-        self.max_results = max_results
-        try:
-            from duckduckgo_search import DDGS
-        except ImportError as e:
-            raise ImportError(
-                "You must install package `duckduckgo_search` to run this tool: for instance run `pip install duckduckgo-search`."
-            ) from e
-        self.ddgs = DDGS(**kwargs)
-
-    def forward(self, query: str) -> str:
-        results = self.ddgs.text(query, max_results=self.max_results)
-        if len(results) == 0:
-            raise Exception("No results found! Try a less restrictive/shorter query.")
-        postprocessed_results = [f"[{result['title']}]({result['href']})\n{result['body']}" for result in results]
-        return "## Search Results\n\n" + "\n\n".join(postprocessed_results)