Deploy to Hugging Face Spaces - Remove Render, enable full features

.dockerignore

@@ -1,6 +1,7 @@
 # Git
 .git
 .gitignore
+.gitattributes
 
 # Python
 __pycache__
@@ -8,11 +9,11 @@ __pycache__
 *$py.class
 *.so
 .Python
-.env
 .venv
 env/
 venv/
 ENV/
+.python-version
 
 # IDE
 .vscode
@@ -28,4 +29,7 @@ Thumbs.db
 
 # Docker
 docker-compose*.yml
+
+# UV
+uv.lock
 .dockerignore

.env

@@ -0,0 +1,3 @@
+# Environment variables
+GROQ_API_KEY=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
+TAVILY_API_KEY=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

Dockerfile

@@ -22,8 +22,8 @@ COPY . .
 # Create directories for data persistence
 RUN mkdir -p /app/workspace_data /app/chroma_db
 
-# Expose ports
-EXPOSE 8000 8501
+# Hugging Face Spaces uses port 7860
+EXPOSE 7860
 
 # Copy startup script
 COPY startup.sh /app/startup.sh

Dockerfile.backend

@@ -1,29 +0,0 @@
-FROM python:3.11-slim
-
-WORKDIR /app
-
-# Install system dependencies
-RUN apt-get update && apt-get install -y \
-    build-essential \
-    gcc \
-    curl \
-    && rm -rf /var/lib/apt/lists/*
-
-# Copy requirements first for caching
-COPY requirements.txt .
-
-# Install Python dependencies
-RUN pip install --no-cache-dir --upgrade pip && \
-    pip install --no-cache-dir -r requirements.txt
-
-# Copy application code
-COPY . .
-
-# Create directories for data persistence
-RUN mkdir -p /app/workspace_data /app/chroma_db
-
-# Expose port
-EXPOSE 8000
-
-# Run FastAPI
-CMD ["uvicorn", "app.api:app", "--host", "0.0.0.0", "--port", "8000"]

Dockerfile.frontend

@@ -1,26 +0,0 @@
-FROM python:3.11-slim
-
-WORKDIR /app
-
-# Install system dependencies
-RUN apt-get update && apt-get install -y \
-    build-essential \
-    gcc \
-    curl \
-    && rm -rf /var/lib/apt/lists/*
-
-# Copy requirements first for caching
-COPY requirements.txt .
-
-# Install Python dependencies
-RUN pip install --no-cache-dir --upgrade pip && \
-    pip install --no-cache-dir -r requirements.txt
-
-# Copy application code
-COPY . .
-
-# Expose port
-EXPOSE 8501
-
-# Run Streamlit
-CMD ["streamlit", "run", "streamlit_app.py", "--server.port=8501", "--server.address=0.0.0.0", "--server.headless=true"]

Dockerfile.render

@@ -1,29 +0,0 @@
-FROM python:3.11-slim
-
-WORKDIR /app
-
-# Install system dependencies
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    curl \
-    dos2unix \
-    && rm -rf /var/lib/apt/lists/* \
-    && apt-get clean
-
-# Copy and install Python dependencies (cached layer)
-COPY requirements.txt .
-RUN pip install --no-cache-dir --upgrade pip && \
-    pip install --no-cache-dir -r requirements.txt
-
-# Copy ALL application code
-COPY . .
-
-# Create workspace directories
-RUN mkdir -p /app/workspace_data /app/chroma_db
-
-# Fix line endings and make startup script executable
-RUN dos2unix /app/start.sh && chmod +x /app/start.sh
-
-# Render requires port 10000
-EXPOSE 10000
-
-CMD ["/bin/bash", "/app/start.sh"]

README.md

@@ -0,0 +1,46 @@
+---
+title: Perplexity AI Clone
+emoji: 🔍
+colorFrom: blue
+colorTo: purple
+sdk: docker
+pinned: false
+license: mit
+---
+
+# Perplexity AI Clone
+
+An AI-powered search and research assistant with multiple modes:
+
+- **Automatic**: Auto-routes to best mode based on query
+- **Web Search**: Real-time web search with citations
+- **RAG**: Search uploaded documents
+- **Deep Research**: Multi-step research with synthesis
+- **Agentic**: Multi-agent RAG with planning
+- **Analysis**: Data analysis and insights
+- **Summarize**: Summarize content and documents
+
+## Features
+
+- 🔍 Real-time web search with Tavily
+- 📚 Document upload and RAG
+- 🤖 LangGraph-powered pipelines
+- 💡 Follow-up question suggestions
+- 🖼️ Image search integration
+- 📊 Knowledge panels
+
+## Environment Variables
+
+Set these secrets in your Hugging Face Space:
+
+- `GROQ_API_KEY`: Your Groq API key
+- `TAVILY_API_KEY`: Your Tavily API key
+
+## Tech Stack
+
+- FastAPI backend
+- Streamlit frontend
+- LangChain + LangGraph
+- Groq LLM (llama3-70b-8192)
+- FAISS vector store
+- Sentence Transformers embeddings

app/api.py

@@ -54,12 +54,12 @@ app.add_middleware(
 
 
 # =======================================================
-# Health Check Endpoint (for Azure Container Apps)
+# Health Check Endpoint
 # =======================================================
 @app.get("/health")
 async def health_check():
     """Health check endpoint for container orchestration"""
-    return {"status": "healthy", "service": "perplexity-clone-api", "lite_mode": Config.LITE_MODE}
+    return {"status": "healthy", "service": "perplexity-clone-api"}
 
 
 # =======================================================
@@ -76,47 +76,35 @@ browse_tool = BrowseTool()
 image_search = TavilyImageSearch()
 summarizer = SummarizerTool()
 
-# Only load heavy components if not in LITE_MODE
-if not Config.LITE_MODE:
-    reranker = Reranker()
-    knowledge_panel = KnowledgePanel()
-    
-    # RAG demo vectorstore
-    processor = DocumentProcessor(
-        chunk_size=Config.CHUNK_SIZE,
-        chunk_overlap=Config.CHUNK_OVERLAP,
-    )
-    demo_docs = processor.load_url("https://lilianweng.github.io/posts/2023-06-23-agent/")
-    demo_splits = processor.split(demo_docs)
-    
-    vector = VectorStore()
-    vector.create(demo_splits)
-else:
-    reranker = None
-    knowledge_panel = None
-    vector = None
-    print("⚡ LITE_MODE: Skipping heavy embeddings to save memory")
+# Load all components
+reranker = Reranker()
+knowledge_panel = KnowledgePanel()
+
+# RAG demo vectorstore
+processor = DocumentProcessor(
+    chunk_size=Config.CHUNK_SIZE,
+    chunk_overlap=Config.CHUNK_OVERLAP,
+)
+demo_docs = processor.load_url("https://lilianweng.github.io/posts/2023-06-23-agent/")
+demo_splits = processor.split(demo_docs)
+
+vector = VectorStore()
+vector.create(demo_splits)
 
 # File manager for per-workspace document RAG
 file_manager = FileManager(base_dir="workspace_data")
 
 # =======================================================
-# Initialize All LangGraph Pipelines (only if not LITE_MODE)
+# Initialize All LangGraph Pipelines
 # =======================================================
-if not Config.LITE_MODE:
-    deep_graph = DeepResearchGraph(vector)
-    rag_graph = RAGOnlyGraph(file_manager)
-    agentic_graph = AgenticRAGGraph(file_manager, vector, image_search)
-else:
-    deep_graph = None
-    rag_graph = None
-    agentic_graph = None
-
+deep_graph = DeepResearchGraph(vector)
+rag_graph = RAGOnlyGraph(file_manager)
+agentic_graph = AgenticRAGGraph(file_manager, vector, image_search)
 web_graph = WebSearchGraph()
 analysis_graph = AnalysisGraph()
 summarize_graph = SummarizeGraph()
 
-print("✅ All LangGraph pipelines initialized!" if not Config.LITE_MODE else "✅ LITE MODE: Core pipelines initialized!")
+print("✅ All LangGraph pipelines initialized!")
 
 
 # =======================================================
@@ -487,15 +475,9 @@ def deep_research(req: ChatRequest):
     memory.add(ws, "user", q)
 
     try:
-        if deep_graph is None:
-            # LITE_MODE fallback - use web search instead
-            state = web_graph.run(q)
-            answer = state.get("answer", "No answer generated.")
-            sources = state.get("sources", [])
-        else:
-            state = deep_graph.run(q)
-            answer = state.get("final_answer", "No answer generated.")
-            sources = state.get("sources", [])
+        state = deep_graph.run(q)
+        answer = state.get("final_answer", "No answer generated.")
+        sources = state.get("sources", [])
     except Exception as e:
         print(f"Deep research error: {e}")
         answer = f"Deep research encountered an error. Please try again."
@@ -971,17 +953,11 @@ def rag_mode(req: ModeRequest):
     memory.add(ws, "user", q)
     
     try:
-        if rag_graph is None:
-            # LITE_MODE fallback
-            answer = "RAG mode requires document uploads. In lite mode, please use Web Search instead."
-            sources = []
-            follow = []
-        else:
-            # Run the RAGOnlyGraph pipeline
-            state = rag_graph.run(q, ws)
-            answer = state.get("answer", "No answer generated.")
-            sources = state.get("sources", [])
-            follow = state.get("followups", [])
+        # Run the RAGOnlyGraph pipeline
+        state = rag_graph.run(q, ws)
+        answer = state.get("answer", "No answer generated.")
+        sources = state.get("sources", [])
+        follow = state.get("followups", [])
     except Exception as e:
         print(f"RAG error: {e}")
         answer = f"RAG mode encountered an error: {str(e)[:100]}"
@@ -1014,22 +990,13 @@ def agentic_mode(req: ModeRequest):
     print(f"\n🤖 AGENTIC MODE (LangGraph): {q}")
     
     try:
-        if agentic_graph is None:
-            # LITE_MODE fallback - use web search
-            state = web_graph.run(q)
-            answer = state.get("answer", "No answer generated.")
-            sources = state.get("sources", [])
-            links = state.get("links", [])
-            images = tavily_images_safe(q)
-            follow = state.get("followups", [])
-        else:
-            # Run the AgenticRAGGraph pipeline
-            state = agentic_graph.run(q, ws)
-            answer = state.get("answer", "No answer generated.")
-            sources = state.get("sources", [])
-            links = state.get("links", [])
-            images = state.get("images", [])
-            follow = state.get("followups", [])
+        # Run the AgenticRAGGraph pipeline
+        state = agentic_graph.run(q, ws)
+        answer = state.get("answer", "No answer generated.")
+        sources = state.get("sources", [])
+        links = state.get("links", [])
+        images = state.get("images", [])
+        follow = state.get("followups", [])
     except Exception as e:
         print(f"Agentic error: {e}")
         answer = f"Agentic mode encountered an error: {str(e)[:100]}"

config/config.py

@@ -8,14 +8,11 @@ class Config:
     GROQ_API_KEY = os.getenv("GROQ_API_KEY")
     TAVILY_API_KEY = os.getenv("TAVILY_API_KEY")
     
-    # Groq model
-    LLM_MODEL = os.getenv("LLM_MODEL", "openai/gpt-oss-20b")
+    # Groq model - using llama3-70b-8192 for best performance
+    LLM_MODEL = os.getenv("LLM_MODEL", "llama3-70b-8192")
 
     CHUNK_SIZE = 400
     CHUNK_OVERLAP = 80
-    
-    # Disable heavy features on free tier (512MB RAM limit)
-    LITE_MODE = os.getenv("LITE_MODE", "true").lower() == "true"
 
     @classmethod
     def get_llm(cls):

render.yaml

@@ -1,20 +0,0 @@
-services:
-  - type: web
-    name: perplexity-clone
-    env: docker
-    dockerfilePath: ./Dockerfile.render
-    dockerContext: .
-    plan: free
-    region: oregon
-    healthCheckPath: /health
-    envVars:
-      - key: GROQ_API_KEY
-        sync: false
-      - key: TAVILY_API_KEY
-        sync: false
-      - key: BACKEND_URL
-        value: http://localhost:8000
-      - key: LITE_MODE
-        value: "true"
-      - key: LLM_MODEL
-        value: "openai/gpt-oss-20b"

requirements.txt

@@ -1,5 +1,5 @@
 # =============================================
-# LIGHTWEIGHT REQUIREMENTS FOR AZURE APP SERVICE  
+# REQUIREMENTS FOR HUGGING FACE SPACES
 # =============================================
 
 # Core LangChain (compatible versions)
@@ -26,9 +26,8 @@ streamlit==1.31.1
 requests==2.31.0
 httpx==0.26.0
 
-# Embeddings - USE CPU-ONLY TORCH (smaller)
---extra-index-url https://download.pytorch.org/whl/cpu
-torch==2.2.0+cpu
+# Embeddings
+torch
 sentence-transformers==2.3.1
 
 # Vector search
@@ -37,7 +36,7 @@ faiss-cpu==1.7.4
 # Web search
 tavily-python==0.3.3
 
-# Scraping (lightweight)
+# Scraping
 beautifulsoup4==4.12.3
 lxml==5.1.0
 

start.sh

@@ -1,19 +0,0 @@
-#!/bin/bash
-set -e
-
-# Start FastAPI backend on port 8000 (internal only)
-echo "Starting FastAPI backend on port 8000..."
-uvicorn app.api:app --host 127.0.0.1 --port 8000 &
-BACKEND_PID=$!
-
-echo "Waiting for backend to initialize..."
-sleep 5
-echo "Backend started (PID: $BACKEND_PID)"
-
-# Start Streamlit on port 10000 (Render's expected port)
-echo "Starting Streamlit on port 10000..."
-exec streamlit run streamlit_app.py \
-    --server.port=10000 \
-    --server.address=0.0.0.0 \
-    --server.headless=true \
-    --browser.gatherUsageStats=false

startup.sh

@@ -16,7 +16,7 @@ if ! kill -0 $BACKEND_PID 2>/dev/null; then
 fi
 
 echo "Backend started successfully (PID: $BACKEND_PID)"
-echo "Starting Streamlit frontend on port 8501..."
+echo "Starting Streamlit frontend on port 7860..."
 
-# Start Streamlit frontend (this will keep the container running)
-streamlit run streamlit_app.py --server.port 8501 --server.address 0.0.0.0 --server.headless true
+# Start Streamlit frontend on port 7860 (Hugging Face Spaces default)
+streamlit run streamlit_app.py --server.port 7860 --server.address 0.0.0.0 --server.headless true

supervisord.conf

@@ -1,29 +0,0 @@
-[supervisord]
-nodaemon=true
-user=root
-logfile=/dev/stdout
-logfile_maxbytes=0
-loglevel=info
-
-[program:fastapi]
-command=uvicorn app.api:app --host 0.0.0.0 --port 8000 --workers 1
-directory=/app
-autostart=true
-autorestart=true
-stdout_logfile=/dev/stdout
-stdout_logfile_maxbytes=0
-stderr_logfile=/dev/stderr
-stderr_logfile_maxbytes=0
-priority=1
-
-[program:streamlit]
-command=streamlit run streamlit_app.py --server.port=10000 --server.address=0.0.0.0 --server.headless=true --browser.gatherUsageStats=false
-directory=/app
-autostart=true
-autorestart=true
-stdout_logfile=/dev/stdout
-stdout_logfile_maxbytes=0
-stderr_logfile=/dev/stderr
-stderr_logfile_maxbytes=0
-priority=2
-startsecs=10