Add Conversational Analytics (Chat with Data) feature

app/main.py

@@ -657,3 +657,72 @@ async def analysis_history(
             for row in rows
         ]
     }
+
+
+class ChatRequest(BaseModel):
+    query: str
+
+@app.post("/chat/{job_id}")
+async def chat_with_data(job_id: str, req: ChatRequest, _: None = Depends(verify_api_key)):
+    """
+    Conversational Analytics feature: Allows users to ask follow-up questions
+    and executes safe Pandas queries using LLM.
+    """
+    from app.agent.nodes import LLM_MODEL
+    from langchain_groq import ChatGroq
+    from langchain_core.messages import HumanMessage
+    import pandas as pd
+    import json
+    
+    with get_db_session() as db:
+        job = db.query(AnalysisJob).filter(AnalysisJob.id == job_id).first()
+        if not job:
+            raise HTTPException(status_code=404, detail="Job not found.")
+            
+    # Load dataset securely
+    df = None
+    if DISABLE_DATA_PERSISTENCE:
+        from app.utils.data_store import get_dataset
+        df = get_dataset(job_id)
+    elif os.path.exists(job.file_path):
+        try:
+            df = pd.read_csv(job.file_path)
+        except UnicodeDecodeError:
+            df = pd.read_csv(job.file_path, encoding="latin-1")
+            
+    if df is None:
+        raise HTTPException(
+            status_code=410, 
+            detail="Raw dataset has been securely discarded from memory (Zero-Retention Mode). Chat feature unavailable."
+        )
+
+    llm = ChatGroq(model=LLM_MODEL, temperature=0.1, max_tokens=1024)
+    # Ask LLM to generate a safe pandas expression
+    schema_info = str(df.dtypes.to_dict())
+    prompt = f"""You are an expert Data Analyst Agent. You have a pandas DataFrame 'df' loaded with these columns and types: {schema_info}.
+The user asks: "{req.query}".
+Write a SINGLE line of Python code using pandas that evaluates to a string or number answering this question. 
+It must be purely an expression (e.g. df['Sales'].mean() or len(df)). No imports, no assignments. Just the expression.
+If the question is just greeting/conversational, reply with 'CONVERSATIONAL: ' followed by your response.
+Output ONLY the expression or conversational answer."""
+    
+    try:
+        ai_expr = llm.invoke([HumanMessage(content=prompt)]).content.strip(" \n`'\"")
+        if ai_expr.startswith("Python") or ai_expr.startswith("python"):
+            ai_expr = ai_expr[6:].strip(" \n`")
+            
+        if ai_expr.startswith("CONVERSATIONAL:"):
+            return {"response": ai_expr.replace("CONVERSATIONAL:", "").strip()}
+            
+        # Safely evaluate
+        allowed_globals = {"__builtins__": {}, "pd": pd}
+        allowed_locals = {"df": df}
+        raw_result = eval(ai_expr, allowed_globals, allowed_locals)
+        
+        # Format the result back into english
+        explanation_prompt = f"The user asked: '{req.query}'. The python result is: {raw_result}. Provide a short, direct human-friendly answer based on this."
+        final_answer = llm.invoke([HumanMessage(content=explanation_prompt)]).content
+        return {"response": final_answer}
+    except Exception as e:
+        logger.error(f"Chat error: {e}")
+        return {"response": "I couldn't calculate that from the data right now. Could you rephrase the question?"}

frontend/static/app.js

@@ -528,4 +528,66 @@ document.addEventListener('DOMContentLoaded', () => {
             logTerminal('[Error]: Download failed - Network Error', true);
         }
     }
+
+    // --- Chat with Data Logic ---
+    const chatBtn = document.getElementById('chat-btn');
+    const chatInput = document.getElementById('chat-input');
+    const chatHistory = document.getElementById('chat-history');
+
+    async function sendChatMessage() {
+        if (!currentJobId) return;
+        const query = chatInput.value.trim();
+        if (!query) return;
+
+        // Add user message
+        const userMsg = document.createElement('div');
+        userMsg.style = "align-self: flex-end; background: #3b82f6; color: white; padding: 8px 12px; border-radius: 8px; max-width: 80%; word-wrap: break-word;";
+        userMsg.innerText = query;
+        chatHistory.appendChild(userMsg);
+        
+        chatInput.value = '';
+        chatInput.disabled = true;
+        chatBtn.disabled = true;
+        chatBtn.innerHTML = '<i class="fa-solid fa-spinner fa-spin"></i>';
+        
+        chatHistory.scrollTop = chatHistory.scrollHeight;
+
+        try {
+            const res = await fetch(buildUrl(`/chat/${currentJobId}`), {
+                method: 'POST',
+                headers: buildHeaders({ 'Content-Type': 'application/json' }),
+                body: JSON.stringify({ query: query })
+            });
+            
+            const data = await res.json();
+            
+            const aiMsg = document.createElement('div');
+            aiMsg.style = "align-self: flex-start; background: #334155; color: #f8fafc; padding: 8px 12px; border-radius: 8px; max-width: 80%; word-wrap: break-word; border:1px solid #475569;";
+            if (res.ok) {
+                aiMsg.innerText = data.response;
+            } else {
+                aiMsg.innerText = `Error: ${data.detail || 'Could not process query.'}`;
+            }
+            chatHistory.appendChild(aiMsg);
+        } catch (error) {
+            const aiMsg = document.createElement('div');
+            aiMsg.style = "align-self: flex-start; background: #7f1d1d; color: #f8fafc; padding: 8px 12px; border-radius: 8px; border:1px solid #991b1b;";
+            aiMsg.innerText = "Network error. Please try again.";
+            chatHistory.appendChild(aiMsg);
+        } finally {
+            chatInput.disabled = false;
+            chatBtn.disabled = false;
+            chatBtn.innerHTML = 'Send <i class="fa-solid fa-paper-plane"></i>';
+            chatHistory.scrollTop = chatHistory.scrollHeight;
+            chatInput.focus();
+        }
+    }
+
+    if (chatBtn && chatInput) {
+        chatBtn.addEventListener('click', sendChatMessage);
+        chatInput.addEventListener('keypress', (e) => {
+            if (e.key === 'Enter') sendChatMessage();
+        });
+    }
+
 });

frontend/templates/index.html

@@ -141,6 +141,19 @@
                               </div>
                           </div>
 
+                          <!-- Conversational AI Data Chat -->
+                          <div class="panel glass-panel chat-panel" style="margin-top:2rem;">
+                              <h3><i class="fa-solid fa-messages"></i> Chat with your Data</h3>
+                              <p style="color:#94a3b8; font-size:0.9rem; margin-bottom:1rem;">Ask follow-up questions. The AI will write and execute safe Pandas code dynamically.</p>
+                              <div id="chat-history" style="min-height:100px; max-height:300px; overflow-y:auto; padding:1rem; background:rgba(0,0,0,0.2); border-radius:8px; margin-bottom:1rem; display:flex; flex-direction:column; gap:10px;">
+                                  <!-- Chat messages injected here -->
+                              </div>
+                              <div style="display:flex; gap:10px;">
+                                  <input type="text" id="chat-input" placeholder="E.g., What was the average salary?" style="flex:1; padding:10px; border-radius:6px; border:1px solid #334155; background:#1e293b; color:white;">
+                                  <button id="chat-btn" class="upload-btn" style="width:100px; margin-top:0;">Send <i class="fa-solid fa-paper-plane"></i></button>
+                              </div>
+                          </div>
+
                           <div class="panel glass-panel">
                               <h3>Export Reports</h3>
                               <div class="export-actions">