Fix response template with strict format, fix data file downloads with proper links, fix plot URL formatting

FRRONTEEEND/components/ChatInterface.tsx

@@ -746,20 +746,29 @@ export const ChatInterface: React.FC<{ onBack: () => void }> = ({ onBack }) => {
                           <h4 className="text-xs font-bold uppercase tracking-wider text-white/60">Visualizations ({allPlots.length})</h4>
                         </div>
                         <div className="space-y-2">
-                          {allPlots.map((plot, idx) => (
-                            <button
-                              key={idx}
-                              onClick={() => setReportModalUrl(plot.url)}
-                              className="w-full p-3 rounded-lg bg-white/5 border border-white/10 hover:bg-emerald-500/10 hover:border-emerald-500/30 transition-all text-left group"
-                            >
-                              <div className="flex items-center justify-between">
-                                <span className="text-sm text-white/80 truncate flex-1">{plot.title}</span>
-                                <ChevronRight className="w-4 h-4 text-white/40 group-hover:text-emerald-400 transition-all" />
-                              </div>
-                              <span className="text-xs text-white/40 mt-1 block">{plot.type || 'interactive'}</span>
-                            </button>
-                          ))}
-                        </div>
+                          {allPlots.map((plot, idx) => {
+                            // Ensure URL is properly formatted
+                            let plotUrl = plot.url;
+                            if (plotUrl && plotUrl.startsWith('./outputs/')) {
+                              plotUrl = plotUrl.replace('./outputs/', '/outputs/');
+                            } else if (plotUrl && !plotUrl.startsWith('/outputs/')) {
+                              plotUrl = `/outputs/${plotUrl.replace(/^outputs\//, '')}`;
+                            }
+                            
+                            return (
+                              <button
+                                key={idx}
+                                onClick={() => setReportModalUrl(plotUrl || plot.url)}
+                                className="w-full p-3 rounded-lg bg-white/5 border border-white/10 hover:bg-emerald-500/10 hover:border-emerald-500/30 transition-all text-left group"
+                              >
+                                <div className="flex items-center justify-between">
+                                  <span className="text-sm text-white/80 truncate flex-1">{plot.title}</span>
+                                  <ChevronRight className="w-4 h-4 text-white/40 group-hover:text-emerald-400 transition-all" />
+                                </div>
+                                <span className="text-xs text-white/40 mt-1 block">{plot.type || 'interactive'}</span>
+                              </button>
+                            );
+                          })}\n                        </div>
                       </div>
                     )}
                     
@@ -774,21 +783,27 @@ export const ChatInterface: React.FC<{ onBack: () => void }> = ({ onBack }) => {
                           {uniqueDataFiles.map((file, idx) => {
                             // Extract filename from path
                             const fileName = file.split('/').pop() || file;
-                            // Create download URL
-                            const downloadUrl = file.startsWith('/') ? file : `/${file}`;
+                            // Create proper download URL
+                            let downloadUrl = file;
+                            if (downloadUrl.startsWith('./outputs/')) {
+                              downloadUrl = downloadUrl.replace('./outputs/', '/outputs/');
+                            } else if (!downloadUrl.startsWith('/outputs/')) {
+                              downloadUrl = `/outputs/${file.replace(/^outputs\//, '')}`;
+                            }
                             
                             return (
-                              <button
+                              <a
                                 key={idx}
-                                onClick={() => window.open(downloadUrl, '_blank')}
-                                className="w-full p-3 rounded-lg bg-white/5 border border-white/10 hover:bg-blue-500/10 hover:border-blue-500/30 transition-all text-left group"
+                                href={downloadUrl}
+                                download={fileName}
+                                className="block w-full p-3 rounded-lg bg-white/5 border border-white/10 hover:bg-blue-500/10 hover:border-blue-500/30 transition-all group"
                               >
                                 <div className="flex items-center justify-between">
                                   <span className="text-sm text-white/80 truncate flex-1">{fileName}</span>
                                   <ChevronRight className="w-4 h-4 text-white/40 group-hover:text-blue-400 transition-all" />
                                 </div>
                                 <span className="text-xs text-white/40 mt-1 block">Click to download</span>
-                              </button>
+                              </a>
                             );
                           })}
                         </div>

src/orchestrator.py

@@ -1099,16 +1099,13 @@ You are a DOER. Complete workflows based on user intent."""
                     "url": f"/outputs/{nested_result['output_path'].replace('./outputs/', '')}"
                 })
         
-        # Build enhanced text summary - start with metrics then LLM explanation
-        summary_lines = [
-            f"## 📊 Analysis Complete",
-            ""
-        ]
+        # Build STRICT response template to prevent malformed output
+        summary_lines = []
         
-        # Show all baseline models comparison FIRST (before LLM summary)
+        # SECTION 1: Model Performance (if models were trained)
         if "all_models" in metrics and metrics["all_models"]:
             summary_lines.extend([
-                "### 🔬 Baseline Models Comparison",
+                "## 🎯 Model Performance",
                 ""
             ])
             
@@ -1124,49 +1121,70 @@ You are a DOER. Complete workflows based on user intent."""
                 rmse = model_metrics.get("rmse", 0)
                 mae = model_metrics.get("mae", 0)
                 
-                # Highlight the best model with emoji
+                # Highlight the best model
                 is_best = (
                     "best_model" in metrics and 
                     metrics["best_model"].get("name", "") == model_name
                 )
-                prefix = "🏆 " if is_best else "  "
+                prefix = "🏆 " if is_best else "📊 "
                 
                 summary_lines.append(
                     f"{prefix}**{model_name.replace('_', ' ').title()}**: "
-                    f"R²={r2:.4f}, RMSE={rmse:.4f}, MAE={mae:.4f}"
+                    f"R²={r2:.4f} | RMSE={rmse:.4f} | MAE={mae:.4f}"
                 )
             
-            summary_lines.append("")
+            summary_lines.extend(["", ""])
         
-        # Show tuned model separately if hyperparameter tuning was done
+        # SECTION 2: Tuning Results (if hyperparameter tuning was done)
         if "tuned_model" in metrics:
             tuned = metrics["tuned_model"]
             summary_lines.extend([
-                "### ⚙️ Hyperparameter Tuning Results",
-                f"- **Model Type**: {tuned.get('model_type', 'N/A')}",
-                f"- **Optimized Score**: {tuned.get('best_score', 0):.4f}",
+                "## ⚙️ Hyperparameter Tuning",
+                "",
+                f"**Model**: {tuned.get('model_type', 'Unknown').title()}",
+                f"**Optimized Score**: {tuned.get('best_score', 0):.4f}",
+                "",
                 ""
             ])
         
+        # SECTION 3: Cross-Validation (if performed)
         if "cross_validation" in metrics:
             cv = metrics["cross_validation"]
             summary_lines.extend([
-                "### ✅ Cross-Validation Results",
-                f"- **Mean Score**: {cv['mean_score']:.4f} (± {cv['std_score']:.4f})",
-                ""
-            ])
-        
-        # Add LLM's explanation after metrics
-        if llm_summary and llm_summary.strip():
-            summary_lines.extend([
-                "---",
+                "## ✅ Cross-Validation",
                 "",
-                "### 📝 Analysis Summary",
+                f"**Mean Score**: {cv['mean_score']:.4f} ± {cv['std_score']:.4f}",
                 "",
-                llm_summary,
                 ""
             ])
         
+        # SECTION 4: Analysis Summary (LLM explanation - cleaned)
+        if llm_summary and llm_summary.strip():
+            # Clean LLM summary aggressively
+            cleaned_summary = llm_summary
+            # Remove all file path patterns
+            import re
+            cleaned_summary = re.sub(r'\./outputs/[^\s\)\]]+', '', cleaned_summary)
+            cleaned_summary = re.sub(r'/outputs/[^\s\)\]]+', '', cleaned_summary)
+            cleaned_summary = re.sub(r'`[^`]*\.(csv|pkl|html|png|json)[^`]*`', '', cleaned_summary)
+            cleaned_summary = re.sub(r'\([^\)]*\.(csv|pkl|html|png|json)[^\)]*\)', '', cleaned_summary)
+            cleaned_summary = re.sub(r'Printed in logs.*?\)', '', cleaned_summary, flags=re.IGNORECASE)
+            cleaned_summary = re.sub(r'\(see above\)', '', cleaned_summary, flags=re.IGNORECASE)
+            cleaned_summary = re.sub(r'see above', '', cleaned_summary, flags=re.IGNORECASE)
+            # Remove broken tables
+            cleaned_summary = re.sub(r'^\s*\|\s*\|\s*$', '', cleaned_summary, flags=re.MULTILINE)
+            cleaned_summary = re.sub(r'^\s*[-|]+\s*$', '', cleaned_summary, flags=re.MULTILINE)
+            cleaned_summary = re.sub(r'\n{3,}', '\n\n', cleaned_summary)
+            cleaned_summary = cleaned_summary.strip()
+            
+            if cleaned_summary:
+                summary_lines.extend([
+                    "## 📝 Workflow Summary",
+                    "",
+                    cleaned_summary,
+                    ""
+                ])
+        
         # Add artifact links
         if artifacts["models"]:
             summary_lines.append("### �� Trained Models")