Update app.py

app.py

@@ -40,36 +40,67 @@ def create_reranking_interface(task_data):
                 </div>
                 '''
         html += '</div>'
-        return html
+        
+        # Also return the computed order for proper initialization
+        return html, order
     
     def save_ranking(order_json, sample_id):
         """Save the current ranking to results."""
         try:
             if not order_json or order_json == "[]":
                 return "⚠️ Drag documents to set the ranking before submitting.", progress_text.value
+            
             order = json.loads(order_json)
-            num_candidates = len(next(s["candidates"] for s in samples if s["id"] == sample_id))
+            sample = next((s for s in samples if s["id"] == sample_id), None)
+            
+            if not sample:
+                return "⚠️ Sample not found.", progress_text.value
+                
+            num_candidates = len(sample["candidates"])
+            
             if len(order) != num_candidates:
                 return f"⚠️ Ranking order length mismatch. Expected {num_candidates}, got {len(order)}.", progress_text.value
+            
             rankings = [0] * num_candidates
-            for rank_minus_1, doc_idx in enumerate(order):
-                if doc_idx < num_candidates:
-                    rankings[doc_idx] = rank_minus_1 + 1
-                else:
-                    raise ValueError(f"Invalid document index {doc_idx} found in order.")
+            try:
+                for rank_minus_1, doc_idx in enumerate(order):
+                    if doc_idx < num_candidates:
+                        rankings[doc_idx] = rank_minus_1 + 1
+                    else:
+                        raise ValueError(f"Invalid document index {doc_idx} found in order.")
+            except Exception as e:
+                return f"⚠️ Error processing ranking order: {str(e)}", progress_text.value
+                
             if sorted(rankings) != list(range(1, num_candidates + 1)):
                 return "⚠️ Ranking validation failed. Ranks are not 1 to N.", progress_text.value
+                
             annotation = {"sample_id": sample_id, "rankings": rankings}
+            
+            # Check if this sample was already annotated
             existing_idx = next((i for i, a in enumerate(results["annotations"]) if a["sample_id"] == sample_id), None)
             if existing_idx is not None:
                 results["annotations"][existing_idx] = annotation
             else:
                 results["annotations"].append(annotation)
+                
             completed_samples[sample_id] = True
-            output_path = f"{task_data['task_name']}_human_results.json"
-            with open(output_path, "w") as f:
-                json.dump(results, f, indent=2)
-            return f"✅ Rankings saved successfully ({len(results['annotations'])}/{len(samples)} completed)", f"Progress: {sum(completed_samples.values())}/{len(samples)}"
+            
+            # Save results with timestamp and better error handling
+            try:
+                output_path = f"{task_data['task_name']}_human_results.json"
+                with open(output_path, "w") as f:
+                    json.dump(results, f, indent=2)
+                    
+                # Check if all samples are complete
+                all_completed = sum(completed_samples.values()) == len(samples)
+                completion_message = "🎉 All samples completed! You can save and submit your results." if all_completed else ""
+                
+                return f"✅ Rankings saved successfully ({len(results['annotations'])}/{len(samples)} completed) {completion_message}", f"Progress: {sum(completed_samples.values())}/{len(samples)}"
+            except Exception as file_error:
+                print(f"Error saving file: {str(file_error)}")
+                # Still mark as completed in memory even if file save fails
+                return f"⚠️ Rankings recorded but file save failed: {str(file_error)}", f"Progress: {sum(completed_samples.values())}/{len(samples)}"
+                
         except json.JSONDecodeError:
             return "⚠️ Error decoding ranking order. Please try again.", progress_text.value
         except Exception as e:
@@ -83,13 +114,23 @@ def create_reranking_interface(task_data):
             sample = next((s for s in samples if s["id"] == sample_id), None)
             if not sample:
                 return gr.update(), gr.update(value="[]"), gr.update(), gr.update()
+            
             existing_ranking = next((anno["rankings"] for anno in results["annotations"] if anno["sample_id"] == sample_id), None)
-            new_html = generate_sortable_html(sample["candidates"], existing_ranking)
+            
+            # Get both the HTML and the initial order
+            new_html, initial_order = generate_sortable_html(sample["candidates"], existing_ranking)
+            
+            # Convert initial order to JSON string for state
+            initial_order_json = json.dumps(initial_order)
+            
             status = "Ready to rank" if not completed_samples.get(sample_id, False) else "Already ranked"
             progress = f"Progress: {sum(completed_samples.values())}/{len(samples)}"
-            return sample["query"], new_html, "[]", progress, status
+            
+            return sample["query"], new_html, initial_order_json, progress, status
         except Exception as e:
-            return gr.update(), gr.update(value="[]"), gr.update(), gr.update(value=f"Error loading sample: {str(e)}")
+            import traceback
+            print(traceback.format_exc())
+            return gr.update(), gr.update(), "[]", gr.update(), f"Error loading sample: {str(e)}"
     
     def next_sample_id(current_id):
         current_idx = next((i for i, s in enumerate(samples) if s["id"] == current_id), -1)
@@ -108,9 +149,20 @@ def create_reranking_interface(task_data):
     def save_results():
         output_path = f"{task_data['task_name']}_human_results.json"
         try:
+            # Create backup with timestamp
+            from datetime import datetime
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            backup_path = f"{task_data['task_name']}_results_{timestamp}.json"
+            
+            # First create a backup
+            with open(backup_path, "w") as f:
+                json.dump(results, f, indent=2)
+                
+            # Then save to the main file
             with open(output_path, "w") as f:
                 json.dump(results, f, indent=2)
-            return f"✅ Results saved to {output_path} ({len(results['annotations'])} annotations)"
+                
+            return f"✅ Results saved to {output_path} ({len(results['annotations'])} annotations)\nBackup created at {backup_path}"
         except Exception as e:
             return f"⚠️ Error saving results file: {str(e)}"
     
@@ -428,18 +480,27 @@ def create_reranking_interface(task_data):
             line-height: 1.5;
             word-break: break-word;
         }
-        .rank-bg-1 .rank-badge { background-color: #198754; }
-        .rank-bg-2 .rank-badge { background-color: #20c997; }
-        .rank-bg-3 .rank-badge { background-color: #ffc107; color: #333; }
-        .rank-bg-4 .rank-badge { background-color: #fd7e14; }
-        .rank-bg-5 .rank-badge { background-color: #dc3545; }
-        .rank-bg-6 .rank-badge, .rank-bg-7 .rank-badge { background-color: #6f42c1; }
-        .rank-bg-8 .rank-badge, .rank-bg-9 .rank-badge { background-color: #d63384; }
-        .rank-bg-10 .rank-badge, .rank-bg-11 .rank-badge, .rank-bg-12 .rank-badge, 
-        .rank-bg-13 .rank-badge, .rank-bg-14 .rank-badge, .rank-bg-15 .rank-badge, 
-        .rank-bg-16 .rank-badge, .rank-bg-17 .rank-badge, .rank-bg-18 .rank-badge, 
-        .rank-bg-19 .rank-badge, .rank-bg-20 .rank-badge { 
-            background-color: #6c757d;
+        /* More professional color scheme for rank badges */
+        .rank-bg-1 .rank-badge { background-color: #1e40af; } /* Deep blue for top rank */
+        .rank-bg-2 .rank-badge { background-color: #3b82f6; } /* Medium blue */
+        .rank-bg-3 .rank-badge { background-color: #60a5fa; } /* Light blue */
+        .rank-bg-4 .rank-badge { background-color: #93c5fd; color: #1e3a8a; } /* Very light blue with dark text */
+        .rank-bg-5 .rank-badge { background-color: #bfdbfe; color: #1e3a8a; } /* Lightest blue with dark text */
+        
+        /* Lower ranks get progressively more gray */
+        .rank-bg-6 .rank-badge, .rank-bg-7 .rank-badge { 
+            background-color: #64748b; 
+        }
+        .rank-bg-8 .rank-badge, .rank-bg-9 .rank-badge, .rank-bg-10 .rank-badge { 
+            background-color: #94a3b8; 
+            color: #0f172a;
+        }
+        .rank-bg-11 .rank-badge, .rank-bg-12 .rank-badge, .rank-bg-13 .rank-badge, 
+        .rank-bg-14 .rank-badge, .rank-bg-15 .rank-badge, .rank-bg-16 .rank-badge, 
+        .rank-bg-17 .rank-badge, .rank-bg-18 .rank-badge, .rank-bg-19 .rank-badge, 
+        .rank-bg-20 .rank-badge { 
+            background-color: #cbd5e1;
+            color: #0f172a;
         }
         </style>
         """