feat: 实现分离式提交评测系统

新功能:
🏗️ 系统架构
- 用户提交 → SAGE-Bench收集 → 本地评测 → 更新排行榜
- 保护API密钥，防止在线仓库暴露评测代码

📁 新增文件
- src/submission/simple_submit.py: 文件收集模式的提交处理
- submissions/: 用户提交文件收集目录

🔧 功能特点
- 只验证和保存JSON文件，不进行在线评测
- 用户提交后立即收到确认信息
- 支持组织信息和邮箱验证
- 完整的提交格式验证

🎯 下一步
- 使用local_eval/目录进行本地评测
- 运行update_leaderboard.py更新排行榜

app.py

@@ -191,10 +191,17 @@ with demo:
         submit_button = gr.Button("Submit Results", variant="primary")
         submission_result = gr.HTML()
         
-        # Simplified submission handling
+        # File collection submission handling
+        def handle_submission(file_upload, org_name, email):
+            try:
+                from src.submission.simple_submit import process_sage_submission_simple
+                return process_sage_submission_simple(file_upload, org_name, email)
+            except ImportError:
+                return format_error("❌ 提交系统暂时不可用，请稍后再试。")
+        
         submit_button.click(
-            lambda: format_warning("📋 Submission feature coming soon! For now, please contact administrators directly."),
-            inputs=[],
+            handle_submission,
+            inputs=[file_upload, org_textbox, email_textbox],
             outputs=[submission_result]
         )
 

src/submission/simple_submit.py

@@ -0,0 +1,189 @@
+#!/usr/bin/env python3
+"""
+简化的SAGE提交处理 - 文件收集模式
+只负责接收和验证提交文件，不进行评测
+"""
+
+import json
+import os
+import shutil
+from datetime import datetime
+from typing import Dict, Any
+
+def format_error(msg):
+    return f"<p style='color: red; font-size: 16px;'>{msg}</p>"
+
+def format_success(msg):
+    return f"<p style='color: green; font-size: 16px;'>{msg}</p>"
+
+def format_warning(msg):
+    return f"<p style='color: orange; font-size: 16px;'>{msg}</p>"
+
+def validate_sage_submission(submission_data: Dict[str, Any]) -> tuple[bool, str]:
+    """验证SAGE基准提交格式"""
+    
+    # 检查必需的顶级字段
+    required_fields = ["submission_org", "submission_email", "predictions"]
+    for field in required_fields:
+        if field not in submission_data:
+            return False, f"缺少必需字段: {field}"
+    
+    # 验证邮箱格式（基本验证）
+    email = submission_data["submission_email"]
+    if "@" not in email or "." not in email:
+        return False, "邮箱格式无效"
+    
+    # 验证predictions
+    predictions = submission_data["predictions"]
+    if not isinstance(predictions, list) or len(predictions) == 0:
+        return False, "predictions必须是非空列表"
+    
+    for i, prediction in enumerate(predictions):
+        # 检查必需的prediction字段
+        pred_required_fields = ["original_question_id", "content", "reasoning_content"]
+        for field in pred_required_fields:
+            if field not in prediction:
+                return False, f"预测{i}中缺少字段: {field}"
+        
+        # 验证content数组
+        content = prediction["content"]
+        reasoning_content = prediction["reasoning_content"]
+        
+        if not isinstance(content, list) or len(content) != 4:
+            return False, f"预测{i}的content必须是包含4个项目的列表"
+        
+        if not isinstance(reasoning_content, list) or len(reasoning_content) != 4:
+            return False, f"预测{i}的reasoning_content必须是包含4个项目的列表"
+        
+        # 验证question ID
+        if not isinstance(prediction["original_question_id"], int):
+            return False, f"预测{i}的question ID必须是整数"
+    
+    return True, "提交格式有效"
+
+def save_submission_file(submission_data: Dict[str, Any], submissions_dir: str = "./submissions") -> str:
+    """保存提交文件到指定目录"""
+    
+    # 确保submissions目录存在
+    os.makedirs(submissions_dir, exist_ok=True)
+    
+    # 生成文件名
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    org_name = submission_data["submission_org"].replace(" ", "_").replace("/", "_").replace("\\", "_")
+    filename = f"submission_{org_name}_{timestamp}.json"
+    
+    # 完整文件路径
+    file_path = os.path.join(submissions_dir, filename)
+    
+    # 保存文件
+    with open(file_path, 'w', encoding='utf-8') as f:
+        json.dump(submission_data, f, indent=2, ensure_ascii=False)
+    
+    return file_path
+
+def process_sage_submission_simple(submission_file, org_name=None, email=None) -> str:
+    """
+    处理SAGE基准提交文件 - 文件收集模式
+    只负责验证和保存，不进行评测
+    """
+    
+    try:
+        # 读取提交的文件
+        if submission_file is None:
+            return format_error("❌ 没有上传文件。请选择一个JSON文件。")
+        
+        # submission_file是文件路径字符串
+        try:
+            with open(submission_file, 'r', encoding='utf-8') as f:
+                content = f.read()
+        except Exception as e:
+            return format_error(f"❌ 读取文件时出错: {str(e)}")
+        
+        # 解析JSON
+        try:
+            submission_data = json.loads(content)
+        except json.JSONDecodeError as e:
+            return format_error(f"❌ JSON格式无效: {str(e)}")
+        
+        # 如果表单提供了组织名和邮箱，使用表单数据
+        if org_name and email:
+            submission_data["submission_org"] = org_name.strip()
+            submission_data["submission_email"] = email.strip()
+        
+        # 验证提交格式
+        is_valid, message = validate_sage_submission(submission_data)
+        if not is_valid:
+            return format_error(f"❌ 提交验证失败: {message}")
+        
+        # 保存提交文件
+        try:
+            saved_path = save_submission_file(submission_data)
+            print(f"✅ 提交文件已保存到: {saved_path}")
+            
+            # 生成成功消息
+            org = submission_data["submission_org"]
+            email_addr = submission_data["submission_email"]
+            num_predictions = len(submission_data["predictions"])
+            
+            success_msg = format_success(f"""
+🎉 <strong>提交成功接收！</strong><br><br>
+📋 <strong>提交信息:</strong><br>
+• 组织: {org}<br>
+• 邮箱: {email_addr}<br>
+• 预测数量: {num_predictions} 个问题<br><br>
+⏳ <strong>下一步:</strong><br>
+您的提交正在排队等待评测。我们的系统将使用LLM-as-Judge对您的答案进行评估，<br>
+并计算各个科学领域的分数。评测完成后，结果将自动添加到排行榜中。<br><br>
+📧 评测完成后，我们会通过邮箱通知您结果。<br>
+🕐 预计评测时间: 1-24小时（取决于队列长度）<br><br>
+感谢您参与SAGE基准测试！🧪
+            """)
+            
+            return success_msg
+            
+        except Exception as e:
+            return format_error(f"❌ 保存提交文件时出错: {str(e)}")
+        
+    except Exception as e:
+        return format_error(f"❌ 提交处理失败: {str(e)}")
+
+def get_submission_stats(submissions_dir: str = "./submissions") -> Dict[str, Any]:
+    """获取提交统计信息"""
+    
+    if not os.path.exists(submissions_dir):
+        return {"total": 0, "recent": []}
+    
+    submissions = []
+    
+    for filename in os.listdir(submissions_dir):
+        if filename.startswith("submission_") and filename.endswith(".json"):
+            file_path = os.path.join(submissions_dir, filename)
+            try:
+                with open(file_path, 'r', encoding='utf-8') as f:
+                    data = json.load(f)
+                
+                # 提取信息
+                timestamp_str = filename.split("_")[-1].replace(".json", "")
+                try:
+                    timestamp = datetime.strptime(timestamp_str, "%Y%m%d_%H%M%S")
+                    formatted_time = timestamp.strftime("%Y-%m-%d %H:%M")
+                except:
+                    formatted_time = timestamp_str
+                
+                submissions.append({
+                    "org": data.get("submission_org", "Unknown"),
+                    "email": data.get("submission_email", ""),
+                    "time": formatted_time,
+                    "predictions": len(data.get("predictions", []))
+                })
+                
+            except Exception:
+                continue
+    
+    # 按时间排序，最新的在前
+    submissions.sort(key=lambda x: x["time"], reverse=True)
+    
+    return {
+        "total": len(submissions),
+        "recent": submissions[:10]  # 最近10个
+    }