feat: Update core agent logic, code execution utilities, and LLM configuration.

utils/code_executor.py

@@ -35,6 +35,7 @@ class CodeExecutor:
         "duckdb",
         "scipy",
         "sklearn",
+        "sklearn.feature_extraction.text",
         "statsmodels",
         "plotly",
         "dash",

utils/extract_code.py

@@ -29,6 +29,22 @@ def extract_code_from_response(response: str) -> Optional[str]:
             end = response.find('```', start)
             if end != -1:
                 return response[start:end].strip()
+        
+        # 尝试提取 code: | 形式的代码块（针对YAML格式错误但结构清晰的情况）
+        import re
+        # 匹配 code: | 后面的内容，直到遇到下一个键（next_key:）或结尾
+        # 假设代码块至少缩进2个空格
+        pattern = r'code:\s*\|\s*\n((?: {2,}.*\n?)+)'
+        match = re.search(pattern, response)
+        if match:
+            code_block = match.group(1)
+            # 尝试去除公共缩进
+            try:
+                import textwrap
+                return textwrap.dedent(code_block).strip()
+            except:
+                return code_block.strip()
+
         elif '```' in response:
             start = response.find('```') + 3
             end = response.find('```', start)

utils/fallback_openai_client.py

@@ -97,23 +97,48 @@ class AsyncFallbackOpenAIClient:
                     print(f"❌ {api_name} API 在达到最大重试次数后仍然失败。")
             except APIStatusError as e: # API 返回的特定状态码错误
                 is_content_filter_error = False
-                if e.status_code == 400:
-                    try:
-                        error_json = e.response.json()
-                        error_details = error_json.get("error", {})
-                        if (error_details.get("code") == self.content_filter_error_code and
-                            self.content_filter_error_field in error_json):
-                            is_content_filter_error = True
-                    except Exception:
-                        pass # 解析错误响应失败，不认为是内容过滤错误
+                retry_after = None
+                
+                # 尝试解析错误详情以获取更多信息（如 Google RPC RetryInfo）
+                try:
+                    error_json = e.response.json()
+                    error_details = error_json.get("error", {})
+                    
+                    # 检查内容过滤错误（针对特定服务商）
+                    if (error_details.get("code") == self.content_filter_error_code and
+                        self.content_filter_error_field in error_json):
+                        is_content_filter_error = True
+                    
+                    # 检查 Google RPC RetryInfo
+                    # 格式示例: {'error': {'details': [{'@type': 'type.googleapis.com/google.rpc.RetryInfo', 'retryDelay': '38s'}]}}
+                    if "details" in error_details:
+                        for detail in error_details["details"]:
+                            if detail.get("@type") == "type.googleapis.com/google.rpc.RetryInfo":
+                                delay_str = detail.get("retryDelay", "")
+                                if delay_str.endswith("s"):
+                                    try:
+                                        retry_after = float(delay_str[:-1])
+                                        print(f"⏳ 收到服务器 RetryInfo，等待时间: {retry_after}秒")
+                                    except ValueError:
+                                        pass
+                except Exception:
+                    pass # 解析错误响应失败，忽略
                 
                 if is_content_filter_error and api_name == "主": # 如果是主 API 的内容过滤错误，则直接抛出以便回退
                     raise e 
                 
                 last_exception = e
                 print(f"⚠️ {api_name} API 调用时发生 APIStatusError ({e.status_code}): {e}. 尝试次数 {attempt + 1}/{max_retries + 1}")
+                
                 if attempt < max_retries:
-                    await asyncio.sleep(self.retry_delay_seconds * (attempt + 1))
+                    # 如果获取到了明确的 retry_after，则使用它；否则使用默认的指数退避
+                    wait_time = retry_after if retry_after is not None else (self.retry_delay_seconds * (attempt + 1))
+                    # 如果是 429 Too Many Requests 且没有解析出 retry_after，建议加大等待时间
+                    if e.status_code == 429 and retry_after is None:
+                        wait_time = max(wait_time, 5.0 * (attempt + 1)) # 429 默认至少等 5 秒
+                        
+                    print(f"💤 将等待 {wait_time:.2f} 秒后重试...")
+                    await asyncio.sleep(wait_time)
                 else:
                     print(f"❌ {api_name} API 在达到最大重试次数后仍然失败 (APIStatusError)。")
             except APIError as e: # 其他不可轻易重试的 OpenAI 错误