diff --git a/backend/app/services/llm/service.py b/backend/app/services/llm/service.py index c13ab87..0d7b913 100644 --- a/backend/app/services/llm/service.py +++ b/backend/app/services/llm/service.py @@ -333,16 +333,13 @@ Please analyze the following code: # 尝试从响应中提取JSON result = self._parse_json(content) + # 验证和清理结果 + result = self._validate_analysis_result(result) + # 记录解析后的问题数量 issues_count = len(result.get("issues", [])) logger.info(f"📊 LLM 分析结果: 发现 {issues_count} 个问题, 质量评分: {result.get('quality_score', 'N/A')}") - # 检查解析结果是否有效（不是默认响应） - if result == self._get_default_response(): - error_msg = f"无法解析LLM响应为有效的分析结果 - Provider: {self.config.provider.value}" - logger.error(error_msg) - raise Exception(error_msg) - return result except Exception as e: @@ -763,6 +760,54 @@ Please analyze the following code: raise ValueError(f"json-repair returned unexpected type: {type(repaired)}") + def _validate_analysis_result(self, result: Dict[str, Any]) -> Dict[str, Any]: + """验证和清理分析结果""" + if not isinstance(result, dict): + logger.warning(f"分析结果不是字典类型: {type(result)}") + return self._get_default_response() + + # 确保 issues 是列表，且每个元素都是字典 + raw_issues = result.get("issues", []) + if not isinstance(raw_issues, list): + logger.warning(f"issues 字段不是列表类型: {type(raw_issues)}") + raw_issues = [] + + valid_issues = [] + for i, issue in enumerate(raw_issues): + if isinstance(issue, dict): + valid_issues.append(issue) + else: + logger.warning(f"忽略无效的问题格式 (index {i}): {type(issue)}") + + result["issues"] = valid_issues + + # 确保 quality_score 是数字 + score = result.get("quality_score") + if score is None or not isinstance(score, (int, float)): + try: + if score is not None: + result["quality_score"] = int(float(str(score))) + else: + result["quality_score"] = 80 + except (ValueError, TypeError): + result["quality_score"] = 80 + + # 确保 summary 和 metrics 存在 + if "summary" not in result or not isinstance(result["summary"], dict): + num_issues = len(valid_issues) + result["summary"] = { + "total_issues": num_issues, + "critical_issues": sum(1 for iss in valid_issues if iss.get("severity") == "critical"), + "high_issues": sum(1 for iss in valid_issues if iss.get("severity") == "high"), + "medium_issues": sum(1 for iss in valid_issues if iss.get("severity") == "medium"), + "low_issues": sum(1 for iss in valid_issues if iss.get("severity") == "low") + } + + if "metrics" not in result or not isinstance(result["metrics"], dict): + result["metrics"] = self._get_default_response()["metrics"] + + return result + def _get_default_response(self) -> Dict[str, Any]: """返回默认响应""" return { @@ -915,6 +960,8 @@ Please analyze the following code: raise Exception("LLM返回空响应") result = self._parse_json(content) + # 验证和清理结果 + result = self._validate_analysis_result(result) return result except Exception as e: diff --git a/backend/app/services/scanner.py b/backend/app/services/scanner.py index 3c1c041..eef9487 100644 --- a/backend/app/services/scanner.py +++ b/backend/app/services/scanner.py @@ -680,36 +680,49 @@ async def scan_repo_task(task_id: str, db_session_factory, user_config: dict = N # 保存问题 issues = analysis.get("issues", []) for issue in issues: - line_num = issue.get("line", 1) - code_snippet = issue.get("code_snippet") - if not code_snippet or len(code_snippet.strip()) < 5: - try: - idx = max(0, int(line_num) - 1) - start = max(0, idx - 2) - end = min(len(file_lines), idx + 3) - code_snippet = '\n'.join(file_lines[start:end]) - except Exception: - code_snippet = "" + try: + # 防御性检查：确保 issue 是字典 + if not isinstance(issue, dict): + print(f"⚠️ 警告: 任务 {task_id} 中文件 {f_path} 的分析结果包含无效的问题格式: {issue}") + continue - audit_issue = AuditIssue( - task_id=task.id, - file_path=f_path, - line_number=line_num, - column_number=issue.get("column"), - issue_type=issue.get("type", "maintainability"), - severity=issue.get("severity", "low"), - title=issue.get("title", "Issue"), - message=issue.get("description") or issue.get("title", "Issue"), - suggestion=issue.get("suggestion"), - code_snippet=code_snippet, - ai_explanation=issue.get("ai_explanation"), - status="open" - ) - db.add(audit_issue) - total_issues += 1 + line_num = issue.get("line", 1) + code_snippet = issue.get("code_snippet") + if not code_snippet or len(code_snippet.strip()) < 5: + try: + idx = max(0, int(line_num) - 1) + start = max(0, idx - 2) + end = min(len(file_lines), idx + 3) + code_snippet = '\n'.join(file_lines[start:end]) + except Exception: + code_snippet = "" + + audit_issue = AuditIssue( + task_id=task.id, + file_path=f_path, + line_number=line_num, + column_number=issue.get("column"), + issue_type=issue.get("type", "maintainability"), + severity=issue.get("severity", "low"), + title=issue.get("title", "Issue"), + message=issue.get("description") or issue.get("title", "Issue"), + suggestion=issue.get("suggestion"), + code_snippet=code_snippet, + ai_explanation=issue.get("ai_explanation"), + status="open" + ) + db.add(audit_issue) + total_issues += 1 + except Exception as e: + print(f"⚠️ 处理单个问题时出错 (文件 {f_path}): {e}") + continue if "quality_score" in analysis: - quality_scores.append(analysis["quality_score"]) + try: + quality_score = float(analysis["quality_score"]) + quality_scores.append(quality_score) + except (ValueError, TypeError): + pass # 更新主任务进度 processed_count = scanned_files + failed_files