#!/usr/bin/env python3 """ 质量验证器：10项质量检查（高级版） 核心算法：基于多维度统计分析和加权评分的质量评估系统 """ import json import re from pathlib import Path from typing import Dict, List, Tuple from collections import Counter import math class QualityValidator: """高级质量验证器""" # 10项质量检查（带详细权重和阈值） QUALITY_CHECKS = { 'mental_model_count': { 'weight': 15, 'description': '心智模型数量', 'pass_criteria': '3-7个', 'pass_score': 15, 'partial_score': 10, 'fail_score': 5, 'algorithm': 'count_and_classify' }, 'model_limitations': { 'weight': 10, 'description': '每个心智模型都有局限性说明', 'pass_criteria': '100%覆盖率', 'pass_score': 10, 'partial_score': 5, 'fail_score': 0, 'algorithm': 'ratio_analysis' }, 'model_evidence': { 'weight': 10, 'description': '每个心智模型有≥2个证据', 'pass_criteria': '100%满足', 'pass_score': 10, 'partial_score': 7, 'fail_score': 3, 'algorithm': 'evidence_counting' }, 'expression_dna': { 'weight': 10, 'description': '表达DNA特征≥3个', 'pass_criteria': '≥3个特征', 'pass_score': 10, 'partial_score': 6, 'fail_score': 2, 'algorithm': 'feature_extraction' }, 'honest_boundaries': { 'weight': 15, 'description': '诚实边界≥3条', 'pass_criteria': '≥3条边界', 'pass_score': 15, 'partial_score': 10, 'fail_score': 5, 'algorithm': 'boundary_detection' }, 'internal_tensions': { 'weight': 10, 'description': '内在张力≥2处', 'pass_criteria': '≥2处张力', 'pass_score': 10, 'partial_score': 6, 'fail_score': 2, 'algorithm': 'tension_identification' }, 'decision_heuristics': { 'weight': 10, 'description': '决策启发式≥5条', 'pass_criteria': '≥5条启发式', 'pass_score': 10, 'partial_score': 6, 'fail_score': 2, 'algorithm': 'pattern_matching' }, 'agentic_protocol': { 'weight': 10, 'description': '包含3步骤的Agentic Protocol', 'pass_criteria': '包含完整3步骤', 'pass_score': 10, 'partial_score': 5, 'fail_score': 0, 'algorithm': 'protocol_validation' }, 'toolkit': { 'weight': 5, 'description': '工具箱≥3个', 'pass_criteria': '≥3个工具', 'pass_score': 5, 'partial_score': 3, 'fail_score': 1, 'algorithm': 'tool_extraction' }, 'primary_sources': { 'weight': 5, 'description': '一手来源>50%（如有research）', 'pass_criteria': '>50%', 'pass_score': 5, 'partial_score': 3, 'fail_score': 1, 'algorithm': 'source_classification' } } # 通过线（总分100分，需≥75分） PASSING_SCORE = 75 def __init__(self): self.validation_result = {} def validate_skill(self, skill_file: str, research_dir: str = None) -> Dict: """验证HumanOS Skill - 主流程""" skill_path = Path(skill_file) # 读取Skill文件 with open(skill_path, 'r', encoding='utf-8') as f: skill_content = f.read() # 执行10项检查 check_results = {} # 检查1: 心智模型数量 check_results['mental_model_count'] = self._check_mental_model_count_advanced( skill_content ) # 检查2: 模型局限性 check_results['model_limitations'] = self._check_model_limitations_advanced( skill_content ) # 检查3: 模型证据 check_results['model_evidence'] = self._check_model_evidence_advanced( skill_content ) # 检查4: 表达DNA check_results['expression_dna'] = self._check_expression_dna_advanced( skill_content ) # 检查5: 诚实边界 check_results['honest_boundaries'] = self._check_honest_boundaries_advanced( skill_content ) # 检查6: 内在张力 check_results['internal_tensions'] = self._check_internal_tensions_advanced( skill_content ) # 检查7: 决策启发式 check_results['decision_heuristics'] = self._check_decision_heuristics_advanced( skill_content ) # 检查8: Agentic Protocol check_results['agentic_protocol'] = self._check_agentic_protocol_advanced( skill_content ) # 检查9: 工具箱 check_results['toolkit'] = self._check_toolkit_advanced( skill_content ) # 检查10: 一手来源 check_results['primary_sources'] = self._check_primary_sources_advanced( skill_content, research_dir ) # 计算总分（加权） total_score = 0 for check_name, result in check_results.items(): weight = self.QUALITY_CHECKS[check_name]['weight'] score = result['score'] # 归一化得分：score / pass_score * weight normalized_score = (score / self.QUALITY_CHECKS[check_name]['pass_score']) * weight result['normalized_score'] = round(normalized_score, 2) total_score += normalized_score total_score = round(total_score, 1) # 判断是否通过 passed = total_score >= self.PASSING_SCORE # 生成详细报告 validation_result = { 'skill_file': skill_file, 'total_score': total_score, 'passing_score': self.PASSING_SCORE, 'passed': passed, 'passing_rate': f"{total_score / 100 * 100:.1f}%", 'check_results': check_results, 'summary': self._generate_summary_advanced(check_results, total_score), 'recommendations': self._generate_recommendations_advanced(check_results), 'quality_grade': self._calculate_quality_grade(total_score), 'analysis': self._generate_detailed_analysis(check_results) } self.validation_result = validation_result return validation_result def _check_mental_model_count_advanced(self, skill_content: str) -> Dict: """高级心智模型数量检查""" # 识别心智模型的多种模式 patterns = [ r'\*\*([^*]+)\*\*.*?(?:思维模型|原理|框架|效应|法则)', r'##\s+(?:心智模型|Mental Models).*?\n(.*?)(?=\n##|\Z)', r'模型\d+\s*[:：]\s*\*\*([^*]+)\*\*', r'([^.\n]+)(?:思维模型|原理|框架)' ] models = set() for pattern in patterns: matches = re.findall(pattern, skill_content, re.MULTILINE | re.DOTALL) for match in matches: if isinstance(match, tuple): model_name = match[0].strip() else: model_name = match.strip() # 过滤无效模型 if len(model_name) >= 2 and not model_name.isdigit(): models.add(model_name) model_list = sorted(list(models)) model_count = len(model_list) # 评分 if 3 <= model_count <= 7: score = self.QUALITY_CHECKS['mental_model_count']['pass_score'] status = 'pass' elif model_count >= 1 and model_count < 3: score = self.QUALITY_CHECKS['mental_model_count']['partial_score'] status = 'partial' elif model_count > 7: score = self.QUALITY_CHECKS['mental_model_count']['pass_score'] # 超过也算通过 status = 'pass' else: score = self.QUALITY_CHECKS['mental_model_count']['fail_score'] status = 'fail' return { 'check': 'mental_model_count', 'description': self.QUALITY_CHECKS['mental_model_count']['description'], 'criteria': self.QUALITY_CHECKS['mental_model_count']['pass_criteria'], 'actual': model_count, 'models': model_list[:10], # 返回前10个 'score': score, 'status': status, 'message': f"发现{model_count}个心智模型: {', '.join(model_list[:5])}{'...' if model_count > 5 else ''}", 'algorithm': self.QUALITY_CHECKS['mental_model_count']['algorithm'] } def _check_model_limitations_advanced(self, skill_content: str) -> Dict: """高级模型局限性检查""" # 获取心智模型 model_check = self._check_mental_model_count_advanced(skill_content) models = model_check['models'] model_count = model_check['actual'] if model_count == 0: return { 'check': 'model_limitations', 'description': self.QUALITY_CHECKS['model_limitations']['description'], 'criteria': self.QUALITY_CHECKS['model_limitations']['pass_criteria'], 'actual': "0%", 'score': self.QUALITY_CHECKS['model_limitations']['fail_score'], 'status': 'fail', 'message': '未发现心智模型', 'algorithm': self.QUALITY_CHECKS['model_limitations']['algorithm'] } # 识别局限性模式 limitation_patterns = [ r'(?:局限|不适用|不能|做不到|边界|限制|缺点)', r'局限性\s*[:：]', r'注意\s*[:：].*?(?:局限|限制)' ] found_limitations = set() for pattern in limitation_patterns: matches = re.finditer(pattern, skill_content) for match in matches: # 获取上下文 context_start = max(0, match.start() - 30) context_end = min(len(skill_content), match.end() + 30) context = skill_content[context_start:context_end] found_limitations.add(context) limitation_count = len(found_limitations) limitation_ratio = limitation_count / model_count # 评分 if limitation_ratio >= 1.0: score = self.QUALITY_CHECKS['model_limitations']['pass_score'] status = 'pass' elif limitation_ratio >= 0.5: score = self.QUALITY_CHECKS['model_limitations']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['model_limitations']['fail_score'] status = 'fail' return { 'check': 'model_limitations', 'description': self.QUALITY_CHECKS['model_limitations']['description'], 'criteria': self.QUALITY_CHECKS['model_limitations']['pass_criteria'], 'actual': f"{int(limitation_ratio * 100)}%", 'limitation_count': limitation_count, 'model_count': model_count, 'score': score, 'status': status, 'message': f"发现{limitation_count}处局限性说明，覆盖{int(limitation_ratio * 100)}%模型", 'algorithm': self.QUALITY_CHECKS['model_limitations']['algorithm'] } def _check_model_evidence_advanced(self, skill_content: str) -> Dict: """高级模型证据检查""" # 获取心智模型 model_check = self._check_mental_model_count_advanced(skill_content) models = model_check['models'] model_count = model_check['actual'] if model_count == 0: return { 'check': 'model_evidence', 'description': self.QUALITY_CHECKS['model_evidence']['description'], 'criteria': self.QUALITY_CHECKS['model_evidence']['pass_criteria'], 'actual': "0个证据/模型", 'score': self.QUALITY_CHECKS['model_evidence']['fail_score'], 'status': 'fail', 'message': '未发现心智模型', 'algorithm': self.QUALITY_CHECKS['model_evidence']['algorithm'] } # 识别证据关键词 evidence_keywords = [ '证据', '案例', '实证', '例子', '例如', '比如', '举例', '证明', '显示', '表明', 'example', 'case', 'evidence', 'proof' ] evidence_count = 0 for keyword in evidence_keywords: # 统计关键词出现次数 count = skill_content.lower().count(keyword.lower()) evidence_count += count # 计算证据/模型比率 evidence_ratio = evidence_count / model_count if model_count > 0 else 0 # 评分 if evidence_ratio >= 2.0: score = self.QUALITY_CHECKS['model_evidence']['pass_score'] status = 'pass' elif evidence_ratio >= 1.0: score = self.QUALITY_CHECKS['model_evidence']['pass_score'] # >=1个也算通过 status = 'pass' elif evidence_ratio > 0: score = self.QUALITY_CHECKS['model_evidence']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['model_evidence']['fail_score'] status = 'fail' return { 'check': 'model_evidence', 'description': self.QUALITY_CHECKS['model_evidence']['description'], 'criteria': self.QUALITY_CHECKS['model_evidence']['pass_criteria'], 'actual': f"{evidence_ratio:.1f}个证据/模型", 'evidence_count': evidence_count, 'model_count': model_count, 'score': score, 'status': status, 'message': f"发现{evidence_count}处证据引用，平均{evidence_ratio:.1f}个证据/模型", 'algorithm': self.QUALITY_CHECKS['model_evidence']['algorithm'] } def _check_expression_dna_advanced(self, skill_content: str) -> Dict: """高级表达DNA检查""" dna_features = { 'sentence_patterns': 0, 'vocabulary': 0, 'rhythm': 0, 'style': 0 } # 检查句式模式 if any(kw in skill_content for kw in ['句式', '句子', '句法']): dna_features['sentence_patterns'] = 1 # 检查词汇 if any(kw in skill_content for kw in ['词汇', '术语', '用词']): dna_features['vocabulary'] = 1 # 检查节奏 if any(kw in skill_content for kw in ['节奏', '语速', '停顿']): dna_features['rhythm'] = 1 # 检查风格 if any(kw in skill_content for kw in ['风格', '表达', '语气']): dna_features['style'] = 1 feature_count = sum(dna_features.values()) # 评分 if feature_count >= 3: score = self.QUALITY_CHECKS['expression_dna']['pass_score'] status = 'pass' elif feature_count >= 1: score = self.QUALITY_CHECKS['expression_dna']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['expression_dna']['fail_score'] status = 'fail' return { 'check': 'expression_dna', 'description': self.QUALITY_CHECKS['expression_dna']['description'], 'criteria': self.QUALITY_CHECKS['expression_dna']['pass_criteria'], 'actual': feature_count, 'features': {k: v for k, v in dna_features.items() if v > 0}, 'score': score, 'status': status, 'message': f"发现{feature_count}个表达DNA特征", 'algorithm': self.QUALITY_CHECKS['expression_dna']['algorithm'] } def _check_honest_boundaries_advanced(self, skill_content: str) -> Dict: """高级诚实边界检查""" # 识别边界关键词和模式 boundary_patterns = [ r'(?:边界|不能|做不到|不擅长|局限|限制|无法|难以)', r'诚实边界\s*[:：]', r'注意事项\s*[:：]', r'局限性\s*[:：]' ] found_boundaries = set() for pattern in boundary_patterns: matches = re.finditer(pattern, skill_content) for match in matches: # 获取上下文 context_start = max(0, match.start() - 50) context_end = min(len(skill_content), match.end() + 50) context = skill_content[context_start:context_end] found_boundaries.add(context[:100]) boundary_count = len(found_boundaries) # 评分 if boundary_count >= 3: score = self.QUALITY_CHECKS['honest_boundaries']['pass_score'] status = 'pass' elif boundary_count >= 1: score = self.QUALITY_CHECKS['honest_boundaries']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['honest_boundaries']['fail_score'] status = 'fail' return { 'check': 'honest_boundaries', 'description': self.QUALITY_CHECKS['honest_boundaries']['description'], 'criteria': self.QUALITY_CHECKS['honest_boundaries']['pass_criteria'], 'actual': boundary_count, 'score': score, 'status': status, 'message': f"发现{boundary_count}条诚实边界", 'algorithm': self.QUALITY_CHECKS['honest_boundaries']['algorithm'] } def _check_internal_tensions_advanced(self, skill_content: str) -> Dict: """高级内在张力检查""" # 识别张力关键词和模式 tension_patterns = [ r'(?:张力|矛盾|冲突|不一致|悖论|两难|困境)', r'内在张力\s*[:：]', r'核心矛盾\s*[:：]' ] found_tensions = set() for pattern in tension_patterns: matches = re.finditer(pattern, skill_content) for match in matches: # 获取上下文 context_start = max(0, match.start() - 50) context_end = min(len(skill_content), match.end() + 50) context = skill_content[context_start:context_end] found_tensions.add(context[:100]) tension_count = len(found_tensions) # 评分 if tension_count >= 2: score = self.QUALITY_CHECKS['internal_tensions']['pass_score'] status = 'pass' elif tension_count >= 1: score = self.QUALITY_CHECKS['internal_tensions']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['internal_tensions']['fail_score'] status = 'fail' return { 'check': 'internal_tensions', 'description': self.QUALITY_CHECKS['internal_tensions']['description'], 'criteria': self.QUALITY_CHECKS['internal_tensions']['pass_criteria'], 'actual': tension_count, 'score': score, 'status': status, 'message': f"发现{tension_count}处内在张力", 'algorithm': self.QUALITY_CHECKS['internal_tensions']['algorithm'] } def _check_decision_heuristics_advanced(self, skill_content: str) -> Dict: """高级决策启发式检查""" # 识别启发式模式 heuristic_patterns = [ r'当(.{1,30})\s*时\s*[，,]?\s*应该?(.{1,60})', r'如果是?(.{1,30})\s*[，,]?\s*则?(.{1,60})', r'(.{1,15})\s*(?:原则|准则|Rule|Principle)\s*[:：]?\s*([^.。]+)', r'在?(.{1,20})\s*时\s*[，,]?\s*优先?(.{1,50})', r'(?:启发式|Heuristic)\s*\d+\s*[:：]' ] found_heuristics = set() for pattern in heuristic_patterns: matches = re.finditer(pattern, skill_content) for match in matches: heuristic = match.group(0).strip() # 过滤和验证质量 if 15 <= len(heuristic) <= 120: found_heuristics.add(heuristic[:80]) heuristic_count = len(found_heuristics) # 评分 if heuristic_count >= 5: score = self.QUALITY_CHECKS['decision_heuristics']['pass_score'] status = 'pass' elif heuristic_count >= 1: score = self.QUALITY_CHECKS['decision_heuristics']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['decision_heuristics']['fail_score'] status = 'fail' return { 'check': 'decision_heuristics', 'description': self.QUALITY_CHECKS['decision_heuristics']['description'], 'criteria': self.QUALITY_CHECKS['decision_heuristics']['pass_criteria'], 'actual': heuristic_count, 'score': score, 'status': status, 'message': f"发现{heuristic_count}条决策启发式", 'algorithm': self.QUALITY_CHECKS['decision_heuristics']['algorithm'] } def _check_agentic_protocol_advanced(self, skill_content: str) -> Dict: """高级Agentic Protocol检查""" # 检查是否存在Agentic Protocol has_agentic_protocol = 'Agentic Protocol' in skill_content or 'agentic protocol' in skill_content.lower() if not has_agentic_protocol: return { 'check': 'agentic_protocol', 'description': self.QUALITY_CHECKS['agentic_protocol']['description'], 'criteria': self.QUALITY_CHECKS['agentic_protocol']['pass_criteria'], 'actual': 0, 'score': self.QUALITY_CHECKS['agentic_protocol']['fail_score'], 'status': 'fail', 'message': '未发现Agentic Protocol', 'algorithm': self.QUALITY_CHECKS['agentic_protocol']['algorithm'] } # 提取Agentic Protocol部分 protocol_pattern = r'Agentic Protocol.*?(?=\n##|\Z)' protocol_match = re.search(protocol_pattern, skill_content, re.DOTALL | re.IGNORECASE) if not protocol_match: return { 'check': 'agentic_protocol', 'description': self.QUALITY_CHECKS['agentic_protocol']['description'], 'criteria': self.QUALITY_CHECKS['agentic_protocol']['pass_criteria'], 'actual': 0, 'score': self.QUALITY_CHECKS['agentic_protocol']['fail_score'], 'status': 'fail', 'message': 'Agentic Protocol格式不正确', 'algorithm': self.QUALITY_CHECKS['agentic_protocol']['algorithm'] } protocol_text = protocol_match.group(0) # 检查步骤 step_patterns = [ r'步骤\s*[123]', r'Step\s*[123]', r'[一二三]\s*[、\.]?\s*步骤' ] steps_found = set() for pattern in step_patterns: matches = re.findall(pattern, protocol_text, re.IGNORECASE) steps_found.update(matches) step_count = len(steps_found) # 检查是否包含必需步骤 required_steps = ['问题', '研究', '回答', 'Question', 'Research', 'Answer'] has_required = any(step in protocol_text for step in required_steps) # 评分 if step_count >= 3 and has_required: score = self.QUALITY_CHECKS['agentic_protocol']['pass_score'] status = 'pass' elif step_count >= 1: score = self.QUALITY_CHECKS['agentic_protocol']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['agentic_protocol']['fail_score'] status = 'fail' return { 'check': 'agentic_protocol', 'description': self.QUALITY_CHECKS['agentic_protocol']['description'], 'criteria': self.QUALITY_CHECKS['agentic_protocol']['pass_criteria'], 'actual': step_count, 'score': score, 'status': status, 'message': f"发现{step_count}个Agentic Protocol步骤", 'algorithm': self.QUALITY_CHECKS['agentic_protocol']['algorithm'] } def _check_toolkit_advanced(self, skill_content: str) -> Dict: """高级工具箱检查""" # 识别工具模式 tool_patterns = [ r'##\s*(?:工具箱|Toolkit|Tools)\s*\n(.*?)(?=\n##|\Z)', r'工具\s*\d*\s*[:：]\s*([^.\n]+)', r'使用\s*(\w+)(?:工具|方法|技巧)', r'Tool\s*\d*\s*[:：]\s*([^.\n]+)' ] found_tools = set() for pattern in tool_patterns: matches = re.finditer(pattern, skill_content, re.DOTALL) for match in matches: tool = match.group(1) if match.lastindex >= 1 else match.group(0) tool = tool.strip() if len(tool) >= 2 and not tool.isdigit(): found_tools.add(tool[:50]) tool_count = len(found_tools) # 评分 if tool_count >= 3: score = self.QUALITY_CHECKS['toolkit']['pass_score'] status = 'pass' elif tool_count >= 1: score = self.QUALITY_CHECKS['toolkit']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['toolkit']['fail_score'] status = 'fail' return { 'check': 'toolkit', 'description': self.QUALITY_CHECKS['toolkit']['description'], 'criteria': self.QUALITY_CHECKS['toolkit']['pass_criteria'], 'actual': tool_count, 'score': score, 'status': status, 'message': f"发现{tool_count}个工具", 'algorithm': self.QUALITY_CHECKS['toolkit']['algorithm'] } def _check_primary_sources_advanced(self, skill_content: str, research_dir: str = None) -> Dict: """高级一手来源检查""" # 如果没有提供research目录，跳过此项 if not research_dir: return { 'check': 'primary_sources', 'description': self.QUALITY_CHECKS['primary_sources']['description'], 'criteria': self.QUALITY_CHECKS['primary_sources']['pass_criteria'], 'actual': 'N/A', 'score': self.QUALITY_CHECKS['primary_sources']['pass_score'], 'status': 'pass', 'message': '未提供research目录，跳过此项', 'algorithm': self.QUALITY_CHECKS['primary_sources']['algorithm'] } research_path = Path(research_dir) if not research_path.exists(): return { 'check': 'primary_sources', 'description': self.QUALITY_CHECKS['primary_sources']['description'], 'criteria': self.QUALITY_CHECKS['primary_sources']['pass_criteria'], 'actual': 'N/A', 'score': self.QUALITY_CHECKS['primary_sources']['pass_score'], 'status': 'pass', 'message': 'Research目录不存在，跳过此项', 'algorithm': self.QUALITY_CHECKS['primary_sources']['algorithm'] } # 统计来源 primary_sources = 0 secondary_sources = 0 total_sources = 0 # 扫描research目录 for file_path in research_path.glob('**/*.md'): total_sources += 1 # 判断来源类型 filename = file_path.stem.lower() # 一手来源指标 primary_indicators = [ 'writings', 'interview', 'speech', 'conversation', 'original', 'primary', 'transcript' ] # 二手来源指标 secondary_indicators = [ 'summary', 'analysis', 'review', 'commentary', 'secondary', 'interpretation' ] if any(ind in filename for ind in primary_indicators): primary_sources += 1 elif any(ind in filename for ind in secondary_indicators): secondary_sources += 1 else: # 默认认为是一手来源 primary_sources += 1 if total_sources > 0: primary_ratio = primary_sources / total_sources else: primary_ratio = 0 # 评分 if primary_ratio > 0.5: score = self.QUALITY_CHECKS['primary_sources']['pass_score'] status = 'pass' elif primary_ratio > 0: score = self.QUALITY_CHECKS['primary_sources']['partial_score'] status = 'partial' else: score = self.QUALITY_CHECKS['primary_sources']['fail_score'] status = 'fail' return { 'check': 'primary_sources', 'description': self.QUALITY_CHECKS['primary_sources']['description'], 'criteria': self.QUALITY_CHECKS['primary_sources']['pass_criteria'], 'actual': f"{int(primary_ratio * 100)}%", 'primary_count': primary_sources, 'secondary_count': secondary_sources, 'total_count': total_sources, 'score': score, 'status': status, 'message': f"发现{primary_sources}/{total_sources}个一手来源 ({int(primary_ratio * 100)}%)", 'algorithm': self.QUALITY_CHECKS['primary_sources']['algorithm'] } def _generate_summary_advanced(self, check_results: Dict, total_score: float) -> Dict: """生成高级验证总结""" passed_checks = sum(1 for result in check_results.values() if result['status'] == 'pass') partial_checks = sum(1 for result in check_results.values() if result['status'] == 'partial') failed_checks = sum(1 for result in check_results.values() if result['status'] == 'fail') return { 'total_checks': len(check_results), 'passed': passed_checks, 'partial': partial_checks, 'failed': failed_checks, 'overall_score': total_score, 'passing_line': self.PASSING_SCORE, 'passed_checks_count': passed_checks, 'pass_rate': f"{passed_checks / len(check_results) * 100:.0f}%" if check_results else "0%" } def _generate_recommendations_advanced(self, check_results: Dict) -> List[Dict]: """生成高级改进建议""" recommendations = [] for check_name, result in check_results.items(): if result['status'] == 'fail': recommendations.append({ 'priority': 'high', 'check': result['description'], 'message': f'未通过，需要补充。当前: {result["actual"]}', 'action': f'添加至少{result["criteria"]}的{result["description"]}' }) elif result['status'] == 'partial': recommendations.append({ 'priority': 'medium', 'check': result['description'], 'message': f'部分通过，建议改进。当前: {result["actual"]}', 'action': f'提升{result["description"]}至{result["criteria"]}' }) return recommendations def _calculate_quality_grade(self, total_score: float) -> str: """计算质量等级""" if total_score >= 90: return 'A+' elif total_score >= 85: return 'A' elif total_score >= 80: return 'A-' elif total_score >= 75: return 'B+' elif total_score >= 70: return 'B' elif total_score >= 65: return 'B-' elif total_score >= 60: return 'C' elif total_score >= 50: return 'D' else: return 'F' def _generate_detailed_analysis(self, check_results: Dict) -> Dict: """生成详细分析""" analysis = { 'strengths': [], 'weaknesses': [], 'opportunities': [] } for check_name, result in check_results.items(): if result['status'] == 'pass': analysis['strengths'].append({ 'area': result['description'], 'performance': f'优秀 (得分: {result["normalized_score"]})' }) elif result['status'] == 'fail': analysis['weaknesses'].append({ 'area': result['description'], 'issue': f'不满足要求 (当前: {result["actual"]})' }) elif result['status'] == 'partial': analysis['opportunities'].append({ 'area': result['description'], 'improvement': f'可以提升 (当前: {result["actual"]}, 目标: {result["criteria"]})' }) return analysis def main(): import argparse parser = argparse.ArgumentParser(description='质量验证（高级版）') parser.add_argument('--skill-file', type=str, required=True, help='Skill文件路径') parser.add_argument('--research-dir', type=str, help='Research目录路径（可选）') parser.add_argument('--output', type=str, help='输出JSON文件路径（可选）') args = parser.parse_args() validator = QualityValidator() # 执行验证 validation_result = validator.validate_skill( args.skill_file, args.research_dir ) # 输出结果 print(json.dumps(validation_result, ensure_ascii=False, indent=2)) # 保存到文件（可选） if args.output: output_path = Path(args.output) output_path.parent.mkdir(parents=True, exist_ok=True) with open(output_path, 'w', encoding='utf-8') as f: json.dump(validation_result, f, ensure_ascii=False, indent=2) print(f"\n验证结果已保存到: {args.output}") if __name__ == '__main__': main()