triage/analysis/workflow_completeness_checker.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
工作流完成度检查器
检查workflow文件是否完成所有任务，生成过滤列表供其他分析脚本使用
"""

import os
import json
import sys
from pathlib import Path
from typing import List, Dict, Tuple


class WorkflowCompletenessChecker:
    """工作流完成度检查器"""
    
    def __init__(self, data_dir: str, output_dir: str):
        """
        初始化检查器
        
        Args:
            data_dir: 数据目录路径
            output_dir: 输出目录路径
        """
        self.data_dir = Path(data_dir)
        self.output_dir = Path(output_dir)
        self.incomplete_files = []
        self.complete_files = []
        self.error_files = []
        
    def check_file_completeness(self, filepath: Path) -> bool:
        """
        检查单个文件是否完成
        
        Args:
            filepath: 文件路径
            
        Returns:
            True if完成，False if未完成
        """
        try:
            with open(filepath, 'r', encoding='utf-8') as f:
                lines = f.readlines()
                if len(lines) < 2:  # 需要至少两行：倒数第二行和最后一行
                    return False
                
                # 检查倒数第二行的task_completion_summary
                second_to_last_line = lines[-2].strip()
                if not second_to_last_line:
                    return False
                
                try:
                    second_to_last_event = json.loads(second_to_last_line)
                    # 检查是否有task_completion_summary字段
                    task_summary = second_to_last_event.get('task_completion_summary', {})
                    if not task_summary:
                        return False
                    
                    # 检查三个阶段的完成状态
                    phases = task_summary.get('phases', {})
                    required_phases = ['triage', 'hpi', 'ph']
                    
                    for phase in required_phases:
                        phase_info = phases.get(phase, {})
                        if not phase_info.get('is_completed', False):
                            return False
                    
                    return True
                except json.JSONDecodeError:
                    return False
                    
        except Exception as e:
            print(f"检查文件 {filepath.name} 时出错: {e}")
            self.error_files.append(filepath.name)
            return False
    
    def scan_directory(self) -> None:
        """扫描目录中的所有workflow文件"""
        if not self.data_dir.exists():
            print(f"数据目录不存在: {self.data_dir}")
            return
        
        # 查找所有jsonl文件
        jsonl_files = list(self.data_dir.glob("*.jsonl"))
        print(f"找到 {len(jsonl_files)} 个数据文件")
        
        for filepath in sorted(jsonl_files):
            if self.check_file_completeness(filepath):
                self.complete_files.append(filepath.name)
            else:
                self.incomplete_files.append(filepath.name)
                
        print(f"完成文件: {len(self.complete_files)} 个")
        print(f"未完成文件: {len(self.incomplete_files)} 个")
        print(f"错误文件: {len(self.error_files)} 个")
    
    def generate_filter_files(self) -> None:
        """生成过滤文件列表"""
        # 创建输出目录
        self.output_dir.mkdir(parents=True, exist_ok=True)
        
        # 保存未完成文件列表（供其他脚本使用）
        incomplete_list_file = self.output_dir / "incomplete_files.txt"
        with open(incomplete_list_file, 'w', encoding='utf-8') as f:
            for filename in self.incomplete_files:
                f.write(f"{filename}\n")
        
        # 保存完成文件列表
        complete_list_file = self.output_dir / "complete_files.txt"
        with open(complete_list_file, 'w', encoding='utf-8') as f:
            for filename in self.complete_files:
                f.write(f"{filename}\n")
        
        # 生成详细统计报告
        report_file = self.output_dir / "completeness_report.json"
        report_data = {
            "scan_directory": str(self.data_dir),
            "total_files": len(self.complete_files) + len(self.incomplete_files) + len(self.error_files),
            "complete_files_count": len(self.complete_files),
            "incomplete_files_count": len(self.incomplete_files),
            "error_files_count": len(self.error_files),
            "completion_rate": len(self.complete_files) / (len(self.complete_files) + len(self.incomplete_files)) if (len(self.complete_files) + len(self.incomplete_files)) > 0 else 0.0,
            "incomplete_files": self.incomplete_files,
            "error_files": self.error_files
        }
        
        with open(report_file, 'w', encoding='utf-8') as f:
            json.dump(report_data, f, ensure_ascii=False, indent=2)
        
        print(f"\n过滤文件已生成:")
        print(f"  - 未完成文件列表: {incomplete_list_file}")
        print(f"  - 完成文件列表: {complete_list_file}")
        print(f"  - 完成度报告: {report_file}")
    
    def print_summary(self) -> None:
        """打印汇总信息"""
        total = len(self.complete_files) + len(self.incomplete_files)
        if total > 0:
            completion_rate = len(self.complete_files) / total * 100
            print(f"\n=== 工作流完成度检查汇总 ===")
            print(f"总文件数: {total}")
            print(f"完成文件: {len(self.complete_files)} 个 ({completion_rate:.1f}%)")
            print(f"未完成文件: {len(self.incomplete_files)} 个")
            
            if self.error_files:
                print(f"错误文件: {len(self.error_files)} 个")
            
            if self.incomplete_files:
                print(f"\n未完成的文件（前10个）:")
                for filename in self.incomplete_files[:10]:
                    print(f"  - {filename}")
                if len(self.incomplete_files) > 10:
                    print(f"  ... 还有 {len(self.incomplete_files) - 10} 个")
    
    def run_check(self) -> None:
        """运行完整的检查流程"""
        print("开始检查工作流完成度...")
        
        # 1. 扫描目录
        self.scan_directory()
        
        # 2. 生成过滤文件
        self.generate_filter_files()
        
        # 3. 打印汇总
        self.print_summary()
        
        print("完成度检查完成！")


def main():
    """主函数"""
    import sys
    
    # 从命令行参数获取路径，如果没有提供则使用默认值
    if len(sys.argv) >= 3:
        data_dir = sys.argv[1]
        output_dir = sys.argv[2]
    else:
        data_dir = "results/results0902"
        output_dir = "analysis/0902"
    
    checker = WorkflowCompletenessChecker(data_dir=data_dir, output_dir=output_dir)
    checker.run_check()


if __name__ == "__main__":
    main()