jarvis-ai-assistant 0.3.30__py3-none-any.whl → 0.7.0__py3-none-any.whl

jarvis/jarvis_code_agent/code_analyzer/impact_analyzer.py

@@ -0,0 +1,781 @@
+"""编辑影响范围分析模块。
+
+提供代码编辑影响范围分析功能，识别可能受影响的文件、函数、测试等。
+"""
+
+import os
+import re
+import ast
+import subprocess
+from dataclasses import dataclass, field
+from typing import List, Optional, Set, Dict
+from enum import Enum
+
+from .context_manager import ContextManager
+from .file_ignore import filter_walk_dirs
+from .symbol_extractor import Symbol
+
+
+
+class ImpactType(Enum):
+    """影响类型枚举"""
+    REFERENCE = "reference"  # 符号引用
+    DEPENDENT = "dependent"  # 依赖的符号
+    TEST = "test"  # 测试文件
+    INTERFACE_CHANGE = "interface_change"  # 接口变更
+    DEPENDENCY_CHAIN = "dependency_chain"  # 依赖链
+
+
+class RiskLevel(Enum):
+    """风险等级枚举"""
+    LOW = "low"
+    MEDIUM = "medium"
+    HIGH = "high"
+
+
+@dataclass
+class Impact:
+    """表示一个影响项"""
+    impact_type: ImpactType
+    target: str  # 受影响的目标（文件路径、符号名等）
+    description: str = ""
+    line: Optional[int] = None
+    severity: RiskLevel = RiskLevel.LOW
+
+
+@dataclass
+class InterfaceChange:
+    """表示接口变更"""
+    symbol_name: str
+    change_type: str  # 'signature', 'return_type', 'parameter', 'removed', 'added'
+    file_path: str
+    line: int
+    before: Optional[str] = None
+    after: Optional[str] = None
+    description: str = ""
+
+
+@dataclass
+class ImpactReport:
+    """影响分析报告"""
+    affected_files: List[str] = field(default_factory=list)
+    affected_symbols: List[Symbol] = field(default_factory=list)
+    affected_tests: List[str] = field(default_factory=list)
+    interface_changes: List[InterfaceChange] = field(default_factory=list)
+    impacts: List[Impact] = field(default_factory=list)
+    risk_level: RiskLevel = RiskLevel.LOW
+    recommendations: List[str] = field(default_factory=list)
+    
+    def to_string(self, project_root: str = "") -> str:
+        """生成可读的影响报告字符串"""
+        lines = []
+        lines.append("=" * 60)
+        lines.append("编辑影响范围分析报告")
+        lines.append("=" * 60)
+        
+        # 风险等级
+        risk_emoji = {
+            RiskLevel.LOW: "🟢",
+            RiskLevel.MEDIUM: "🟡",
+            RiskLevel.HIGH: "🔴"
+        }
+        lines.append(f"\n风险等级: {risk_emoji.get(self.risk_level, '⚪')} {self.risk_level.value.upper()}")
+        
+        # 受影响文件
+        if self.affected_files:
+            lines.append(f"\n受影响文件 ({len(self.affected_files)}):")
+            for file_path in self.affected_files[:10]:
+                rel_path = os.path.relpath(file_path, project_root) if project_root else file_path
+                lines.append(f"  - {rel_path}")
+            if len(self.affected_files) > 10:
+                lines.append(f"  ... 还有 {len(self.affected_files) - 10} 个文件")
+        
+        # 受影响符号
+        if self.affected_symbols:
+            lines.append(f"\n受影响符号 ({len(self.affected_symbols)}):")
+            for symbol in self.affected_symbols[:10]:
+                lines.append(f"  - {symbol.kind} {symbol.name} ({os.path.basename(symbol.file_path)}:{symbol.line_start})")
+            if len(self.affected_symbols) > 10:
+                lines.append(f"  ... 还有 {len(self.affected_symbols) - 10} 个符号")
+        
+        # 受影响测试
+        if self.affected_tests:
+            lines.append(f"\n受影响测试 ({len(self.affected_tests)}):")
+            for test_file in self.affected_tests[:10]:
+                rel_path = os.path.relpath(test_file, project_root) if project_root else test_file
+                lines.append(f"  - {rel_path}")
+            if len(self.affected_tests) > 10:
+                lines.append(f"  ... 还有 {len(self.affected_tests) - 10} 个测试文件")
+        
+        # 接口变更
+        if self.interface_changes:
+            lines.append(f"\n接口变更 ({len(self.interface_changes)}):")
+            for change in self.interface_changes[:10]:
+                lines.append(f"  - {change.symbol_name}: {change.change_type}")
+                if change.description:
+                    lines.append(f"    {change.description}")
+            if len(self.interface_changes) > 10:
+                lines.append(f"  ... 还有 {len(self.interface_changes) - 10} 个接口变更")
+        
+        # 建议
+        if self.recommendations:
+            lines.append("\n建议:")
+            for i, rec in enumerate(self.recommendations, 1):
+                lines.append(f"  {i}. {rec}")
+        
+        lines.append("\n" + "=" * 60)
+        return "\n".join(lines)
+
+
+@dataclass
+class Edit:
+    """表示一个编辑操作"""
+    file_path: str
+    line_start: int
+    line_end: int
+    before: str = ""
+    after: str = ""
+    edit_type: str = "modify"  # 'modify', 'add', 'delete'
+
+
+class TestDiscoverer:
+    """测试文件发现器"""
+    
+    # 测试文件命名模式
+    TEST_PATTERNS = {
+        'python': [
+            r'test_.*\.py$',
+            r'.*_test\.py$',
+        ],
+        'javascript': [
+            r'.*\.test\.(js|ts|jsx|tsx)$',
+            r'.*\.spec\.(js|ts|jsx|tsx)$',
+        ],
+        'rust': [
+            r'.*_test\.rs$',
+        ],
+        'java': [
+            r'.*Test\.java$',
+            r'.*Tests\.java$',
+        ],
+        'go': [
+            r'.*_test\.go$',
+        ],
+    }
+    
+    def __init__(self, project_root: str):
+        self.project_root = project_root
+    
+    def find_test_files(self, file_path: str) -> List[str]:
+        """查找与文件相关的测试文件"""
+        test_files: List[str] = []
+        
+        # 检测语言
+        language = self._detect_language(file_path)
+        if not language:
+            return test_files
+        
+        # 获取测试文件模式
+        patterns = self.TEST_PATTERNS.get(language, [])
+        if not patterns:
+            return test_files
+        
+        # 获取文件的基础名称（不含扩展名）
+        base_name = os.path.splitext(os.path.basename(file_path))[0]
+        
+        # 在项目根目录搜索测试文件
+        for root, dirs, files in os.walk(self.project_root):
+            # 跳过隐藏目录和常见忽略目录
+            dirs[:] = filter_walk_dirs(dirs)
+            
+            for file in files:
+                file_path_full = os.path.join(root, file)
+                
+                # 检查是否匹配测试文件模式
+                for pattern in patterns:
+                    if re.match(pattern, file, re.IGNORECASE):
+                        # 检查测试文件是否可能测试目标文件
+                        if self._might_test_file(file_path_full, file_path, base_name):
+                            test_files.append(file_path_full)
+                        break
+        
+        return list(set(test_files))
+    
+    def _might_test_file(self, test_file: str, target_file: str, base_name: str) -> bool:
+        """判断测试文件是否可能测试目标文件"""
+        # 读取测试文件内容，查找目标文件的引用
+        try:
+            with open(test_file, 'r', encoding='utf-8', errors='replace') as f:
+                content = f.read()
+            
+            # 检查是否导入或引用了目标文件
+            # 简单的启发式方法：检查文件名、模块名等
+            target_base = os.path.splitext(os.path.basename(target_file))[0]
+            
+            # 检查导入语句
+            import_patterns = [
+                rf'import\s+.*{re.escape(target_base)}',
+                rf'from\s+.*{re.escape(target_base)}',
+                rf'use\s+.*{re.escape(target_base)}',  # Rust
+            ]
+            
+            for pattern in import_patterns:
+                if re.search(pattern, content, re.IGNORECASE):
+                    return True
+            
+            # 检查文件名是否出现在测试文件中
+            if target_base.lower() in content.lower():
+                return True
+                
+        except Exception:
+            pass
+        
+        return False
+    
+    def _detect_language(self, file_path: str) -> Optional[str]:
+        """检测文件语言"""
+        ext = os.path.splitext(file_path)[1].lower()
+        ext_map = {
+            '.py': 'python',
+            '.js': 'javascript',
+            '.ts': 'javascript',
+            '.jsx': 'javascript',
+            '.tsx': 'javascript',
+            '.rs': 'rust',
+            '.java': 'java',
+            '.go': 'go',
+        }
+        return ext_map.get(ext)
+
+
+class ImpactAnalyzer:
+    """编辑影响范围分析器"""
+    
+    def __init__(self, context_manager: ContextManager):
+        self.context_manager = context_manager
+        self.project_root = context_manager.project_root
+        self.test_discoverer = TestDiscoverer(self.project_root)
+    
+    def analyze_edit_impact(self, file_path: str, edits: List[Edit]) -> ImpactReport:
+        """分析编辑的影响范围
+        
+        Args:
+            file_path: 被编辑的文件路径
+            edits: 编辑操作列表
+            
+        Returns:
+            ImpactReport: 影响分析报告
+        """
+        impacts: List[Impact] = []
+        affected_symbols: Set[Symbol] = set()
+        affected_files: Set[str] = {file_path}
+        interface_changes: List[InterfaceChange] = []
+        
+        # 1. 分析每个编辑的影响
+        for edit in edits:
+            # 分析符号影响
+            symbols_in_edit = self._find_symbols_in_edit(file_path, edit)
+            for symbol in symbols_in_edit:
+                affected_symbols.add(symbol)
+                symbol_impacts = self._analyze_symbol_impact(symbol, edit)
+                impacts.extend(symbol_impacts)
+                
+                # 收集受影响的文件
+                for impact in symbol_impacts:
+                    if impact.impact_type == ImpactType.REFERENCE:
+                        affected_files.add(impact.target)
+                    elif impact.impact_type == ImpactType.DEPENDENT:
+                        affected_files.add(impact.target)
+        
+        # 2. 分析依赖链影响
+        dependency_impacts = self._analyze_dependency_chain(file_path)
+        impacts.extend(dependency_impacts)
+        for impact in dependency_impacts:
+            affected_files.add(impact.target)
+        
+        # 3. 检测接口变更
+        if edits:
+            # 需要读取文件内容来比较
+            interface_changes = self._detect_interface_changes(file_path, edits)
+            for change in interface_changes:
+                affected_files.add(change.file_path)
+        
+        # 4. 查找相关测试
+        test_files = self.test_discoverer.find_test_files(file_path)
+        for test_file in test_files:
+            impacts.append(Impact(
+                impact_type=ImpactType.TEST,
+                target=test_file,
+                description=f"可能测试 {os.path.basename(file_path)} 的测试文件"
+            ))
+            affected_files.add(test_file)
+        
+        # 5. 评估风险等级
+        risk_level = self._assess_risk(impacts, interface_changes)
+        
+        # 6. 生成建议
+        recommendations = self._generate_recommendations(
+            impacts, interface_changes, affected_files, test_files
+        )
+        
+        return ImpactReport(
+            affected_files=list(affected_files),
+            affected_symbols=list(affected_symbols),
+            affected_tests=test_files,
+            interface_changes=interface_changes,
+            impacts=impacts,
+            risk_level=risk_level,
+            recommendations=recommendations,
+        )
+    
+    def _find_symbols_in_edit(self, file_path: str, edit: Edit) -> List[Symbol]:
+        """查找编辑区域内的符号"""
+        symbols = self.context_manager.symbol_table.get_file_symbols(file_path)
+        
+        # 找出在编辑范围内的符号
+        affected_symbols = []
+        for symbol in symbols:
+            # 检查符号是否与编辑区域重叠
+            if (symbol.line_start <= edit.line_end and 
+                symbol.line_end >= edit.line_start):
+                affected_symbols.append(symbol)
+        
+        return affected_symbols
+    
+    def _analyze_symbol_impact(self, symbol: Symbol, edit: Edit) -> List[Impact]:
+        """分析符号编辑的影响"""
+        impacts = []
+        
+        # 1. 查找所有引用该符号的位置
+        references = self.context_manager.find_references(symbol.name, symbol.file_path)
+        for ref in references:
+            impacts.append(Impact(
+                impact_type=ImpactType.REFERENCE,
+                target=ref.file_path,
+                description=f"引用符号 {symbol.name}",
+                line=ref.line,
+                severity=RiskLevel.MEDIUM if symbol.kind in ('function', 'class') else RiskLevel.LOW
+            ))
+        
+        # 2. 查找依赖该符号的其他符号（在同一文件中）
+        if symbol.kind in ('function', 'class'):
+            dependents = self._find_dependent_symbols(symbol)
+            for dep in dependents:
+                impacts.append(Impact(
+                    impact_type=ImpactType.DEPENDENT,
+                    target=dep.file_path,
+                    description=f"依赖符号 {symbol.name}",
+                    line=dep.line_start,
+                    severity=RiskLevel.MEDIUM
+                ))
+        
+        return impacts
+    
+    def _find_dependent_symbols(self, symbol: Symbol) -> List[Symbol]:
+        """查找依赖该符号的其他符号"""
+        dependents = []
+        
+        # 获取同一文件中的所有符号
+        file_symbols = self.context_manager.symbol_table.get_file_symbols(symbol.file_path)
+        
+        # 查找在符号定义之后的符号（可能使用该符号）
+        for other_symbol in file_symbols:
+            if (other_symbol.line_start > symbol.line_end and
+                other_symbol.name != symbol.name):
+                # 简单检查：如果符号名出现在其他符号的范围内，可能依赖
+                # 这里使用简单的启发式方法
+                content = self.context_manager._get_file_content(symbol.file_path)
+                if content:
+                    # 提取其他符号的代码区域
+                    lines = content.split('\n')
+                    if other_symbol.line_start <= len(lines) and other_symbol.line_end <= len(lines):
+                        region = '\n'.join(lines[other_symbol.line_start-1:other_symbol.line_end])
+                        if symbol.name in region:
+                            dependents.append(other_symbol)
+        
+        return dependents
+    
+    def _analyze_dependency_chain(self, file_path: str) -> List[Impact]:
+        """分析依赖链，找出所有可能受影响的文件"""
+        impacts = []
+        
+        # 获取依赖该文件的所有文件（传递闭包）
+        visited = set()
+        to_process = [file_path]
+        
+        while to_process:
+            current = to_process.pop(0)
+            if current in visited:
+                continue
+            visited.add(current)
+            
+            dependents = self.context_manager.dependency_graph.get_dependents(current)
+            for dependent in dependents:
+                if dependent not in visited:
+                    impacts.append(Impact(
+                        impact_type=ImpactType.DEPENDENCY_CHAIN,
+                        target=dependent,
+                        description=f"间接依赖 {os.path.basename(file_path)}",
+                        severity=RiskLevel.LOW
+                    ))
+                    to_process.append(dependent)
+        
+        return impacts
+    
+    def _detect_interface_changes(self, file_path: str, edits: List[Edit]) -> List[InterfaceChange]:
+        """检测接口变更（函数签名、类定义等）"""
+        changes: List[InterfaceChange] = []
+        
+        # 读取文件内容
+        content_before = self._get_file_content_before_edit(file_path, edits)
+        content_after = self._get_file_content_after_edit(file_path, edits)
+        
+        if not content_before or not content_after:
+            return changes
+        
+        # 解析AST并比较
+        try:
+            tree_before = ast.parse(content_before, filename=file_path)
+            tree_after = ast.parse(content_after, filename=file_path)
+            
+            # 提取函数和类定义
+            defs_before = self._extract_definitions(tree_before)
+            defs_after = self._extract_definitions(tree_after)
+            
+            # 比较定义
+            for name, def_before in defs_before.items():
+                if name in defs_after:
+                    def_after = defs_after[name]
+                    change = self._compare_definition(name, def_before, def_after, file_path)
+                    if change:
+                        changes.append(change)
+                else:
+                    # 定义被删除
+                    changes.append(InterfaceChange(
+                        symbol_name=name,
+                        change_type='removed',
+                        file_path=file_path,
+                        line=def_before['line'],
+                        description=f"符号 {name} 被删除"
+                    ))
+            
+            # 检查新增的定义
+            for name, def_after in defs_after.items():
+                if name not in defs_before:
+                    changes.append(InterfaceChange(
+                        symbol_name=name,
+                        change_type='added',
+                        file_path=file_path,
+                        line=def_after['line'],
+                        description=f"新增符号 {name}"
+                    ))
+                    
+        except SyntaxError:
+            # 如果解析失败，跳过接口变更检测
+            pass
+        
+        return changes
+    
+    def _extract_definitions(self, tree: ast.AST) -> Dict[str, Dict]:
+        """从AST中提取函数和类定义"""
+        definitions = {}
+        
+        for node in ast.walk(tree):
+            if isinstance(node, ast.FunctionDef):
+                # 提取函数签名
+                args = [arg.arg for arg in node.args.args]
+                signature = f"{node.name}({', '.join(args)})"
+                definitions[node.name] = {
+                    'type': 'function',
+                    'line': node.lineno,
+                    'signature': signature,
+                    'args': args,
+                    'node': node,
+                }
+            elif isinstance(node, ast.ClassDef):
+                definitions[node.name] = {
+                    'type': 'class',
+                    'line': node.lineno,
+                    'signature': node.name,
+                    'node': node,
+                }
+        
+        return definitions
+    
+    def _compare_definition(self, name: str, def_before: Dict, def_after: Dict, file_path: str) -> Optional[InterfaceChange]:
+        """比较两个定义，检测接口变更"""
+        if def_before['type'] != def_after['type']:
+            return InterfaceChange(
+                symbol_name=name,
+                change_type='signature',
+                file_path=file_path,
+                line=def_after['line'],
+                before=def_before['signature'],
+                after=def_after['signature'],
+                description=f"符号 {name} 的类型从 {def_before['type']} 变为 {def_after['type']}"
+            )
+        
+        if def_before['type'] == 'function':
+            # 比较函数参数
+            args_before = def_before.get('args', [])
+            args_after = def_after.get('args', [])
+            
+            if args_before != args_after:
+                return InterfaceChange(
+                    symbol_name=name,
+                    change_type='signature',
+                    file_path=file_path,
+                    line=def_after['line'],
+                    before=def_before['signature'],
+                    after=def_after['signature'],
+                    description=f"函数 {name} 的参数从 ({', '.join(args_before)}) 变为 ({', '.join(args_after)})"
+                )
+        
+        return None
+    
+    def _get_file_content_before_edit(self, file_path: str, edits: List[Edit]) -> Optional[str]:
+        """获取编辑前的文件内容"""
+        try:
+            with open(file_path, 'r', encoding='utf-8', errors='replace') as f:
+                return f.read()
+        except Exception:
+            return None
+    
+    def _get_file_content_after_edit(self, file_path: str, edits: List[Edit]) -> Optional[str]:
+        """获取编辑后的文件内容（模拟）"""
+        # 这里应该根据edits模拟编辑后的内容
+        # 为了简化，我们直接读取当前文件内容
+        # 在实际使用中，应该根据edits应用变更
+        try:
+            with open(file_path, 'r', encoding='utf-8', errors='replace') as f:
+                return f.read()
+        except Exception:
+            return None
+    
+    def _assess_risk(self, impacts: List[Impact], interface_changes: List[InterfaceChange]) -> RiskLevel:
+        """评估编辑风险等级"""
+        # 统计高风险因素
+        high_risk_count = 0
+        medium_risk_count = 0
+        
+        # 接口变更通常是高风险
+        if interface_changes:
+            high_risk_count += len(interface_changes)
+        
+        # 统计影响数量
+        reference_count = sum(1 for i in impacts if i.impact_type == ImpactType.REFERENCE)
+        if reference_count > 10:
+            high_risk_count += 1
+        elif reference_count > 5:
+            medium_risk_count += 1
+        
+        # 检查是否有高风险的影响
+        for impact in impacts:
+            if impact.severity == RiskLevel.HIGH:
+                high_risk_count += 1
+            elif impact.severity == RiskLevel.MEDIUM:
+                medium_risk_count += 1
+        
+        # 评估风险等级
+        if high_risk_count > 0 or medium_risk_count > 3:
+            return RiskLevel.HIGH
+        elif medium_risk_count > 0 or len(impacts) > 5:
+            return RiskLevel.MEDIUM
+        else:
+            return RiskLevel.LOW
+    
+    def _generate_recommendations(
+        self,
+        impacts: List[Impact],
+        interface_changes: List[InterfaceChange],
+        affected_files: Set[str],
+        test_files: List[str],
+    ) -> List[str]:
+        """生成修复建议"""
+        recommendations = []
+        
+        # 如果有接口变更，建议检查所有调用点
+        if interface_changes:
+            recommendations.append(
+                f"检测到 {len(interface_changes)} 个接口变更，请检查所有调用点并更新相关代码"
+            )
+        
+        # 如果有测试文件，建议运行测试
+        if test_files:
+            recommendations.append(
+                f"发现 {len(test_files)} 个相关测试文件，建议运行测试确保功能正常"
+            )
+        
+        # 如果影响文件较多，建议增量测试
+        if len(affected_files) > 5:
+            recommendations.append(
+                f"编辑影响了 {len(affected_files)} 个文件，建议进行增量测试"
+            )
+        
+        # 如果有大量引用，建议代码审查
+        reference_count = sum(1 for i in impacts if i.impact_type == ImpactType.REFERENCE)
+        if reference_count > 10:
+            recommendations.append(
+                f"检测到 {reference_count} 个符号引用，建议进行代码审查"
+            )
+        
+        if not recommendations:
+            recommendations.append("编辑影响范围较小，建议进行基本测试")
+        
+        return recommendations
+
+
+def parse_git_diff_to_edits(file_path: str, project_root: str) -> List[Edit]:
+    """从git diff中解析编辑操作
+    
+    Args:
+        file_path: 文件路径
+        project_root: 项目根目录
+        
+    Returns:
+        List[Edit]: 编辑操作列表
+    """
+    edits: List[Edit] = []
+    
+    try:
+        # 获取文件的git diff
+        abs_path = os.path.abspath(file_path)
+        if not os.path.exists(abs_path):
+            return edits
+        
+        # 检查是否有git仓库
+        try:
+            subprocess.run(
+                ["git", "rev-parse", "--git-dir"],
+                cwd=project_root,
+                check=True,
+                capture_output=True,
+                stderr=subprocess.DEVNULL,
+            )
+        except (subprocess.CalledProcessError, FileNotFoundError):
+            # 不是git仓库或git不可用，返回空列表
+            return edits
+        
+        # 获取HEAD的hash
+        try:
+            result = subprocess.run(
+                ["git", "rev-parse", "HEAD"],
+                cwd=project_root,
+                capture_output=True,
+                text=True,
+                check=False,
+            )
+            head_exists = result.returncode == 0 and result.stdout.strip()
+        except Exception:
+            head_exists = False
+        
+        # 临时添加文件到git索引（如果是新文件）
+        subprocess.run(
+            ["git", "add", "-N", "--", abs_path],
+            cwd=project_root,
+            check=False,
+            stdout=subprocess.DEVNULL,
+            stderr=subprocess.DEVNULL,
+        )
+        
+        try:
+            # 获取diff
+            cmd = ["git", "diff"] + (["HEAD"] if head_exists else []) + ["--", abs_path]
+            result = subprocess.run(
+                cmd,
+                cwd=project_root,
+                capture_output=True,
+                text=True,
+                encoding="utf-8",
+                errors="replace",
+                check=False,
+            )
+            
+            if result.returncode != 0 or not result.stdout:
+                return edits
+            
+            diff_text = result.stdout
+            
+            # 解析diff文本
+            lines = diff_text.split('\n')
+            current_hunk_start = None
+            current_line_num: Optional[int] = None
+            before_lines: List[str] = []
+            after_lines: List[str] = []
+            in_hunk = False
+            
+            for line in lines:
+                # 解析hunk header: @@ -start,count +start,count @@
+                if line.startswith('@@'):
+                    # 保存之前的hunk
+                    if in_hunk and current_hunk_start is not None:
+                        if before_lines or after_lines:
+                            edits.append(Edit(
+                                file_path=abs_path,
+                                line_start=current_hunk_start,
+                                line_end=current_hunk_start + len(after_lines) - 1 if after_lines else current_hunk_start,
+                                before='\n'.join(before_lines),
+                                after='\n'.join(after_lines),
+                                edit_type='modify' if before_lines and after_lines else ('delete' if before_lines else 'add')
+                            ))
+                    
+                    # 解析新的hunk
+                    match = re.search(r'@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@', line)
+                    if match:
+                        old_start = int(match.group(1))
+                        new_start = int(match.group(3))
+                        
+                        current_hunk_start = new_start
+                        current_line_num = old_start
+                        before_lines = []
+                        after_lines = []
+                        in_hunk = True
+                    continue
+                
+                if not in_hunk:
+                    continue
+                
+                # 解析diff行
+                if line.startswith('-') and not line.startswith('---'):
+                    # 删除的行
+                    before_lines.append(line[1:])
+                    if current_line_num is not None:
+                        current_line_num += 1
+                elif line.startswith('+') and not line.startswith('+++'):
+                    # 新增的行
+                    after_lines.append(line[1:])
+                elif line.startswith(' '):
+                    # 未改变的行
+                    before_lines.append(line[1:])
+                    after_lines.append(line[1:])
+                    if current_line_num is not None:
+                        current_line_num += 1
+            
+            # 保存最后一个hunk
+            if in_hunk and current_hunk_start is not None:
+                if before_lines or after_lines:
+                    edits.append(Edit(
+                        file_path=abs_path,
+                        line_start=current_hunk_start,
+                        line_end=current_hunk_start + len(after_lines) - 1 if after_lines else current_hunk_start,
+                        before='\n'.join(before_lines),
+                        after='\n'.join(after_lines),
+                        edit_type='modify' if before_lines and after_lines else ('delete' if before_lines else 'add')
+                    ))
+        
+        finally:
+            # 清理临时添加的文件
+            subprocess.run(
+                ["git", "reset", "--", abs_path],
+                cwd=project_root,
+                check=False,
+                stdout=subprocess.DEVNULL,
+                stderr=subprocess.DEVNULL,
+            )
+    
+    except Exception:
+        # 解析失败时返回空列表
+        pass
+    
+    return edits
+