auto-coder 0.1.400__py3-none-any.whl → 2.0.0__py3-none-any.whl

autocoder/common/pruner/tools/query_message_ids.py

@@ -0,0 +1,264 @@
+#!/usr/bin/env python3
+"""
+会话消息ID配置查询工具
+
+命令行工具，用于查询和显示指定会话ID的消息删除配置信息。
+
+消息ID格式支持：
+    1. 直接字段格式: {"message_id": "9226b3a4-1234-5678-9abc-def012345678"}
+    2. Hint嵌入格式: {"content": "消息内容 [[message_id: 9226b3a4]]"}
+    3. 兼容格式: {"content": "消息内容 message_id: 9226b3a4"}
+
+用法示例:
+    python -m autocoder.common.pruner.tools.query_message_ids --conversation-id 796e1c2e-1ab8-46d1-9448-f9bd29d5c095
+    python -m autocoder.common.pruner.tools.query_message_ids -c 796e1c2e-1ab8-46d1-9448-f9bd29d5c095 --verbose
+    python -m autocoder.common.pruner.tools.query_message_ids --list-all
+"""
+
+import argparse
+import sys
+import json
+from datetime import datetime
+from typing import Optional
+
+from ..conversation_message_ids_api import get_conversation_message_ids_api
+
+
+def format_timestamp(timestamp: str) -> str:
+    """格式化时间戳显示"""
+    try:
+        dt = datetime.fromisoformat(timestamp.replace('Z', '+00:00'))
+        return dt.strftime('%Y-%m-%d %H:%M:%S')
+    except:
+        return timestamp
+
+
+def print_separator(char: str = "=", length: int = 80):
+    """打印分隔线"""
+    print(char * length)
+
+
+def print_section_header(title: str):
+    """打印节标题"""
+    print()
+    print(f"📋 {title}:")
+
+
+def query_conversation_message_ids(conversation_id: str, verbose: bool = False):
+    """查询并打印指定会话ID的消息删除配置"""
+    
+    print(f"🔍 查询会话ID: {conversation_id}")
+    print_separator()
+    
+    try:
+        # 获取API实例
+        api = get_conversation_message_ids_api()
+        
+        # 获取消息ID配置
+        message_ids_config = api.get_conversation_message_ids(conversation_id)
+        
+        if not message_ids_config:
+            print(f"❌ 未找到会话ID '{conversation_id}' 的消息删除配置")
+            return False
+        
+        print(f"✅ 找到会话ID '{conversation_id}' 的消息删除配置")
+        
+        # 基本信息
+        print_section_header("基本信息")
+        print(f"  会话ID: {message_ids_config.conversation_id}")
+        print(f"  描述: {message_ids_config.description or '无描述'}")
+        print(f"  保持成对删除: {'是' if message_ids_config.preserve_pairs else '否'}")
+        print(f"  创建时间: {format_timestamp(message_ids_config.created_at)}")
+        print(f"  更新时间: {format_timestamp(message_ids_config.updated_at)}")
+        
+        # 消息ID列表
+        print_section_header("要删除的消息ID列表")
+        if message_ids_config.message_ids:
+            for i, msg_id in enumerate(message_ids_config.message_ids, 1):
+                print(f"  {i:2d}. {msg_id}")
+            print(f"\n📊 总计: {len(message_ids_config.message_ids)} 个消息ID")
+        else:
+            print("  (无消息ID)")
+        
+        # 获取统计信息
+        if verbose:
+            stats = api.get_message_ids_statistics(conversation_id)
+            if stats:
+                print_section_header("详细统计信息")
+                print(f"  配置的消息ID数量: {stats['total_message_ids']}")
+                print(f"  唯一消息ID数量: {stats['unique_message_ids']}")
+                print(f"  保持成对删除: {'是' if stats['preserve_pairs'] else '否'}")
+                print(f"  创建时间: {format_timestamp(stats['created_at'])}")
+                print(f"  更新时间: {format_timestamp(stats['updated_at'])}")
+        
+        # JSON格式输出
+        if verbose:
+            print_section_header("完整配置 (JSON格式)")
+            config_dict = message_ids_config.to_dict()
+            print(json.dumps(config_dict, ensure_ascii=False, indent=2))
+        
+        return True
+        
+    except Exception as e:
+        print(f"❌ 查询过程中发生错误: {str(e)}")
+        if verbose:
+            import traceback
+            traceback.print_exc()
+        return False
+
+
+def list_all_conversations(verbose: bool = False):
+    """列出所有已配置的会话ID"""
+    
+    print("📋 所有已配置消息删除的会话列表")
+    print_separator()
+    
+    try:
+        api = get_conversation_message_ids_api()
+        
+        # 获取所有会话ID
+        conversation_ids = api.list_conversation_ids()
+        
+        if not conversation_ids:
+            print("❌ 未找到任何已配置消息删除的会话")
+            return False
+        
+        print(f"✅ 找到 {len(conversation_ids)} 个已配置的会话:")
+        print()
+        
+        # 列出每个会话的基本信息
+        for i, conv_id in enumerate(conversation_ids, 1):
+            config = api.get_conversation_message_ids(conv_id)
+            if config:
+                print(f"  {i:2d}. {conv_id}")
+                if verbose:
+                    print(f"      描述: {config.description or '无描述'}")
+                    print(f"      消息ID数量: {len(config.message_ids)}")
+                    print(f"      创建时间: {format_timestamp(config.created_at)}")
+                    print(f"      保持成对删除: {'是' if config.preserve_pairs else '否'}")
+                    print()
+        
+        if not verbose:
+            print(f"\n💡 使用 --verbose 参数查看详细信息")
+        
+        return True
+        
+    except Exception as e:
+        print(f"❌ 列表查询过程中发生错误: {str(e)}")
+        if verbose:
+            import traceback
+            traceback.print_exc()
+        return False
+
+
+def export_all_configs(output_file: Optional[str] = None, verbose: bool = False):
+    """导出所有消息ID配置"""
+    
+    print("📤 导出所有消息ID配置")
+    print_separator()
+    
+    try:
+        api = get_conversation_message_ids_api()
+        
+        # 导出数据
+        export_data = api.export_message_ids_to_dict()
+        
+        if output_file:
+            # 导出到文件
+            with open(output_file, 'w', encoding='utf-8') as f:
+                json.dump(export_data, f, ensure_ascii=False, indent=2)
+            print(f"✅ 已导出 {export_data['total_configs']} 个配置到文件: {output_file}")
+        else:
+            # 打印到控制台
+            print(f"✅ 共 {export_data['total_configs']} 个配置:")
+            print(json.dumps(export_data, ensure_ascii=False, indent=2))
+        
+        return True
+        
+    except Exception as e:
+        print(f"❌ 导出过程中发生错误: {str(e)}")
+        if verbose:
+            import traceback
+            traceback.print_exc()
+        return False
+
+
+def main():
+    """主函数"""
+    parser = argparse.ArgumentParser(
+        description="会话消息ID配置查询工具",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+示例用法:
+  # 查询特定会话的消息删除配置
+  %(prog)s --conversation-id 796e1c2e-1ab8-46d1-9448-f9bd29d5c095
+  
+  # 详细查询（包含JSON输出）
+  %(prog)s -c 796e1c2e-1ab8-46d1-9448-f9bd29d5c095 --verbose
+  
+  # 列出所有已配置的会话
+  %(prog)s --list-all
+  
+  # 详细列出所有会话
+  %(prog)s --list-all --verbose
+  
+  # 导出所有配置到文件
+  %(prog)s --export --output configs.json
+        """
+    )
+    
+    # 互斥的主要操作选项
+    action_group = parser.add_mutually_exclusive_group(required=True)
+    action_group.add_argument(
+        '-c', '--conversation-id',
+        type=str,
+        help='要查询的会话ID'
+    )
+    action_group.add_argument(
+        '--list-all',
+        action='store_true',
+        help='列出所有已配置消息删除的会话'
+    )
+    action_group.add_argument(
+        '--export',
+        action='store_true',
+        help='导出所有消息ID配置'
+    )
+    
+    # 辅助选项
+    parser.add_argument(
+        '-v', '--verbose',
+        action='store_true',
+        help='显示详细信息（包含JSON格式输出）'
+    )
+    parser.add_argument(
+        '--output',
+        type=str,
+        help='导出文件路径（仅在使用 --export 时有效）'
+    )
+    
+    args = parser.parse_args()
+    
+    # 打印工具标题和时间戳
+    print("🔍 会话消息ID配置查询工具")
+    print(f"📅 执行时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
+    print()
+    
+    success = False
+    
+    if args.conversation_id:
+        # 查询特定会话
+        success = query_conversation_message_ids(args.conversation_id, args.verbose)
+    elif args.list_all:
+        # 列出所有会话
+        success = list_all_conversations(args.verbose)
+    elif args.export:
+        # 导出配置
+        success = export_all_configs(args.output, args.verbose)
+    
+    # 根据结果设置退出码
+    sys.exit(0 if success else 1)
+
+
+if __name__ == "__main__":
+    main() 

autocoder/common/pruner/tools/test_agentic_pruning_logic.py

@@ -0,0 +1,432 @@
+#!/usr/bin/env python3
+"""
+AgenticConversationPruner 消息ID裁剪逻辑验证脚本
+
+本脚本用于验证 agentic_conversation_pruner.py 中基于消息ID的裁剪逻辑，
+使用真实的会话ID和消息配置来测试裁剪功能。
+
+支持的消息ID格式：
+1. 直接字段格式: {"message_id": "9226b3a4-1234-5678-9abc-def012345678"}
+2. Hint嵌入格式: {"content": "消息内容 [[message_id: 9226b3a4]]"}
+3. 兼容格式: {"content": "消息内容 message_id: 9226b3a4"}
+
+测试场景：
+1. 有消息ID配置的会话 - 应用消息ID裁剪
+2. 无消息ID配置的会话 - 跳过消息ID裁剪  
+3. 裁剪统计信息验证
+4. 成对裁剪逻辑验证
+5. 两种消息ID格式的混合验证
+"""
+
+import sys
+import os
+import json
+from datetime import datetime
+from typing import List, Dict, Any
+
+# 添加src路径到Python路径
+sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), '../../../..'))
+
+from autocoder.common.pruner.agentic_conversation_pruner import AgenticConversationPruner
+from autocoder.common.pruner.conversation_message_ids_api import get_conversation_message_ids_api
+from autocoder.common import AutoCoderArgs
+from autocoder.sdk import get_llm
+
+
+def create_test_conversations() -> List[Dict[str, Any]]:
+    """创建测试用的会话数据，包含我们已知要删除的消息ID
+    
+    展示两种消息ID格式的支持：
+    1. 直接字段格式（推荐）- 从message_id字段提取
+    2. Hint嵌入格式（兼容性）- 从content中提取
+    """
+    
+    # 根据之前的查询，我们知道要删除的消息ID是 642f04ee
+    # 我们创建一个包含这个ID的会话数据来测试
+    conversations = [
+        {
+            "role": "system",
+            "content": "You are a helpful assistant.",
+            "message_id": "12345678abcdef01"  # 格式1: 直接字段格式（完整UUID）
+        },
+        {
+            "role": "user", 
+            "content": "Hello, can you help me with some code? [[message_id: 87654321]]",  # 格式2: 标准hint格式
+            "message_id": "87654321fedcba02"  # 同时包含两种格式用于验证优先级
+        },
+        {
+            "role": "assistant",
+            "content": "Of course! I'd be happy to help you with your code. What do you need assistance with?",
+            "message_id": "642f04ee11223344"  # 格式1: 这个消息ID的前8位匹配要删除的 642f04ee
+        },
+        {
+            "role": "user",
+            "content": "I want to read a file and analyze it. message_id: 13579bdf",  # 格式2: 兼容hint格式
+            # 没有message_id字段，只从content中提取
+        },
+        {
+            "role": "assistant", 
+            "content": "I'll help you read and analyze the file.\n\n<read_file>\n<path>example.py</path>\n</read_file>",
+            "message_id": "97531eca"  # 格式1: 8字符短ID
+        },
+        {
+            "role": "user",
+            "content": "<tool_result tool_name='read_file' success='true'><message>File read successfully</message><content>def hello():\n    print('Hello, World!')\n    return 'success'</content></tool_result> [[message_id: abcdef01]]",  # 格式2: 标准hint格式
+            # 没有message_id字段，只从content中提取
+        },
+        {
+            "role": "assistant",
+            "content": "I can see the file contains a simple hello function. The code looks good!",
+            "message_id": "fedcba9876543210"  # 格式1: 完整UUID
+        }
+    ]
+    
+    return conversations
+
+
+def create_test_args() -> AutoCoderArgs:
+    """创建测试用的参数"""
+    return AutoCoderArgs(
+        source_dir=".",
+        conversation_prune_safe_zone_tokens=1000,  # 设置较小的值，强制触发裁剪逻辑
+        context_prune=True,
+        context_prune_strategy="extract",
+        query="测试消息ID裁剪逻辑"
+    )
+
+
+def print_conversations(conversations: List[Dict[str, Any]], title: str):
+    """打印会话列表，方便查看"""
+    print(f"\n📋 {title}")
+    print("=" * 80)
+    
+    for i, conv in enumerate(conversations):
+        role = conv.get("role", "unknown")
+        content = conv.get("content", "")
+        message_id = conv.get("message_id", "no_id")
+        
+        # 截断长内容
+        if len(content) > 100:
+            content_display = content[:97] + "..."
+        else:
+            content_display = content
+            
+        print(f"  {i+1:2d}. [{role:9}] {message_id[:8]} | {content_display}")
+
+
+def verify_pruning_with_message_ids():
+    """验证有消息ID配置的裁剪逻辑"""
+    
+    print("🔍 测试场景1: 验证有消息ID配置的会话裁剪")
+    print("=" * 80)
+    
+    # 使用已知的会话ID，该会话配置了要删除消息ID 642f04ee
+    conversation_id = "796e1c2e-1ab8-46d1-9448-f9bd29d5c095"
+    
+    # 创建测试数据
+    original_conversations = create_test_conversations()
+    args = create_test_args()
+    llm = get_llm("v3_chat", product_mode="lite")
+    
+    # 打印原始会话
+    print_conversations(original_conversations, "原始会话数据")
+    
+    # 查询消息ID配置
+    api = get_conversation_message_ids_api()
+    message_ids_config = api.get_conversation_message_ids(conversation_id)
+    
+    if message_ids_config:
+        print(f"\n✅ 找到消息ID配置:")
+        print(f"   会话ID: {message_ids_config.conversation_id}")
+        print(f"   要删除的消息ID: {message_ids_config.message_ids}")
+        print(f"   保持成对删除: {message_ids_config.preserve_pairs}")
+    else:
+        print(f"❌ 未找到会话ID {conversation_id} 的消息ID配置")
+        return False
+    
+    # 创建裁剪器并执行裁剪
+    from autocoder.common.conversations.get_conversation_manager import get_conversation_manager
+    
+    # 不设置 conversation manager，因为我们直接通过构造函数传递 conversation_id
+    
+    pruner = AgenticConversationPruner(args=args, llm=llm, conversation_id=conversation_id)
+    
+    # 调试：检查消息ID提取和验证
+    print(f"\n🔍 调试信息:")
+    print(f"   测试数据中的消息ID:")
+    for i, conv in enumerate(original_conversations):
+        message_id = conv.get("message_id", "no_id")
+        extracted_id = message_id[:8] if len(message_id) >= 8 else "invalid"
+        print(f"     {i+1}. {message_id} -> {extracted_id}")
+    
+    print(f"   配置中要删除的消息ID: {message_ids_config.message_ids}")
+    
+    # 手动验证消息ID匹配
+    conversation_msg_ids = [conv.get("message_id", "")[:8] for conv in original_conversations if len(conv.get("message_id", "")) >= 8]
+    print(f"   从对话中提取的消息ID: {conversation_msg_ids}")
+    
+    # 检查验证逻辑
+    api = get_conversation_message_ids_api()
+    validation_result = api.validate_message_ids(
+        message_ids_config.message_ids, 
+        [conv.get("message_id", "") for conv in original_conversations]
+    )
+    print(f"   消息ID验证结果: {validation_result.is_valid}")
+    if not validation_result.is_valid:
+        print(f"   验证失败原因: {validation_result.error_message}")
+    if validation_result.warnings:
+        for warning in validation_result.warnings:
+            print(f"   验证警告: {warning}")
+    
+    print(f"\n🚀 开始执行裁剪...")
+    
+    # 首先计算token数，看看是否需要裁剪
+    from autocoder.common.tokens import count_string_tokens
+    import json
+    original_tokens = count_string_tokens(json.dumps(original_conversations, ensure_ascii=False))
+    print(f"   原始对话token数: {original_tokens}")
+    print(f"   安全区域阈值: {args.conversation_prune_safe_zone_tokens}")
+    print(f"   是否超过阈值: {'是' if original_tokens > args.conversation_prune_safe_zone_tokens else '否'}")
+    
+    if original_tokens <= args.conversation_prune_safe_zone_tokens:
+        print(f"⚠️ 注意: token数未超过阈值，可能不会触发裁剪逻辑")
+    
+    # 直接测试 _apply_message_ids_pruning 方法
+    print(f"\n🔧 直接测试 _apply_message_ids_pruning 方法:")
+    try:
+        range_pruned = pruner._apply_message_ids_pruning(original_conversations)
+        print(f"   Message IDs pruning 结果: {len(original_conversations)} -> {len(range_pruned)} 消息")
+        
+        if len(range_pruned) < len(original_conversations):
+            print(f"   ✅ Message IDs pruning 成功删除了消息")
+            # 检查哪些消息被删除了
+            original_ids = [conv.get("message_id", "")[:8] for conv in original_conversations]
+            pruned_ids = [conv.get("message_id", "")[:8] for conv in range_pruned]
+            deleted_ids = [msg_id for msg_id in original_ids if msg_id not in pruned_ids]
+            print(f"   删除的消息ID: {deleted_ids}")
+        else:
+            print(f"   ⚠️ Message IDs pruning 没有删除任何消息")
+    except Exception as e:
+        print(f"   ❌ Message IDs pruning 测试失败: {e}")
+        import traceback
+        traceback.print_exc()
+    
+    # 执行完整的裁剪流程
+    pruned_conversations = pruner.prune_conversations(original_conversations)
+    
+    # 打印裁剪后的会话
+    print_conversations(pruned_conversations, "裁剪后的会话数据")
+    
+    # 验证裁剪结果
+    print(f"\n📊 裁剪结果分析:")
+    print(f"   原始消息数量: {len(original_conversations)}")
+    print(f"   裁剪后数量: {len(pruned_conversations)}")
+    print(f"   删除的消息数: {len(original_conversations) - len(pruned_conversations)}")
+    
+    # 检查特定消息ID是否被删除
+    pruned_message_ids = [conv.get("message_id", "")[:8] for conv in pruned_conversations]
+    target_message_id = "642f04ee"
+    
+    if target_message_id in pruned_message_ids:
+        print(f"❌ 错误: 消息ID {target_message_id} 应该被删除但仍然存在")
+        return False
+    else:
+        print(f"✅ 正确: 消息ID {target_message_id} 已成功删除")
+    
+    # 获取裁剪统计信息
+    stats = pruner.get_pruning_statistics()
+    print(f"\n📈 裁剪统计信息:")
+    print(f"   消息ID裁剪已应用: {stats['range_pruning']['applied']}")
+    print(f"   消息ID裁剪成功: {stats['range_pruning']['success']}")
+    print(f"   会话ID: {stats['range_pruning']['conversation_id']}")
+    print(f"   总压缩比: {stats['compression']['total_compression_ratio']:.2%}")
+    
+    # 验证成对删除逻辑（如果启用）
+    if message_ids_config.preserve_pairs:
+        print(f"\n🔗 成对删除逻辑验证:")
+        # 分析是否正确处理了用户/助手配对
+        user_assistant_pairs = []
+        i = 0
+        while i < len(pruned_conversations):
+            if (i + 1 < len(pruned_conversations) and 
+                pruned_conversations[i].get("role") == "user" and 
+                pruned_conversations[i + 1].get("role") == "assistant"):
+                user_assistant_pairs.append((i, i + 1))
+                i += 2
+            else:
+                i += 1
+        
+        print(f"   检测到 {len(user_assistant_pairs)} 个用户/助手配对")
+        print(f"   配对完整性: {'✅ 良好' if len(user_assistant_pairs) > 0 else '⚠️ 需要检查'}")
+    
+    return True
+
+
+def verify_pruning_without_message_ids():
+    """验证无消息ID配置的裁剪逻辑"""
+    
+    print("\n🔍 测试场景2: 验证无消息ID配置的会话裁剪")
+    print("=" * 80)
+    
+    # 使用一个不存在的会话ID
+    conversation_id = "nonexistent-conversation-id-12345"
+    
+    # 创建测试数据
+    original_conversations = create_test_conversations()
+    args = create_test_args()
+    llm = get_llm("v3_chat", product_mode="lite")
+    
+    # 创建裁剪器并执行裁剪
+    from autocoder.common.conversations.get_conversation_manager import get_conversation_manager
+    
+    # 不设置 conversation manager，因为我们直接通过构造函数传递 conversation_id
+    
+    pruner = AgenticConversationPruner(args=args, llm=llm, conversation_id=conversation_id)
+    
+    print(f"🚀 开始执行裁剪 (无消息ID配置)...")
+    pruned_conversations = pruner.prune_conversations(original_conversations)
+    
+    # 验证结果
+    print(f"\n📊 裁剪结果分析:")
+    print(f"   原始消息数量: {len(original_conversations)}")
+    print(f"   裁剪后数量: {len(pruned_conversations)}")
+    
+    # 获取统计信息
+    stats = pruner.get_pruning_statistics()
+    print(f"\n📈 裁剪统计信息:")
+    print(f"   消息ID裁剪已应用: {stats['range_pruning']['applied']}")
+    print(f"   消息ID裁剪成功: {stats['range_pruning']['success']}")
+    
+    # 验证没有应用消息ID裁剪
+    if not stats['range_pruning']['applied']:
+        print(f"✅ 正确: 无消息ID配置时跳过了消息ID裁剪")
+        return True
+    else:
+        print(f"❌ 错误: 无消息ID配置时不应该应用消息ID裁剪")
+        return False
+
+
+def verify_tool_cleanup_logic():
+    """验证工具输出清理逻辑"""
+    
+    print("\n🔍 测试场景3: 验证工具输出清理逻辑")
+    print("=" * 80)
+    
+    # 创建包含大量工具输出的会话数据
+    large_tool_output = "def example_function():\n    pass\n" * 100  # 创建大量内容
+    
+    conversations_with_tools = [
+        {
+            "role": "user",
+            "content": "Please read a file",
+            "message_id": "tool_test_001"
+        },
+        {
+            "role": "assistant", 
+            "content": "I'll read the file for you.\n\n<read_file>\n<path>example.py</path>\n</read_file>",
+            "message_id": "tool_test_002"
+        },
+        {
+            "role": "user",
+            "content": f"<tool_result tool_name='read_file' success='true'><message>File read successfully</message><content>{large_tool_output}</content></tool_result>",
+            "message_id": "tool_test_003"
+        },
+        {
+            "role": "assistant",
+            "content": "I can see the file content. It contains many function definitions.",
+            "message_id": "tool_test_004"
+        }
+    ]
+    
+    # 使用小的token阈值来触发工具清理
+    args = AutoCoderArgs(
+        source_dir=".",
+        conversation_prune_safe_zone_tokens=1000,  # 小阈值触发清理
+        context_prune=True,
+        context_prune_strategy="extract",
+        query="测试工具清理逻辑"
+    )
+    
+    llm = get_llm("v3_chat", product_mode="lite")
+    
+    # 不使用消息ID配置，只测试工具清理
+    # 为工具清理测试使用一个测试会话ID
+    test_conversation_id = "tool-cleanup-test-conversation"
+    pruner = AgenticConversationPruner(args=args, llm=llm, conversation_id=test_conversation_id)
+    
+    print(f"🚀 开始执行工具输出清理...")
+    
+    # 计算原始token数
+    from autocoder.common.tokens import count_string_tokens
+    original_tokens = count_string_tokens(json.dumps(conversations_with_tools, ensure_ascii=False))
+    print(f"   原始token数: {original_tokens}")
+    
+    pruned_conversations = pruner.prune_conversations(conversations_with_tools)
+    
+    # 计算清理后token数
+    final_tokens = count_string_tokens(json.dumps(pruned_conversations, ensure_ascii=False))
+    print(f"   清理后token数: {final_tokens}")
+    print(f"   Token减少: {original_tokens - final_tokens} ({((original_tokens - final_tokens) / original_tokens * 100):.1f}%)")
+    
+    # 检查工具结果是否被清理
+    tool_result_cleaned = False
+    for conv in pruned_conversations:
+        content = conv.get("content", "")
+        if "tool_result" in content and "Content cleared to save tokens" in content:
+            tool_result_cleaned = True
+            break
+    
+    if tool_result_cleaned:
+        print(f"✅ 正确: 工具输出已被清理")
+        return True
+    else:
+        print(f"⚠️ 注意: 工具输出可能未被清理（可能因为token数仍在阈值内）")
+        return True
+
+
+def main():
+    """主测试函数"""
+    print("🧪 AgenticConversationPruner 消息ID裁剪逻辑验证")
+    print(f"📅 测试时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
+    print("=" * 80)
+    
+    test_results = []
+    
+    try:
+        # 测试场景1: 有消息ID配置的裁剪
+        result1 = verify_pruning_with_message_ids()
+        test_results.append(("有消息ID配置的裁剪", result1))
+        
+        # 测试场景2: 无消息ID配置的裁剪  
+        result2 = verify_pruning_without_message_ids()
+        test_results.append(("无消息ID配置的裁剪", result2))
+        
+        # 测试场景3: 工具输出清理
+        result3 = verify_tool_cleanup_logic()
+        test_results.append(("工具输出清理逻辑", result3))
+        
+    except Exception as e:
+        print(f"❌ 测试过程中发生错误: {str(e)}")
+        import traceback
+        traceback.print_exc()
+        return False
+    
+    # 汇总测试结果
+    print(f"\n🎯 测试结果汇总")
+    print("=" * 80)
+    
+    all_passed = True
+    for test_name, result in test_results:
+        status = "✅ 通过" if result else "❌ 失败"
+        print(f"   {test_name}: {status}")
+        if not result:
+            all_passed = False
+    
+    print(f"\n{'🎉 所有测试通过!' if all_passed else '⚠️ 部分测试失败，请检查上述结果'}")
+    
+    return all_passed
+
+
+if __name__ == "__main__":
+    success = main()
+    sys.exit(0 if success else 1)