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

autocoder/{agent → index/filter}/agentic_filter.py

@@ -8,29 +8,30 @@ from typing import List, Dict, Any, Union, Callable, Optional
 from autocoder.common.printer import Printer
 from rich.console import Console
 from rich.panel import Panel
-from pydantic import SkipValidation
 
 from autocoder.common.result_manager import ResultManager
 from autocoder.utils.auto_coder_utils.chat_stream_out import stream_out
 from byzerllm.utils.str2model import to_model
 from autocoder.common import git_utils
 from autocoder.commands.tools import AutoCommandTools
-from autocoder.auto_coder import AutoCoderArgs
+from autocoder.common import AutoCoderArgs
+from autocoder.common.autocoderargs_parser import AutoCoderArgsParser
 from autocoder.common import detect_env
 from autocoder.common import shells
 from loguru import logger
 from autocoder.utils import llms as llms_utils
-from autocoder.rag.token_counter import count_tokens
+from autocoder.common.tokens import count_string_tokens as count_tokens
 from autocoder.common.global_cancel import global_cancel
 from autocoder.common.auto_configure import config_readme
 from autocoder.utils.auto_project_type import ProjectTypeAnalyzer
 from rich.text import Text
-from autocoder.common.mcp_server import get_mcp_server, McpServerInfoRequest
+from autocoder.common.mcp_tools.server import get_mcp_server, McpServerInfoRequest
 from autocoder.common.action_yml_file_manager import ActionYmlFileManager
 from autocoder.events.event_manager_singleton import get_event_manager
 from autocoder.events import event_content as EventContentCreator
 from autocoder.run_context import get_run_context
 from autocoder.common.stream_out_type import AgenticFilterStreamOutType
+from autocoder.common.core_config import get_memory_manager
 
 
 class AgenticFilterRequest(BaseModel):
@@ -43,8 +44,8 @@ class FileOperation(BaseModel):
 
 
 class AgenticFilterResponse(BaseModel):
-    files: List[FileOperation]  # 文件列表，包含path和operation字段
-    reasoning: str  # 决策过程说明
+    files: List[FileOperation]  # File list containing path and operation fields
+    reasoning: str  # Decision process explanation
 
 
 class CommandSuggestion(BaseModel):
@@ -63,53 +64,12 @@ class AutoCommandRequest(BaseModel):
     user_input: str
 
 
-class MemoryConfig(BaseModel):
-    """
-    A model to encapsulate memory configuration and operations.
-    """
-
-    memory: Dict[str, Any]
-    save_memory_func: SkipValidation[Callable]
-
-    class Config:
-        arbitrary_types_allowed = True
-
-
-class CommandConfig(BaseModel):
-    coding: SkipValidation[Callable]
-    chat: SkipValidation[Callable]
-    add_files: SkipValidation[Callable]
-    remove_files: SkipValidation[Callable]
-    index_build: SkipValidation[Callable]
-    index_query: SkipValidation[Callable]
-    list_files: SkipValidation[Callable]
-    ask: SkipValidation[Callable]
-    revert: SkipValidation[Callable]
-    commit: SkipValidation[Callable]
-    help: SkipValidation[Callable]
-    exclude_dirs: SkipValidation[Callable]
-    summon: SkipValidation[Callable]
-    design: SkipValidation[Callable]
-    mcp: SkipValidation[Callable]
-    models: SkipValidation[Callable]
-    lib: SkipValidation[Callable]
-    execute_shell_command: SkipValidation[Callable]
-    generate_shell_command: SkipValidation[Callable]
-    conf_export: SkipValidation[Callable]
-    conf_import: SkipValidation[Callable]
-    index_export: SkipValidation[Callable]
-    index_import: SkipValidation[Callable]
-    exclude_files: SkipValidation[Callable]
-
-
 class AgenticFilter:
     def __init__(
         self,
         llm: Union[byzerllm.ByzerLLM, byzerllm.SimpleByzerLLM],
         conversation_history: List[Dict[str, Any]],
         args: AutoCoderArgs,
-        memory_config: MemoryConfig,
-        command_config: Optional[CommandConfig] = None,
     ):
         self.llm = llm
         self.args = args
@@ -119,9 +79,11 @@ class AgenticFilter:
         # Use existing args for max iterations
         self.max_iterations = args.auto_command_max_iterations
         self.conversation_history = conversation_history
-        self.memory_config = memory_config
-        self.command_config = command_config
-        self.project_type_analyzer = ProjectTypeAnalyzer(args=args, llm=self.llm)
+        self.project_type_analyzer = ProjectTypeAnalyzer(
+            args=args, llm=self.llm)
+
+        # Initialize AutoCoderArgs parser for flexible parameter parsing
+        self.args_parser = AutoCoderArgsParser()
         try:
             self.mcp_server = get_mcp_server()
             mcp_server_info_response = self.mcp_server.send_request(
@@ -134,94 +96,156 @@ class AgenticFilter:
         except Exception as e:
             logger.error(f"Error getting MCP server info: {str(e)}")
             self.mcp_server_info = ""
+        self.memory_config = get_memory_manager()
+
+    def _get_parsed_safe_zone_tokens(self) -> int:
+        """
+        Parse conversation_prune_safe_zone_tokens parameter, supports multiple formats
+
+        Returns:
+            Parsed token count
+        """
+        return self.args_parser.parse_conversation_prune_safe_zone_tokens(
+            self.args.conversation_prune_safe_zone_tokens,
+            self.args.code_model
+        )
 
     @byzerllm.prompt()
     def _analyze(self, request: AgenticFilterRequest) -> str:
         """        
-        ## 当前用户环境信息如下:
+        ## Current User Environment Information:
         <os_info>
-        操作系统: {{ env_info.os_name }} {{ env_info.os_version }}
-        操作系统发行版: {{ os_distribution }}
-        Python版本: {{ env_info.python_version }}
-        终端类型: {{ env_info.shell_type }}
-        终端编码: {{ env_info.shell_encoding }}
-        当前用户: {{ current_user }}
+        Operating System: {{ env_info.os_name }} {{ env_info.os_version }}
+        OS Distribution: {{ os_distribution }}
+        Python Version: {{ env_info.python_version }}
+        Terminal Type: {{ env_info.shell_type }}
+        Terminal Encoding: {{ env_info.shell_encoding }}
+        Current User: {{ current_user }}
 
         {%- if shell_type %}
-        脚本类型：{{ shell_type }}
+        Script Type: {{ shell_type }}
         {%- endif %}
 
         {%- if env_info.conda_env %}
-        Conda环境: {{ env_info.conda_env }}
+        Conda Environment: {{ env_info.conda_env }}
         {%- endif %}
         {%- if env_info.virtualenv %}
-        虚拟环境: {{ env_info.virtualenv }}
+        Virtual Environment: {{ env_info.virtualenv }}
         {%- endif %}
         </os_info>
 
-        当前项目根目录：
+        Current Project Root Directory:
         {{ current_project }}
 
         {% if current_files %}
-        ## 当前用户手动添加关注的文件列表：
+        ## Current User Manually Added File List:
         <current_files>
         {% for file in current_files %}
         - {{ file }}
         {% endfor %}
         </current_files>
         {% endif %}
+        
+        ## Available Function List:
+        {{ available_commands }}  
+
+        ====        
+
+        AC MOD FILES
+
+        # AC Modules (.ac.mod.md)         
+
+        ## What is an AC Module?
+
+        Any directory containing a `.ac.mod.md` file is considered an AC Module - a language-agnostic module that provides complete functionality and can be used as an API. 
+        These modules are self-contained units with well-defined interfaces and comprehensive documentation. AC Modules  basically should not large than 60K tokens, 
+        you can use count_tokens tool to get the tokens number of the module.                 
+
+        ## AC Module Structure        
+        - .ac.mod.md contains detailed information about:
+            - Module overview and purpose: One-sentence description of core functionality
+            - Directory structure: Complete file tree with functionality descriptions
+            - Quick start guide: Basic usage examples and code snippets
+            - Core components: Main classes, methods, and their responsibilities (make it simple)
+            - Sequence diagrams: Visual flow showing component interactions and calling relationships
+            - Dependency graphs: Internal dependencies and relationships between components
+            - External dependencies: References to other AC modules it depends on
+            - Configuration management: Setup examples and configuration files
+            - Testing instructions: Executable commands and test examples            
+
+        ## When to Use AC Modules
+
+        1. **Avoid duplicate implementation**: Check if functionality already exists in project AC modules before implementing new features
+        2. **Project understanding**: Review multiple AC modules to gain comprehensive knowledge of the entire project architecture
+        3. **File modification context**: When modifying files in a directory, check if it's an AC module or contains AC modules to understand the full impact
 
-        ## 可用函数列表:
-        {{ available_commands }}
+        ## execute_shell_command
 
-        ## 当前大模型窗口安全值
-        {{ conversation_safe_zone_tokens }}
+        When to use:
+        - When you want to list all AC modules in the project
+        - When you want to list all AC modules in a specific directory
 
-        ## Token 安全区
-        对话和文件内容的总Token数不应超过 {{ conversation_safe_zone_tokens }}。请谨慎读取大文件。  
+        Example:
+
+        execute_shell_command(command="find . -name '*.ac.mod.md' -type f | sed 's|/[^/]*$||' | sort | uniq")
+
+
+        ## read_files
+
+        When to use:
+        - Use the this tool to retrieve comprehensive information about an AC module
+        - The tool reads the `.ac.mod.md` file and provides structured information about the module        
+
+        Example:
+
+        read_files(paths="src/autocoder/agent/.ac.mod.md")        
+               
+       ====     
         
-        ## 对话历史
+        {% if conversation_history %}
+        ## Conversation History
         <conversation_history>
         {% for msg in conversation_history %}
         **{{ msg.role }}**: {{ msg.content }}
         {% endfor %}
         </conversation_history>
+        {% endif %}
         
-        ## 完成任务的一些实践指导
+        ## Practical Guidance for Task Completion
         {{ command_combination_readme }}
-        
-        ## 你的任务以及要求
-        你是一个代码分析专家，需要根据用户的需求分析项目中相关的文件。你的工作是确定：
 
-        1. **需要修改的文件**（标记为"MODIFY"）：直接需要更改的文件
-        2. **需要参考的文件**（标记为"REFERENCE"）：理解需求或实现修改所需参考的文件
-        3. **需要新增的文件**（标记为"ADD"）：实现需求可能需要创建的新文件
-        4. **需要删除的文件**（标记为"REMOVE"）：实现需求可能需要删除的文件
+        ## Your Task and Requirements
+        You are a code analysis expert who needs to analyze project-related files based on user requirements. Your job is to determine:
+
+        1. **Files to Modify** (marked as "MODIFY"): Files that need direct changes
+        2. **Files to Reference** (marked as "REFERENCE"): Files needed for understanding requirements or implementing modifications
+        3. **Files to Add** (marked as "ADD"): New files that may need to be created to fulfill requirements
+        4. **Files to Remove** (marked as "REMOVE"): Files that may need to be deleted to fulfill requirements
 
-        请通过以下步骤进行分析：
-        1. 理解用户需求的核心目标
-        2. 使用提供的工具函数探索项目结构
-        3. 分析相关文件的内容和依赖关系
-        4. 确定需要修改、参考、新增或删除的文件列表
+        Please analyze through the following steps:
+        1. Understand the core objectives of user requirements
+        2. Use the provided tool functions to explore project structure
+        3. Analyze the content and dependencies of relevant files
+        4. Determine the list of files to modify, reference, add, or remove
 
-        ## 返回格式要求
-        返回格式必须是严格的JSON格式：
+        ## Return Format Requirements
+        The return format must be strict JSON format:
 
         ```json
         {
             "suggestions": [
                 {
-                    "command": "函数名称",
+                    "command": "function_name",
                     "parameters": {},
                     "confidence": 0.9,
-                    "reasoning": "推荐理由"
+                    "reasoning": "recommendation_reason"
                 }
             ],
-            "reasoning": "整体推理说明"
+            "reasoning": "overall_reasoning_explanation"
         }
         ```
 
-        请返回第一个建议的函数调用。我会将每个函数的执行结果提供给你，然后你可以根据这些结果确定下一步要执行的函数，直到完成分析。
+        Please return the first suggested function call. I will provide you with the execution results of each function, and then you can determine the next function to execute based on these results until the analysis is complete.
         """
         env_info = detect_env()
         shell_type = "bash"
@@ -231,37 +255,37 @@ class AgenticFilter:
             shell_type = "powershell"
         return {
             "user_input": request.user_input,
-            "current_files": self.memory_config.memory["current_files"]["files"],
+            "current_files": self.memory_config.get_current_files(),
             "conversation_history": self.conversation_history,
             "available_commands": self._command_readme.prompt(),
-            "current_conf": json.dumps(self.memory_config.memory["conf"], indent=2),
+            "current_conf": json.dumps(self.memory_config.get_all_config(), indent=2),
             "env_info": env_info,
             "shell_type": shell_type,
             "shell_encoding": shells.get_terminal_encoding(),
-            "conversation_safe_zone_tokens": self.args.conversation_prune_safe_zone_tokens,
+            "conversation_safe_zone_tokens": self._get_parsed_safe_zone_tokens(),
             "os_distribution": shells.get_os_distribution(),
             "current_user": shells.get_current_username(),
             "command_combination_readme": self._command_combination_readme.prompt(
                 user_input=request.user_input
             ),
-            "current_project": os.path.abspath(self.args.source_dir),
-        }
+            "current_project": os.path.abspath(self.args.source_dir or "."),
+        }  # type: ignore
 
     @byzerllm.prompt()
     def _command_combination_readme(self, user_input: str) -> str:
         """
-        ## 操作流程建议
-        1.  **理解需求**: 分析用户输入 `{{ user_input }}`。
-        2.  **探索项目**:
-            *   使用 `list_files` 层层递进了解项目结构，用来确定需要关注的文件。
-            *   如果用户提及文件名等，则可以使用 `find_files_by_name` 或，如果用户提到关键字则可以使用 `find_files_by_content` 定位可能相关的文件。
-            *   如果用户提到了具体的符号（函数，类名等）则可以使用 `get_project_map` 获取候选文件的详细信息（如符号）。
-        3.  **深入分析**:
-            *   使用 `read_files` 读取关键文件的内容进行确认。如果文件过大，使用 `line_ranges` 参数分段读取。
-            *   如有必要，使用 `run_python` 或 `execute_shell_command` 执行代码或命令进行更复杂的分析。
-        4.  **迭代决策**: 根据工具的返回结果，你可能需要多次调用不同的工具来逐步缩小范围或获取更多信息。        
-        6.  **最终响应**: 当你确定了所有需要参考和修改的文件后，**必须**调用 `output_result` 工具，并提供符合其要求格式的JSON字符串作为其 `response` 参数。
-            该json格式要求为：
+        ## Recommended Operation Flow
+        1.  **Understand Requirements**: Analyze user input `{{ user_input }}`.
+        2.  **Explore Project**:
+            *   List all current AC modules using `execute_shell_command` to enumerate them, then read `.ac.mod.md` files to understand the purpose of each AC module.
+            *   Use `list_files` to progressively understand project structure and determine files that need attention.
+            *   If user mentions file names, use `find_files_by_name`; if user mentions keywords, use `find_files_by_content` to locate potentially relevant files.            
+        3.  **In-depth Analysis**:
+            *   Use `read_files` to read key file contents for confirmation. If files are too large, use `line_ranges` parameter for segmented reading.
+            *   If necessary, use `execute_shell_command` to execute code or commands for more complex analysis.
+        4.  **Iterative Decision Making**: Based on tool return results, you may need to call different tools multiple times to gradually narrow down scope or obtain more information.        
+        6.  **Final Response**: When you have determined all files that need to be referenced and modified, you **must** call the `output_result` tool and provide a JSON string that meets its format requirements as the `response` parameter.
+            The required JSON format is:
             ```json
             {
                 "files": [
@@ -270,36 +294,33 @@ class AgenticFilter:
                     {"path": "/path/to/new_file.txt", "operation": "ADD"},
                     {"path": "/path/to/old_file.log", "operation": "REMOVE"}
                 ],
-                "reasoning": "详细说明你是如何通过分析和使用工具得出这个文件列表的。"
+                "reasoning": "Detailed explanation of how you analyzed and used tools to arrive at this file list."
             }
             ```
-        
-        {% if enable_active_context %}
-        ** 非常非常重要的提示 **
-        每一个目录都有一个描述信息，比如 {{ project_path }}/src/abc/bbc 的目录描述信息会放在 {{ project_path }}/.auto-coder/active-context/src/abc/bbc/active.md 文件中。
-        你可以使用 read_files 函数读取，从而帮助你更好的挑选要详细阅读哪个文件。值得注意的是，active.md 并不会包含该目录下所有的文件信息，只保存最近发生变更的文件的信息。
-        {% endif %}
-        """        
+
+        ** Very Important Note **
+        You should try to read the .ac.mod.md file for every directory to better understand the directory content. For example, directory description information for {{ project_path }}/src/abc/bbc would be located in {{ project_path }}/src/abc/bbc/.ac.mod.md file.
+        You can use the read_files function to read these files, helping you better choose which files to read in detail.        
+        """
         return {
-            "project_path": os.path.abspath(self.args.source_dir),
-            "enable_active_context": self.args.enable_active_context,
-        }
+            "project_path": os.path.abspath(self.args.source_dir or ".")
+        }  # type: ignore
 
     @byzerllm.prompt()
     def _execute_command_result(self, result: str) -> str:
         """
-        根据函数执行结果，返回下一个函数。
+        Based on function execution results, return the next function.
 
-        下面是我们上一个函数执行结果:
+        Below is our previous function execution result:
 
         <function_result>
         {{ result }}
         </function_result>
 
-        请根据命令执行结果以及前面的对话，返回下一个函数。
+        Please return the next function based on the command execution results and previous conversation.
 
-        *** 非常非常重要的提示 ***
-        1. 如果你认为已经收集到足够信息来确定最终的文件列表，请务必调用 `output_result` 并以如下格式要求的JSON字符串作为 `response` 参数。最多允许 {{ max_iterations }} 次工具调用。
+        *** Very Important Notes ***
+        1. If you think you have collected enough information to determine the final file list, you must call `output_result` with a JSON string in the required format as the `response` parameter. Maximum {{ max_iterations }} tool calls allowed.
             ```json
             {
                 "files": [
@@ -308,114 +329,46 @@ class AgenticFilter:
                     {"path": "/path/to/new_file.txt", "operation": "ADD"},
                     {"path": "/path/to/old_file.log", "operation": "REMOVE"}
                 ],
-                "reasoning": "详细说明你是如何通过分析和使用工具得出这个文件列表的。"
+                "reasoning": "Detailed explanation of how you analyzed and used tools to arrive at this file list."
             }
             ```
-        2. 你最多尝试 {{ auto_command_max_iterations }} 次，如果 {{ auto_command_max_iterations }} 次都没有满足要求，则不要返回任何函数，确保 suggestions 为空。
+        2. You can try at most {{ auto_command_max_iterations }} times. If {{ auto_command_max_iterations }} attempts do not meet requirements, do not return any function and ensure suggestions is empty.
         """
         return {
             "auto_command_max_iterations": self.args.auto_command_max_iterations,
-            "conversation_safe_zone_tokens": self.args.conversation_prune_safe_zone_tokens,
-        }
+            "conversation_safe_zone_tokens": self._get_parsed_safe_zone_tokens(),
+        }  # type: ignore
 
     @byzerllm.prompt()
     def _command_readme(self) -> str:
         """
-        你有如下函数可供使用：
-        <commands>
-        <name>ask_user</name>
-        <description>
-        如果你对用户的问题有什么疑问，或者你想从用户收集一些额外信息，可以调用此方法。
-        输入参数 question 是你对用户的提问。
-        返回值是 用户对你问题的回答。
-        ** 如果你的问题比较多，建议一次就问一个，然后根据用户回答再问下一个。 **
-        </description>
-        <usage>
-         该命令接受一个参数 question，为需要向用户询问的问题字符串。
-
-         使用例子：
-         ask_user(question="请输入火山引擎的 R1 模型推理点")
-        </command>
-
-        <command>
-        <name>run_python</name>
-        <description>运行指定的Python代码。主要用于执行一些Python脚本或测试代码。</description>
-        <usage>
-         该命令接受一个参数 code，为要执行的Python代码字符串。
-
-         使用例子：
-
-         run_python(code="print('Hello World')")
-
-         注意：
-         - 代码将在项目根目录下执行
-         - 可以访问项目中的所有文件
-         - 输出结果会返回给用户
-        </usage>
-        </command>
-
+        You have the following functions available:
+        <commands>        
         <command>
         <name>execute_shell_command</name>
-        <description>运行指定的Shell脚本。主要用于编译、运行、测试等任务。</description>
+        <description>Execute specified shell scripts. Mainly used for compilation, running, testing, and other tasks.</description>
         <usage>
-         该命令接受一个参数 command，为要执行的Shell脚本字符串。
+         This command accepts one parameter 'command', which is the shell script string to execute.
 
 
-         使用例子：
+         Usage example:
 
          execute_shell_command(command="ls -l")
 
-         注意：
-         - 脚本将在项目根目录下执行
-         - 禁止执行包含 rm 命令的脚本
-         - 输出结果会返回给用户
-         - 执行该命令的时候，需要通过 ask_user 询问用户是否同意执行，如果用户拒绝，则不再执行当前想执行的脚本呢。
-        </usage>
-        </command>
-
-        <command>
-        <name>generate_shell_command</name>
-        <description>
-        根据用户需求描述，生成shell脚本。
-        </description>
-        <usage>
-          支持的参数名为 input_text， 字符串类型，用户的需求，使用该函数，会打印生成结果，用户可以更加清晰
-          的看到生成的脚本。然后配合 ask_user, execute_shell_command 两个函数，最终完成
-          脚本执行。
-        </usage>
-        </command>
-
-        <command>
-        <name>get_project_map</name>
-        <description>返回项目中指定文件包括文件用途、导入的包、定义的类、函数、变量等。</description>
-        <usage>
-         该命令接受一个参数 file_paths，路径list,或者是以逗号分割的多个文件路径。
-         路径支持相对路径和绝对路径。
-
-         使用例子：
-
-         get_project_map(file_paths=["full/path/to/main.py","partial/path/to/utils.py"])，
-
-         或者：
-
-         get_project_map(file_paths="full/path/to/main.py,partial/path/to/utils.py")
-
-         该函数特别适合你想要了解某个文件的用途，以及该文件的导入的包，定义的类，函数，变量等信息。
-         同时，你还能看到文件的大小（tokens数），以及索引的大小（tokens数），以及构建索引花费费用等信息。
-         如果你觉得该文件确实是你关注的，你可以通过 read_files 函数来读取文件完整内容，从而帮你做更好的决策。
-
-         注意：
-         - 返回值为JSON格式文本
-         - 只能返回已被索引的文件
+         Notes:
+         - Scripts will be executed in the project root directory
+         - Scripts containing rm commands are prohibited
+         - Output results will be returned to the user
+         - When executing this command, you need to ask the user through ask_user whether they agree to execute it. If the user refuses, do not execute the script.
         </usage>
-        </command>
+        </command>                
 
         <command>
         <name>list_files</name>
-        <description>list_files 查看某个目录下的所有文件</description>
+        <description>list_files views all files in a specific directory</description>
         <usage>
-         该命令接受一个参数 path, 为要查看的目录路径。
-         使用例子：
+         This command accepts one parameter 'path', which is the directory path to view.
+         Usage example:
          list_files(path="path/to/dir")
 
         </usage>
@@ -423,117 +376,117 @@ class AgenticFilter:
 
         <command>
         <name>read_files</name>
-        <description>读取指定文件的内容（支持指定行范围），支持文件名或绝对路径。</description>
+        <description>Read specified file content (supports line range specification), supports filename or absolute path.</description>
         <usage>
-        该函数用于读取指定文件的内容。
+        This function is used to read specified file content.
 
-        参数说明:
+        Parameter description:
         1. paths (str):
-           - 以逗号分隔的文件路径列表
-           - 支持两种格式:
-             a) 文件名: 如果多个文件匹配该名称，将选择第一个匹配项
-             b) 绝对路径: 直接指定文件的完整路径
-           - 示例: "main.py,utils.py" 或 "/path/to/main.py,/path/to/utils.py"
-           - 建议: 每次调用推荐一个文件，最多不要超过3个文件。
+           - Comma-separated list of file paths
+           - Supports two formats:
+             a) Filename: If multiple files match this name, the first match will be selected
+             b) Absolute path: Directly specify the complete file path
+           - Example: "main.py,utils.py" or "/path/to/main.py,/path/to/utils.py"
+           - Recommendation: Each call should read one file, maximum of 3 files.
 
         2. line_ranges (Optional[str]):
-           - 可选参数，用于指定每个文件要读取的具体行范围
-           - 格式说明:
-             * 使用逗号分隔不同文件的行范围
-             * 每个文件可以指定多个行范围，用/分隔
-             * 每个行范围使用-连接起始行和结束行
-           - 示例:
-             * "1-100,2-50" (为两个文件分别指定一个行范围)
-             * "1-100/200-300,50-100" (第一个文件指定两个行范围，第二个文件指定一个行范围)
-           - 注意: line_ranges中的文件数量必须与paths中的文件数量一致，否则会抛出错误
-
-        返回值:
-        - 返回str类型，包含所有请求文件的内容
-        - 每个文件内容前会标注文件路径和行范围信息（如果指定了行范围）
-
-        使用例子：
+           - Optional parameter to specify the specific line ranges to read for each file
+           - Format description:
+             * Use commas to separate line ranges for different files
+             * Each file can specify multiple line ranges, separated by /
+             * Each line range uses - to connect start and end lines
+           - Examples:
+             * "1-100,2-50" (specify one line range for each of two files)
+             * "1-100/200-300,50-100" (first file has two line ranges, second file has one line range)
+           - Note: The number of files in line_ranges must match the number of files in paths, otherwise an error will be thrown
+
+        Return value:
+        - Returns str type, containing content of all requested files
+        - Each file content will be annotated with file path and line range information (if line ranges are specified)
+
+        Usage examples:
 
         read_files(paths="main.py,utils.py", line_ranges="1-100/200-300,50-100")
 
         read_files(paths="main.py,utils.py")
 
-        你可以使用 get_project_structure 函数获取项目结构后，然后再通过 get_project_map 函数获取某个文件的用途，符号列表，以及
-        文件大小（tokens数）,最后再通过 read_files 函数来读取文件内容，从而帮你做更好的决策。如果需要读取的文件过大，
+        You can use get_project_structure function to get project structure, then use get_project_map function to get a file's purpose, symbol list, and
+        file size (token count), and finally use read_files function to read file content, helping you make better decisions. If the file to be read is too large,
 
-        特别注意：使用 read_files 时，一次性读取文件数量不要超过1个,每次只读取200行。如果发现读取的内容不够，则继续读取下面200行。
+        Special note: When using read_files, do not read more than 1 file at a time, only read 200 lines each time. If you find the read content is insufficient, continue reading the next 200 lines.
 
         </usage>
         </command>
 
         <command>
         <name>find_files_by_name</name>
-        <description>根据文件名中的关键字搜索文件。</description>
+        <description>Search files based on keywords in filename.</description>
         <usage>
-         该命令接受一个参数 keyword，为要搜索的关键字字符串。
+         This command accepts one parameter 'keyword', which is the keyword string to search for.
 
-         使用例子：
+         Usage example:
 
          find_files_by_name(keyword="test")
 
-         注意：
-         - 搜索不区分大小写
-         - 返回所有匹配的文件路径，逗号分隔
+         Notes:
+         - Search is case-insensitive
+         - Returns all matching file paths, comma-separated
         </usage>
         </command>
 
         <command>
         <name>find_files_by_content</name>
-        <description>根据文件内容中的关键字搜索文件。</description>
+        <description>Search files based on keywords in file content.</description>
         <usage>
-         该命令接受一个参数 keyword，为要搜索的关键字字符串。
+         This command accepts one parameter 'keyword', which is the keyword string to search for.
 
-         使用例子：
+         Usage example:
 
          find_files_by_content(keyword="TODO")
 
-         注意：
-         - 搜索不区分大小写
-         - 如果结果过多，只返回前10个匹配项
+         Notes:
+         - Search is case-insensitive
+         - If too many results, only returns the first 10 matches
         </usage>
         </command>
 
         <command>
         <name>read_file_with_keyword_ranges</name>
-        <description>读取包含指定关键字的行及其前后指定范围的行。</description>
+        <description>Read lines containing specified keywords and their surrounding lines within specified range.</description>
         <usage>
-         该函数用于读取包含关键字的行及其前后指定范围的行。
+         This function is used to read lines containing keywords and their surrounding lines within specified range.
 
-         参数说明:
-         1. file_path (str): 文件路径，可以是相对路径或绝对路径
-         2. keyword (str): 要搜索的关键字
-         3. before_size (int): 关键字行之前要读取的行数，默认100
-         4. after_size (int): 关键字行之后要读取的行数，默认100
+         Parameter description:
+         1. file_path (str): File path, can be relative or absolute path
+         2. keyword (str): Keyword to search for
+         3. before_size (int): Number of lines to read before keyword line, default 100
+         4. after_size (int): Number of lines to read after keyword line, default 100
 
-         返回值:
-         - 返回str类型，包含关键字的行及其前后指定范围的行
-         - 格式如下：
+         Return value:
+         - Returns str type, containing keyword lines and their surrounding lines within specified range
+         - Format as follows:
            ```
            ##File: /path/to/file.py
            ##Line: 10-20
 
-           内容
+           Content
            ```
 
-         使用例子：
+         Usage example:
          read_file_with_keyword_ranges(file_path="main.py", keyword="TODO", before_size=5, after_size=5)
 
-         注意：
-         - 如果文件中有多个匹配的关键字，会返回多个内容块
-         - 搜索不区分大小写
+         Notes:
+         - If there are multiple matching keywords in the file, multiple content blocks will be returned
+         - Search is case-insensitive
         </usage>
         </command>
 
         <command>
         <name>output_result</name>
-        <description>输出最后需要的结果</description>
+        <description>Output the final required result</description>
         <usage>
-         只有一个参数：
-         response: 字符串类型，需要返回给用户的内容。 response 必须满足如下Json格式：
+         Only one parameter:
+         response: String type, content to return to user. response must satisfy the following Json format:
 
          ```json
         {
@@ -543,70 +496,70 @@ class AgenticFilter:
                 {"path": "/path/to/new_file.txt", "operation": "ADD"},
                 {"path": "/path/to/old_file.log", "operation": "REMOVE"}
             ],
-            "reasoning": "详细说明你是如何通过分析和使用工具得出这个文件列表的。"
+            "reasoning": "Detailed explanation of how you analyzed and used tools to arrive at this file list."
         }
         ```
 
-        使用例子：
-        output_result(response='{"files": [{"path": "/path/to/file1.py", "operation": "MODIFY"}, {"path": "/path/to/file2.md", "operation": "REFERENCE"}, {"path": "/path/to/new_file.txt", "operation": "ADD"}, {"path": "/path/to/old_file.log", "operation": "REMOVE"}], "reasoning": "详细说明你是如何通过分析和使用工具得出这个文件列表的。"}')
+        Usage example:
+        output_result(response='{"files": [{"path": "/path/to/file1.py", "operation": "MODIFY"}, {"path": "/path/to/file2.md", "operation": "REFERENCE"}, {"path": "/path/to/new_file.txt", "operation": "ADD"}, {"path": "/path/to/old_file.log", "operation": "REMOVE"}], "reasoning": "Detailed explanation of how you analyzed and used tools to arrive at this file list."}')
         </usage>
         </command>
 
         <command>
         <name>count_file_tokens</name>
-        <description>计算指定文件的token数量。</description>
+        <description>Calculate the token count of specified file.</description>
         <usage>
-         该函数接受一个参数 file_path, 为要计算的文件路径。
+         This function accepts one parameter file_path, which is the file path to calculate.
 
-         使用例子：
+         Usage example:
          count_file_tokens(file_path="full")
 
-         注意：
-         - 返回值为int类型，表示文件的token数量。
+         Notes:
+         - Return value is int type, representing the token count of the file.
 
         </usage>
         </command>
 
         <command>
         <name>count_string_tokens</name>
-        <description>计算指定字符串的token数量。</description>
+        <description>Calculate the token count of specified string.</description>
         <usage>
-         该函数接受一个参数 text, 为要计算的文本。
+         This function accepts one parameter text, which is the text to calculate.
 
-         使用例子：
-         count_string_tokens(text="你好，世界")
+         Usage example:
+         count_string_tokens(text="Hello, World")
 
-         注意：
-         - 返回值为int类型，表示文本的token数量。
+         Notes:
+         - Return value is int type, representing the token count of the text.
 
         </usage>
         </command>
 
         <command>
         <n>find_symbol_definition</n>
-        <description>查找指定符号的定义所在的文件路径。</description>
+        <description>Find the file path where the specified symbol is defined.</description>
         <usage>
-         该函数接受一个参数 symbol, 为要查找的符号名称。
+         This function accepts one parameter symbol, which is the symbol name to search for.
 
-         使用例子：
+         Usage examples:
          find_symbol_definition(symbol="MyClass")
          find_symbol_definition(symbol="process_data")
 
-         注意：
-         - 返回值为字符串，包含符号定义所在的文件路径列表，以逗号分隔
-         - 支持精确匹配和模糊匹配（不区分大小写）
-         - 如果未找到匹配项，会返回提示信息
+         Notes:
+         - Return value is string, containing file path list where symbols are defined, comma-separated
+         - Supports exact matching and fuzzy matching (case-insensitive)
+         - If no matches found, returns prompt information
 
         </usage>
         </command>        
         <command>
         <n>execute_mcp_server</n>
-        <description>执行MCP服务器</description>
+        <description>Execute MCP server</description>
         <usage>
-         该函数接受一个参数 query, 为要执行的MCP服务器查询字符串。
+         This function accepts one parameter query, which is the MCP server query string to execute.
 
-         你可以根据下面已经连接的 mcp server 信息，来决定个是否调用该函数，注意该函数会更具你的 query
-         自动选择合适的 mcp server 来执行。如果你想某个特定的 server 来执行，你可以在 query 中说明你想哪个 server 执行。
+         You can decide whether to call this function based on the connected mcp server information below. Note that this function will automatically
+         select the appropriate mcp server to execute based on your query. If you want a specific server to execute, you can specify which server you want in the query.
 
          <mcp_server_info>
          {{ mcp_server_info }}
@@ -621,10 +574,10 @@ class AgenticFilter:
         }
 
     def analyze(self, request: AgenticFilterRequest) -> Optional[AgenticFilterResponse]:
-        # 获取 prompt 内容
+        # Get prompt content
         prompt = self._analyze.prompt(request)
 
-        # 获取对当前项目变更的最近8条历史人物
+        # Get the recent 8 history records of current project changes
         action_yml_file_manager = ActionYmlFileManager(self.args.source_dir)
         history_tasks = action_yml_file_manager.to_tasks_prompt(limit=8)
         new_messages = []
@@ -633,20 +586,20 @@ class AgenticFilter:
             new_messages.append(
                 {
                     "role": "assistant",
-                    "content": "好的，我知道最近的任务对项目的变更了，我会参考这些来更好的理解你的需求。",
+                    "content": "Okay, I understand the recent project changes from tasks, and I will refer to these to better understand your requirements.",
                 }
             )
 
-        # 构造对话上下文
+        # Construct conversation context
         conversations = new_messages + [{"role": "user", "content": prompt}]
 
-        # 使用 stream_out 进行输出
+        # Use stream_out for output
         printer = Printer()
         title = printer.get_message_from_key("auto_command_analyzing")
         final_title = printer.get_message_from_key("auto_command_analyzed")
 
         def extract_command_response(content: str) -> str:
-            # 提取 JSON 并转换为 AutoCommandResponse
+            # Extract JSON and convert to AutoCommandResponse
             try:
                 response = to_model(content, AutoCommandResponse)
                 if response.suggestions:
@@ -669,26 +622,29 @@ class AgenticFilter:
         success_flag = False
 
         get_event_manager(self.args.event_file).write_result(
-            EventContentCreator.create_result(content=printer.get_message_from_key("agenticFilterContext")),
+            EventContentCreator.create_result(
+                content=printer.get_message_from_key("agenticFilterContext")),
             metadata={
-                "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
             }
         )
 
         while True:
-            global_cancel.check_and_raise(token=self.args.event_file)
+            global_cancel.check_and_raise(token=self.args.cancel_token)
             # print(json.dumps(conversations, ensure_ascii=False, indent=4))
             model_name = ",".join(llms_utils.get_llm_names(self.llm))
             start_time = time.monotonic()
             result, last_meta = stream_out(
-                self.llm.stream_chat_oai(conversations=conversations, delta_mode=True),
+                self.llm.stream_chat_oai(
+                    conversations=conversations, delta_mode=True),
                 model_name=model_name,
                 title=title,
                 final_title=final_title,
                 display_func=extract_command_response,
                 args=self.args,
+                cancel_token=self.args.cancel_token,
                 extra_meta={
-                    "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                    "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
                 },
             )
 
@@ -700,9 +656,11 @@ class AgenticFilter:
                 from autocoder.utils import llms as llm_utils
 
                 model_info = (
-                    llm_utils.get_model_info(model_name, self.args.product_mode) or {}
+                    llm_utils.get_model_info(
+                        model_name, self.args.product_mode) or {}
                 )
-                input_price = model_info.get("input_price", 0.0) if model_info else 0.0
+                input_price = model_info.get(
+                    "input_price", 0.0) if model_info else 0.0
                 output_price = (
                     model_info.get("output_price", 0.0) if model_info else 0.0
                 )
@@ -743,17 +701,42 @@ class AgenticFilter:
                 )
 
             conversations.append({"role": "assistant", "content": result})
-            # 提取 JSON 并转换为 AutoCommandResponse
-            response = to_model(result, AutoCommandResponse)
+            # Extract JSON and convert to AutoCommandResponse
+            try:
+                response = to_model(result, AutoCommandResponse)
+            except Exception as e:
+                error_content = printer.get_message_from_key_with_format(
+                    "invalid_output_format_error",
+                    error=str(e)
+                )
+                # If this key doesn't exist, use default error message
+                if not error_content or error_content.startswith("Message key"):
+                                    error_content = f'''Invalid output format, error message: {str(e)}
+Please output in the required JSON format:
+```json
+{{{{
+    "files": [
+        {{"path": "/path/to/file1.py", "operation": "MODIFY"}},
+        {{"path": "/path/to/file2.md", "operation": "REFERENCE"}},
+        {{"path": "/path/to/new_file.txt", "operation": "ADD"}},
+        {{"path": "/path/to/old_file.log", "operation": "REMOVE"}}
+    ],
+                    "reasoning": "Detailed explanation of how you analyzed and used tools to arrive at this file list."
+}}}}
+```                    
+'''
+                conversations.append(
+                    {"role": "user", "content": error_content})
+                continue
 
             if not response or not response.suggestions:
                 break
 
-            # 执行命令
+            # Execute command
             command = response.suggestions[0].command
             parameters = response.suggestions[0].parameters
 
-            # 打印正在执行的命令
+                            # Print the command being executed
             temp_content = printer.get_message_from_key_with_format(
                 "auto_command_executing", command=command
             )
@@ -764,15 +747,16 @@ class AgenticFilter:
                     content=EventContentCreator.ResultCommandPrepareStatContent(
                         command=command, parameters=parameters
                     ).to_dict()
-                ),metadata={
-                    "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                ), metadata={
+                    "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
                 }
             )
             try:
                 self.execute_auto_command(command, parameters)
             except Exception as e:
-                error_content = f"执行命令失败，错误信息：{e}"
-                conversations.append({"role": "user", "content": error_content})
+                error_content = f"Command execution failed, error message: {e}"
+                conversations.append(
+                    {"role": "user", "content": error_content})
                 continue
 
             content = ""
@@ -780,30 +764,32 @@ class AgenticFilter:
             if last_result:
                 action = last_result.meta["action"]
                 if action == "coding":
-                    # 如果上一步是 coding，则需要把上一步的更改前和更改后的内容作为上下文
+                    # If the previous step was coding, need to include before and after changes as context
                     changes = git_utils.get_changes_by_commit_message(
                         "", last_result.meta["commit_message"]
                     )
                     if changes.success:
                         for file_path, change in changes.changes.items():
                             if change:
-                                content += f"## File: {file_path}[更改前]\n{change.before or 'New File'}\n\nFile: {file_path}\n\n[更改后]\n{change.after or 'Deleted File'}\n\n"
+                                content += f"## File: {file_path}[Before Changes]\n{change.before or 'New File'}\n\nFile: {file_path}\n\n[After Changes]\n{change.after or 'Deleted File'}\n\n"
                     else:
-                        content = printer.get_message_from_key("no_changes_made")
+                        content = printer.get_message_from_key(
+                            "no_changes_made")
                 else:
-                    # 其他的直接获取执行结果
+                    # For others, directly get execution results
                     content = last_result.content
 
                 if action != command:
-                    # command 和 action 不一致，则认为命令执行失败，退出
+                    # If command and action don't match, consider command execution failed and exit
                     temp_content = printer.get_message_from_key_with_format(
                         "auto_command_action_break", command=command, action=action
                     )
                     printer.print_str_in_terminal(temp_content, style="yellow")
                     get_event_manager(self.args.event_file).write_result(
-                        EventContentCreator.create_result(content=temp_content),
+                        EventContentCreator.create_result(
+                            content=temp_content),
                         metadata={
-                            "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                            "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
                         }
                     )
                     break
@@ -818,14 +804,14 @@ class AgenticFilter:
                             command=command, content=content
                         ).to_dict(),
                         metadata={
-                            "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                            "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
                         }
                     )
                 )
 
-                # 打印执行结果
+                # Print execution results
                 console = Console()
-                # 截取content前后200字符
+                # Truncate content to 200 characters before and after
                 truncated_content = (
                     content[:200] + "\n...\n" + content[-200:]
                     if len(content) > 400
@@ -834,7 +820,7 @@ class AgenticFilter:
                 title = printer.get_message_from_key_with_format(
                     "command_execution_result", action=action
                 )
-                # 转义内容，避免Rich将内容中的[]解释为markup语法
+                # Escape content to prevent Rich from interpreting [] in content as markup syntax
                 text_content = Text(truncated_content)
                 console.print(
                     Panel(
@@ -842,24 +828,25 @@ class AgenticFilter:
                     )
                 )
 
-                # 添加新的对话内容
+                # Add new conversation content
                 new_content = self._execute_command_result.prompt(content)
                 conversations.append({"role": "user", "content": new_content})
 
-                # 统计 token 数量
+                # Count token numbers
                 total_tokens = count_tokens(
                     json.dumps(conversations, ensure_ascii=False)
                 )
 
-                # 如果对话过长，使用默认策略进行修剪
-                if total_tokens > self.args.conversation_prune_safe_zone_tokens:
+                # If conversation is too long, use default strategy for pruning
+                parsed_safe_zone_tokens = self._get_parsed_safe_zone_tokens()
+                if total_tokens > parsed_safe_zone_tokens:
                     self.printer.print_in_terminal(
                         "conversation_pruning_start",
                         style="yellow",
                         total_tokens=total_tokens,
-                        safe_zone=self.args.conversation_prune_safe_zone_tokens,
+                        safe_zone=parsed_safe_zone_tokens,
                     )
-                    from autocoder.common.conversation_pruner import ConversationPruner
+                    from autocoder.common.pruner.conversation_pruner import ConversationPruner
 
                     pruner = ConversationPruner(self.args, self.llm)
                     conversations = pruner.prune_conversations(conversations)
@@ -872,27 +859,28 @@ class AgenticFilter:
                 get_event_manager(self.args.event_file).write_result(
                     EventContentCreator.create_result(content=temp_content),
                     metadata={
-                        "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                        "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
                     }
                 )
                 break
 
         get_event_manager(self.args.event_file).write_result(
-            EventContentCreator.create_result(content=printer.get_message_from_key("agenticFilterCommandResult")),
+            EventContentCreator.create_result(
+                content=printer.get_message_from_key("agenticFilterCommandResult")),
             metadata={
-                "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value                    
+                "stream_out_type": AgenticFilterStreamOutType.AGENTIC_FILTER.value
             }
-        )    
+        )
 
         if success_flag:
-            to_model(content, AgenticFilterResponse)
+            return to_model(content, AgenticFilterResponse)
             # return AgenticFilterResponse(**json.loads(content))
         else:
             return None
 
     def execute_auto_command(self, command: str, parameters: Dict[str, Any]) -> None:
         """
-        执行自动生成的命令
+        Execute automatically generated commands
         """
         command_map = {
             "run_python": self.tools.run_python_code,
@@ -912,6 +900,7 @@ class AgenticFilter:
             "count_file_tokens": self.tools.count_file_tokens,
             "count_string_tokens": self.tools.count_string_tokens,
             "find_symbol_definition": self.tools.find_symbol_definition,
+            "execute_shell_command": shells.execute_shell_command,
         }
 
         if command not in command_map:
@@ -922,7 +911,7 @@ class AgenticFilter:
             return
 
         try:
-            # 将参数字典转换为命令所需的格式
+            # Convert parameter dictionary to the format required by commands
             if parameters:
                 command_map[command](**parameters)
             else: