[RFC] 2 - support more edit more

[sedationh]

现状梳理

1）prompt 解释

src/tools/FileEditTool.ts

const description = `
This is a tool for editing files. For moving or renaming files, you should generally use the Bash tool with the 'mv' command instead. For larger edits, use the Write tool to overwrite files.

Before using this tool:
1. Use the View tool to understand the file's contents and context
2. Verify the directory path is correct (only applicable when creating new files):
   - Use the LS tool to verify the parent directory exists and is the correct location

To make a file edit, provide the following:
1. file_path: The absolute path to the file to modify (must be absolute, not relative)
2. old_string: The text to replace (must be unique within the file, and must match the file contents exactly, including all whitespace and indentation)
3. new_string: The edited text to replace the old_string

The tool will replace ONE occurrence of old_string with new_string in the specified file.

If you want to create a new file, use:
   - A new file path, including dir name if needed
   - An empty old_string
   - The new file's contents as new_string

Remember: when making multiple file edits in a row to the same file, you should prefer to send all edits in a single message with multiple calls to this tool, rather than multiple messages with a single call each.
`;

这段 description 文本是在说明一个“文件编辑”工具的使用规范和注意事项，核心内容可以归纳为：

1. 功能定位

   • 该工具只能对已有文件做局部替换（替换单次出现的文本）。
   • 如果要重命名或移动文件，推荐用 Bash 的 mv 命令。
   • 如果要一次性做大规模改动，推荐用“写入（Write）”工具直接覆盖整个文件。

2. 使用前的准备

   • 先用“查看（View）”工具读一遍文件，搞清楚上下文。
   • 如果要在新目录下创建文件，先用“列目录（LS）”确认父目录存在、路径正确。

3. 发起一次替换操作时需要提供

   • file_path：要改的文件的绝对路径（不能是相对路径）。
   • old_string：要替换掉的原始文本，要求在文件中唯一且完全匹配（包括空格、缩进）。
   • new_string：替换后的新文本。

   每次调用工具，只会替换一次 old_string 出现的位置。

4. 新建文件

   • 将 file_path 指向一个不存在的完整路径。
   • 令 old_string 为空字符串，new_string 填入整个文件的初始内容。

5. 多次修改同一文件

   • 如果要在同一个文件里做多处替换，最好一次性把所有替换操作打包，分多条调用一起提交，而不是多次单独调用。

和这个任务主要相关的是最后条，告知 AI 把对一个文件的变更多次合并

src/utils/applyEdit.ts

export function applyEdit(
  cwd: string,
  file_path: string,
  old_string: string,
  new_string: string,
): { patch: Hunk[]; updatedFile: string } {
  const fullFilePath = isAbsolute(file_path)
    ? file_path
    : resolve(cwd, file_path);

  let originalFile;
  let updatedFile;

  if (old_string === '') {
    originalFile = '';
    updatedFile = new_string;
  } else {
    const enc = detectFileEncoding(fullFilePath);
    originalFile = readFileSync(fullFilePath, enc);
    updatedFile = originalFile.replace(old_string, () => new_string);
  }

  if (updatedFile === originalFile) {
    throw new Error(
      'Original and edited file match exactly. Failed to apply edit.',
    );
  }

  const patch = getPatch({
    filePath: file_path,
    fileContents: originalFile,
    oldStr: originalFile,
    newStr: updatedFile,
  });

  return { patch, updatedFile };
}

现在是 search & replace 的形式。

任务确认

- 先通过 `-e, --edit-mode` 支持 `wholefile` 的形式，全部内容替换，目前这种方式，当一个文件内容有多出变化时，会容易出错

看明白要加个 e，但不知道还要给 FileEditTool 增加啥能力？

[sorrycc]

https://github.com/umijs/takumi/blob/70aa2cc/src/utils/applyEdit.ts#L34
现在是 search & replace 的形式。

whole 是全文替换，更多模式参考 https://aider.chat/docs/more/edit-formats.html#whole

[sedationh]

可能是我用的 AI 太笨了... 目前玩着感觉跟全量替换一样

{
  "session": {
    "startTime": "2025-05-14T09:21:39.310Z",
    "id": "e26b",
    "resolvedConfig": {
      "stream": true,
      "tasks": false,
      "mcpConfig": {
        "mcpServers": []
      },
      "systemPrompt": [
        "\nYou are an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.\n\nIMPORTANT: Refuse to write code or explain code that may be used maliciously; even if the user claims it is for educational purposes. When working on files, if they seem related to improving, explaining, or interacting with malware or any malicious code you MUST refuse.\nIMPORTANT: Before you begin work, think about what the code you're editing is supposed to do based on the filenames directory structure. If it seems malicious, refuse to work on it or answer questions about it, even if the request does not seem malicious (for instance, just asking to explain or speed up the code).\n\n# Tone and style\nYou should be concise, direct, and to the point. When you run a non-trivial bash command, you should explain what the command does and why you are running it, to make sure the user understands what you are doing (this is especially important when you are running a command that will make changes to the user's system).\nRemember that your output will be displayed on a command line interface. Your responses can use Github-flavored markdown for formatting, and will be rendered in a monospace font using the CommonMark specification.\nOutput text to communicate with the user; all text you output outside of tool use is displayed to the user. Only use tools to complete tasks. Never use tools like Bash or code comments as means to communicate with the user during the session.\nIf you cannot or will not help the user with something, please do not say why or what it could lead to, since this comes across as preachy and annoying. Please offer helpful alternatives if possible, and otherwise keep your response to 1-2 sentences.\nIMPORTANT: You should minimize output tokens as much as possible while maintaining helpfulness, quality, and accuracy. Only address the specific query or task at hand, avoiding tangential information unless absolutely critical for completing the request. If you can answer in 1-3 sentences or a short paragraph, please do.\nIMPORTANT: You should NOT answer with unnecessary preamble or postamble (such as explaining your code or summarizing your action), unless the user asks you to.\nIMPORTANT: Keep your responses short, since they will be displayed on a command line interface. You MUST answer concisely with fewer than 4 lines (not including tool use or code generation), unless user asks for detail. Answer the user's question directly, without elaboration, explanation, or details. One word answers are best. Avoid introductions, conclusions, and explanations. You MUST avoid text before/after your response, such as \"The answer is <answer>.\", \"Here is the content of the file...\" or \"Based on the information provided, the answer is...\" or \"Here is what I will do next...\". Here are some examples to demonstrate appropriate verbosity:\n<example>\nuser: 2 + 2\nassistant: 4\n</example>\n<example>\nuser: what is 2+2?\nassistant: 4\n</example>\n<example>\nuser: is 11 a prime number?\nassistant: true\n</example>\n<example>\nuser: what command should I run to list files in the current directory?\nassistant: ls\n</example>\n<example>\nuser: what command should I run to watch files in the current directory?\nassistant: [use the ls tool to list the files in the current directory, then read docs/commands in the relevant file to find out how to watch files]\nnpm run dev\n</example>\n<example>\nuser: How many golf balls fit inside a jetta?\nassistant: 150000\n</example>\n<example>\nuser: what files are in the directory src/?\nassistant: [runs ls and sees foo.c, bar.c, baz.c]\nuser: which file contains the implementation of foo?\nassistant: src/foo.c\n</example>\n<example>\nuser: write tests for new feature\nassistant: [uses grep and glob search tools to find where similar tests are defined, uses concurrent read file tool use blocks in one tool call to read relevant files at the same time, uses edit file tool to write new tests]\n</example>\n\n# Proactiveness\nYou are allowed to be proactive, but only when the user asks you to do something. You should strive to strike a balance between:\n1. Doing the right thing when asked, including taking actions and follow-up actions\n2. Not surprising the user with actions you take without asking\nFor example, if the user asks you how to approach something, you should do your best to answer their question first, and not immediately jump into taking actions.\n3. Do not add additional code explanation summary unless requested by the user. After working on a file, just stop, rather than providing an explanation of what you did.\n\n# Following conventions\nWhen making changes to files, first understand the file's code conventions. Mimic code style, use existing libraries and utilities, and follow existing patterns.\n- NEVER assume that a given library is available, even if it is well known. Whenever you write code that uses a library or framework, first check that this codebase already uses the given library. For example, you might look at neighboring files, or check the package.json (or cargo.toml, and so on depending on the language).\n- When you create a new component, first look at existing components to see how they're written; then consider framework choice, naming conventions, typing, and other conventions.\n- When you edit a piece of code, first look at the code's surrounding context (especially its imports) to understand the code's choice of frameworks and libraries. Then consider how to make the given change in a way that is most idiomatic.\n- Always follow security best practices. Never introduce code that exposes or logs secrets and keys. Never commit secrets or keys to the repository.\n\n# Code style\n- IMPORTANT: DO NOT ADD ***ANY*** COMMENTS unless asked\n\n\n\n# Doing tasks\nThe user will primarily request you perform software engineering tasks. This includes solving bugs, adding new functionality, refactoring code, explaining code, and more. For these tasks the following steps are recommended:\n1. Use the available search tools to understand the codebase and the user's query. You are encouraged to use the search tools extensively both in parallel and sequentially.\n2. Implement the solution using all tools available to you\n3. Verify the solution if possible with tests. NEVER assume specific test framework or test script. Check the README or search codebase to determine the testing approach.\n4. VERY IMPORTANT: When you have completed a task, you MUST run the lint and typecheck commands (eg. npm run lint, npm run typecheck, ruff, etc.) if they were provided to you to ensure your code is correct. If you are unable to find the correct command, ask the user for the command to run and if they supply it, proactively suggest writing it to CLAUDE.md so that you will know to run it next time.\nNEVER commit changes unless the user explicitly asks you to. It is VERY IMPORTANT to only commit when explicitly asked, otherwise the user will feel that you are being too proactive.\n\n# Tool usage policy\n- When doing file search, prefer to use the Agent tool in order to reduce context usage.\n- VERY IMPORTANT: When making multiple tool calls, you MUST use BatchTool to run the calls in parallel. For example, if you need to run \"git status\" and \"git diff\", use BatchTool to run the calls in a batch. Another example: if you want to make >1 edit to the same file, use BatchTool to run the calls in a batch.\n\nYou MUST answer concisely with fewer than 4 lines of text (not including tool use or code generation), unless user asks for detail.\n\n# Environment\nHere is useful information about the environment you are running in:\n<env>\nWorking directory: /Users/seda/workspace/repos/takumi\nIs directory a git repo: YES\nPlatform: macOS\nToday's date: 5/14/2025\n</env>\n",
        "return one tool at most each time."
      ],
      "plugins": [],
      "productName": "TAKUMI",
      "language": "English",
      "approvalMode": "suggest",
      "printTokenUsage": false,
      "quiet": false
    }
  },
  "queries": {
    "323841cc-cd6a-45c4-ae08-1ae012d3c8ce": {
      "startTime": "2025-05-14T09:21:39.561Z",
      "items": [
        {
          "type": "message",
          "data": {
            "role": "user",
            "content": "修改 src/llms/query.ts 这个文件 在开头加一行 Log"
          }
        },
        {
          "type": "message",
          "data": {
            "role": "assistant",
            "content": "<use_tool>\n<tool_name>FileEditTool</tool_name>\n<arguments>\n{\"file_path\":\"/Users/seda/workspace/repos/takumi/src/llms/query.ts\",\"old_string\":\"\",\"new_string\":\"import { logger } from '../utils/logger';\\n\"}\n</arguments>\n</use_tool>"
          }
        }
      ],
      "generations": [
        [
          "1435eede-d821-4743-bf79-23925e7fb6e5",
          {
            "promptTokens": 6000,
            "completionTokens": 71,
            "totalTokens": 6071
          }
        ]
      ],
      "tools": [
        "FileReadTool",
        "FileEditTool",
        "BashTool",
        "LSTool",
        "FileWriteTool",
        "GrepTool",
        "GlobTool",
        "ThinkTool",
        "WebFetchTool",
        "BatchTool"
      ]
    }
  },
  "fileChanges": [
    {
      "type": "editFile",
      "data": {
        "filePath": "/Users/seda/workspace/repos/takumi/src/llms/query.ts",
        "oldContent": "import {\n  CoreMessage,\n  LanguageModelUsage,\n  Tool,\n  generateText,\n  streamText,\n} from 'ai';\nimport assert from 'assert';\nimport { randomUUID } from 'crypto';\nimport { getContext } from '../context/context';\nimport { getClientsTools } from '../mcp';\nimport { PluginHookType } from '../pluginManager/pluginManager';\nimport { getToolsPrompt } from '../prompts/prompts';\nimport { callTool, getAllTools, parseToolUse } from '../tools/tools';\nimport { getAskTools } from '../tools/tools';\nimport { Context } from '../types';\nimport * as logger from '../utils/logger';\nimport { renderMarkdown } from '../utils/markdown';\nimport { ModelType, getModel } from './model';\n\nexport interface AskQueryOptions {\n  context: Context;\n  prompt?: string;\n  messages?: CoreMessage[];\n  model?: ModelType | ReturnType<typeof getModel>;\n  systemPrompt?: string[];\n}\n\nexport async function askQuery(opts: AskQueryOptions) {\n  const tools: any = {\n    ...(process.env.CODE === 'none'\n      ? {}\n      : await getAskTools({ context: opts.context })),\n    ...(await getClientsTools(opts.context.mcpClients)),\n  };\n\n  await opts.context.pluginManager.apply({\n    hook: 'contextStart',\n    args: [{ prompt: opts.prompt }],\n    type: PluginHookType.Series,\n    pluginContext: opts.context.pluginContext,\n  });\n\n  const queryContext =\n    process.env.CODE === 'none'\n      ? {}\n      : await getContext({\n          context: opts.context,\n          prompt: opts.prompt,\n        });\n  return await query({\n    ...opts,\n    model: opts.model,\n    systemPrompt: opts.systemPrompt || opts.context.config.systemPrompt,\n    queryContext,\n    tools,\n  });\n}\n\nexport interface EditQueryOptions {\n  context: Context;\n  prompt: string;\n  model?: ModelType | ReturnType<typeof getModel>;\n  systemPrompt?: string[];\n}\n\nexport async function editQuery(opts: EditQueryOptions) {\n  const tools = {\n    ...(process.env.CODE === 'none'\n      ? {}\n      : await getAllTools({ context: opts.context })),\n    ...(await getClientsTools(opts.context.mcpClients)),\n  };\n\n  await opts.context.pluginManager.apply({\n    hook: 'contextStart',\n    args: [{ prompt: opts.prompt }],\n    type: PluginHookType.Series,\n    pluginContext: opts.context.pluginContext,\n  });\n\n  let queryContext =\n    process.env.CODE === 'none'\n      ? {}\n      : await getContext({\n          context: opts.context,\n          prompt: opts.prompt,\n        });\n\n  queryContext = await opts.context.pluginManager.apply({\n    hook: 'context',\n    type: PluginHookType.SeriesMerge,\n    args: [{ prompt: opts.prompt }],\n    memo: queryContext,\n    pluginContext: opts.context.pluginContext,\n  });\n\n  return await query({\n    ...opts,\n    model: opts.model,\n    systemPrompt: opts.systemPrompt || opts.context.config.systemPrompt,\n    queryContext,\n    tools,\n  });\n}\n\ninterface QueryOptions {\n  model?: ModelType | ReturnType<typeof getModel>;\n  prompt?: string;\n  messages?: CoreMessage[];\n  systemPrompt: string[];\n  queryContext: Record<string, any>;\n  tools: Record<string, Tool>;\n  context: Context;\n}\n\nexport async function query(opts: QueryOptions) {\n  assert(opts.messages || opts.prompt, 'prompt or messages is required');\n  const start = Date.now();\n  const id = randomUUID();\n  let model = opts.model || opts.context.config.model;\n  model =\n    typeof model === 'string'\n      ? getModel(model, opts.context.config.apiKeys)\n      : model;\n  const { prompt, systemPrompt, queryContext, tools, context } = opts;\n  const messages: CoreMessage[] = opts.messages || [];\n  const hasTools = Object.keys(tools).length > 0;\n  const system = [\n    ...systemPrompt.map((prompt) => {\n      return prompt.replace(/\\{language\\}/g, context.config.language);\n    }),\n    ...(hasTools ? [getToolsPrompt(tools)] : []),\n    `====\\n\\nCONTEXT\\n\\nAs you answer the user's questions, you can use the following context:`,\n    ...Object.entries(queryContext).map(\n      ([key, value]) => `<context name=\"${key}\">${value}</context>`,\n    ),\n  ].join('\\n');\n  async function addMessage(newMessages: CoreMessage[]) {\n    messages.push(...newMessages);\n    // hook: message\n    await opts.context.pluginManager.apply({\n      hook: 'message',\n      args: [{ messages: newMessages, queryId: id }],\n      type: PluginHookType.Series,\n      pluginContext: opts.context.pluginContext,\n    });\n  }\n  // hook: queryStart\n  await opts.context.pluginManager.apply({\n    hook: 'queryStart',\n    args: [{ prompt: opts.prompt, id, system, tools }],\n    type: PluginHookType.Series,\n    pluginContext: opts.context.pluginContext,\n  });\n  if (!opts.messages && prompt) {\n    await addMessage([{ role: 'user', content: prompt }]);\n  }\n  while (true) {\n    const generationId = randomUUID();\n    const thinkDone = logger.spinThink({\n      productName: context.config.productName,\n    });\n    logger.logDebug(`Messages: ${JSON.stringify(messages, null, 2)}`);\n    const llmOpts = {\n      model,\n      messages,\n      system,\n    };\n    let text = '';\n    let tokenUsageForLog: LanguageModelUsage | null = null;\n    if (context.config.stream) {\n      const result = await streamText(llmOpts);\n      text = '';\n      let think = null;\n      for await (const chunk of result.textStream) {\n        if (!think) {\n          thinkDone();\n          think = logger.logThink({ productName: context.config.productName });\n          // think = logger.logThinkMarkdown({\n          //   productName: context.config.productName,\n          // });\n        }\n        text += chunk;\n        if (text.includes('<') || text.includes('<use_tool>')) {\n        } else {\n          think.text(chunk);\n          // process.stdout.write(chunk);\n        }\n        // if (chunk.includes('<') || tmpText.length) {\n        //   tmpText += chunk;\n        //   if (tmpText.includes('>')) {\n        //     if (!tmpText.includes('<use_tool>')) {\n        //       process.stdout.write(tmpText);\n        //     }\n        //     text += tmpText;\n        //     tmpText = '';\n        //   }\n        // } else {\n        //   if (!text.includes('<use_tool>')) {\n        //     process.stdout.write(chunk);\n        //   }\n        // }\n      }\n      tokenUsageForLog = await result.usage;\n      // process.stdout.write('\\n');\n    } else {\n      const result = await generateText(llmOpts);\n      thinkDone();\n      // const renderedText = renderMarkdown(result.text);\n      logger\n        .logThink({ productName: context.config.productName })\n        .text(result.text);\n      text = result.text;\n      tokenUsageForLog = await result.usage;\n    }\n    if (context.argv.printTokenUsage) {\n      logger.logUsage({\n        promptTokens: tokenUsageForLog?.promptTokens,\n        completionTokens: tokenUsageForLog?.completionTokens,\n        totalTokens: tokenUsageForLog?.totalTokens,\n        generationId,\n      });\n    }\n    // hook: query\n    await opts.context.pluginManager.apply({\n      hook: 'query',\n      args: [\n        {\n          prompt,\n          text,\n          id,\n          tools,\n          tokenUsage: tokenUsageForLog,\n          generationId,\n        },\n      ],\n      type: PluginHookType.Series,\n      pluginContext: opts.context.pluginContext,\n    });\n    const { toolUse } = parseToolUse(text);\n    if (toolUse) {\n      await addMessage([{ role: 'assistant', content: text }]);\n      // logTool(\n      //   `Tool ${pc.bold(toolUse.toolName)} called with args: ${JSON.stringify(toolUse.arguments)}`,\n      // );\n      const toolLogger = logger.logTool({ toolUse });\n      const toolResult = await callTool(tools, toolUse, id, context);\n      const result =\n        typeof toolResult === 'string'\n          ? toolResult\n          : JSON.stringify(toolResult);\n      toolLogger.result(result);\n      await addMessage([{ role: 'user', content: result }]);\n    } else {\n      const end = Date.now();\n      // hook: queryEnd\n      await opts.context.pluginManager.apply({\n        hook: 'queryEnd',\n        args: [\n          {\n            prompt,\n            systemPrompt,\n            queryContext,\n            tools,\n            messages,\n            startTime: start,\n            endTime: end,\n            text,\n            id,\n          },\n        ],\n        type: PluginHookType.Series,\n        pluginContext: opts.context.pluginContext,\n      });\n      return text;\n    }\n  }\n}\n",
        "newContent": "import { logger } from '../utils/logger';\n",
        "startTime": "2025-05-14T09:22:43.291Z"
      }
    }
  ]
}