yaoyao/src/shared/grokBotImport.ts

/** Grok Bot 模板导入：解析 x.ai/bot/<id> 公开页并翻译为夭夭 Bot 的人设草稿。 */
export interface GrokBotMeta {
  name: string
  description: string
  imageUrl?: string
}

export type GrokBotTemplateKind = 'housekeeper' | 'manager' | 'general'

export interface GrokBotDraft {
  kind: GrokBotTemplateKind
  job?: string
  antiJobs?: string[]
  actBias?: 'ask_first' | 'ask_key_then_act' | 'act_now'
  canManageTeam?: boolean
  instructions: string
  suggestions: string[]
}

export const GROK_BOT_URL_PATTERN = /^https:\/\/x\.ai\/bot\/([A-Za-z0-9_-]+)/

function decodeEntities(text: string): string {
  return text
    .replace(/&amp;/g, '&')
    .replace(/&lt;/g, '<')
    .replace(/&gt;/g, '>')
    .replace(/&quot;/g, '"')
    .replace(/&#39;|&apos;/g, "'")
    .replace(/&#x27;/g, "'")
    .trim()
}

function metaContent(html: string, keys: string[]): string | undefined {
  for (const key of keys) {
    // Attributes on real pages are double-quoted; only exclude the quote
    // actually in use so apostrophes inside descriptions survive.
    const pattern = new RegExp(
      `<meta[^>]+(?:property|name)="${key}"[^>]+content="([^"]*)"|<meta[^>]+content="([^"]*)"[^>]+(?:property|name)="${key}"`,
      'i',
    )
    const match = pattern.exec(html)
    const value = match?.[1] ?? match?.[2]
    if (value) return decodeEntities(value)
  }
  return undefined
}

/** Extract the public metadata from a fetched x.ai bot page. */
export function parseGrokBotPage(html: string): GrokBotMeta | null {
  const title = metaContent(html, ['og:title', 'twitter:title']) ?? (() => {
    const match = /<title[^>]*>([^<]*)<\/title>/i.exec(html)
    return match ? decodeEntities(match[1]!.replace(/\s*[-|]\s*Grok.*$/i, '')) : undefined
  })()
  const description = metaContent(html, ['og:description', 'twitter:description', 'description'])
  if (!title) return null
  return { name: title.slice(0, 100), description: (description ?? '').slice(0, 1200), imageUrl: metaContent(html, ['og:image', 'twitter:image']) }
}

export class GrokBotImportError extends Error {
  code: string
  status: number
  constructor(message: string, code: string, status = 400) {
    super(message)
    this.code = code
    this.status = status
  }
}

/** Optional outbound proxy for x.ai only: HTTPS_PROXY / GROK_BOT_PROXY. */
async function proxyDispatcher(target: string): Promise<unknown> {
  const proxy = process.env.GROK_BOT_PROXY ?? process.env.HTTPS_PROXY ?? process.env.https_proxy
  if (!proxy || !/^https:\/\//.test(target)) return undefined
  try {
    const undici = await import('undici')
    return new undici.ProxyAgent(proxy)
  } catch {
    return undefined
  }
}

export async function fetchGrokBotPage(url: string, fetchImpl: typeof fetch = fetch, timeoutMs = 12_000): Promise<string> {
  if (!GROK_BOT_URL_PATTERN.test(url.trim())) throw new GrokBotImportError('请输入 x.ai/bot/<id> 形式的 Grok Bot 链接', 'invalid_grok_bot_url')
  let response: Response
  try {
    const dispatcher = await proxyDispatcher(url)
    response = await fetchImpl(url.trim(), { headers: { 'user-agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15' , accept: 'text/html' }, signal: AbortSignal.timeout(timeoutMs), redirect: 'follow', ...(dispatcher ? { dispatcher } : {}) } as RequestInit)
  } catch {
    throw new GrokBotImportError('无法访问 x.ai：当前网络可能需要代理（可配置 GROK_BOT_PROXY/HTTPS_PROXY），或改用“粘贴描述导入”', 'grok_bot_fetch_failed', 502)
  }
  if (!response.ok) throw new GrokBotImportError(`x.ai 返回 ${response.status}，请确认链接可公开访问`, 'grok_bot_fetch_failed', 502)
  const html = await response.text()
  if (!parseGrokBotPage(html)) throw new GrokBotImportError('页面里找不到 Bot 信息，请确认这是 Grok Bot 分享链接', 'grok_bot_page_invalid', 502)
  return html
}

export async function fetchAvatarDataUrl(imageUrl: string | undefined, fetchImpl: typeof fetch = fetch, maxBytes = 2_800_000): Promise<string | undefined> {
  if (!imageUrl || !/^https:\/\//.test(imageUrl)) return undefined
  try {
    const dispatcher = await proxyDispatcher(imageUrl)
    const response = await fetchImpl(imageUrl, { signal: AbortSignal.timeout(10_000), redirect: 'follow', ...(dispatcher ? { dispatcher } : {}) } as RequestInit)
    if (!response.ok) return undefined
    const type = (response.headers.get('content-type') ?? '').toLowerCase()
    if (!/^image\/(png|jpeg|webp)/.test(type)) return undefined
    const buffer = Buffer.from(await response.arrayBuffer())
    if (buffer.byteLength > maxBytes) return undefined
    return `data:${type.split(';')[0]};base64,${buffer.toString('base64')}`
  } catch {
    return undefined
  }
}

const HOUSEKEEPER_PATTERN = /email|mail|calendar|邮件|日历|日程|invoice|账单|form|表单|rsvp|邀请函/i
const MANAGER_PATTERN = /notion|project|项目|团队|team|task|排期|进度|delegate|委派|分派/i

/** Translate a Grok Bot description into a yaoyao persona draft. */
export function grokBotDraft(meta: GrokBotMeta): GrokBotDraft {
  const summary = `${meta.description}`
  if (MANAGER_PATTERN.test(summary)) {
    return {
      kind: 'manager',
      job: '管理团队项目：每个项目一份任务计划和一个群聊，任务交给合适的 Bot 并跟进到完成',
      antiJobs: ['绝不替专家 Bot 完成具体工作', '不隐瞒风险和延误，坏消息第一时间上报'],
      actBias: 'ask_key_then_act',
      canManageTeam: true,
      instructions: [
        `你从 Grok Bot 模板「${meta.name}」导入，原始描述：${summary}`,
        '',
        '工作方式（夭夭版）：',
        '1. 每个新项目：创建一份任务计划（写清目标、验收标准、截止时间），并建一个以项目命名的群聊，把相关 Bot 拉进来。',
        '2. 任务拆解后写给最合适的成员 Bot，用 @成员名 明确委派；你是管理员，负责分派、跟进、汇总，不亲自做专家工作。',
        '3. 每天在群里汇总一次进度：完成、进行中、阻塞，以及需要用户决策的事项。',
        '4. 用户是最终决策者：涉及范围、预算、优先级的变化，先问再做。',
      ].join('\n'),
      suggestions: ['为它准备几个专家 Bot 作为队友（可再从 Grok Bot 模板导入或自建）'],
    }
  }
  if (HOUSEKEEPER_PATTERN.test(summary)) {
    return {
      kind: 'housekeeper',
      job: '盯家庭邮箱和日程，学校的表单、账单和回复邀请不遗漏',
      antiJobs: ['绝不直接发送邮件或回复邀请', '绝不替用户提交任何表单', '不擅自改期或取消日程'],
      actBias: 'ask_key_then_act',
      instructions: [
        `你从 Grok Bot 模板「${meta.name}」导入，原始描述：${summary}`,
        '',
        '工作方式（夭夭版）：',
        '1. 巡检时用 desktop_shell 在服务器 Mac 上执行 osascript 读取「邮件」App 的未读邮件和「日历」App 今明两天的日程。',
        '2. 发现需要处理的事项（表单截止、账单、接送冲突、RSVP）时，在对话里列出清单并给出建议动作。',
        '3. 需要回复的邮件：在对话里起草完整回复文本，等用户确认后仍由用户发送。',
        '4. 保持每条汇报简短：事项、截止时间、建议，不写长篇解释。',
      ].join('\n'),
      suggestions: ['在「电脑与定时任务」勾选本机控制并完成文件与命令授权（邮件/日历靠它读取）', '创建每天早晚各一次的定时任务让它自动巡检'],
    }
  }
  return {
    kind: 'general',
    instructions: `你从 Grok Bot 模板「${meta.name}」导入，原始描述：${summary}\n\n（此 Bot 从 Grok Bot 模板导入。上方为原始描述，可在「角色与规则」中补充一句话职责、反任务清单与行动偏好，让它更贴合你的使用。）`,
    suggestions: ['打开「角色与规则」补一句话职责和反任务清单，结构化规范会自动生效'],
  }
}