feat: 重写agent hub连接器，新增对话记录和微信输入状态

- 新增 Agent Hub 对话记录面板，按会话回看机器人与微信用户的完整收发内容 - 新增微信原生正在输入状态，长任务维持 typing，异常路径强制收尾
2026-10-07 14:20:05 +08:00 · 2026-09-17 17:32:37 +08:00
parent 1337bcb7de
commit 4b2395a8d2
73 changed files with 8855 additions and 2890 deletions
@@ -0,0 +1,175 @@
+import { describe, expect, it } from 'vitest'
+import {
+  buildAuthorizedHeaders,
+  buildBaseInfo,
+  buildCommonHeaders,
+  buildQrStatusHeaders,
+  generateWechatUin,
+  sanitizeBotAgent
+} from '../../src/main/services/wechat-ilink/headers'
+import {
+  extractMarkdownImageUrls,
+  markdownToPlainText
+} from '../../src/main/services/wechat-ilink/markdown'
+import {
+  extractInboundText,
+  normalizeInboundMessage
+} from '../../src/main/services/wechat-ilink/messages'
+
+describe('wechat-ilink headers', () => {
+  it('X-WECHAT-UIN 是随机 uint32 的 base64，并且每次请求都重新生成', () => {
+    const decoded = Buffer.from(generateWechatUin(), 'base64').toString('utf8')
+    expect(decoded).toMatch(/^\d+$/)
+    expect(Number(decoded)).toBeGreaterThanOrEqual(0)
+    expect(Number(decoded)).toBeLessThanOrEqual(0xffffffff)
+
+    const first = buildAuthorizedHeaders('token')['X-WECHAT-UIN']
+    const second = buildAuthorizedHeaders('token')['X-WECHAT-UIN']
+    expect(first).toBeTruthy()
+    // 官方要求每次请求重新生成以防重放：连续两次不应相同。
+    expect(second).not.toBe(first)
+  })
+
+  it('登录后的业务头包含鉴权字段与固定应用头', () => {
+    const headers = buildAuthorizedHeaders('bot-token')
+    expect(headers.AuthorizationType).toBe('ilink_bot_token')
+    expect(headers.Authorization).toBe('Bearer bot-token')
+    expect(headers['iLink-App-Id']).toBe('bot')
+    expect(headers['iLink-App-ClientVersion']).toBe('132102')
+  })
+
+  it('没有 token 时拒绝构造鉴权头', () => {
+    expect(() => buildAuthorizedHeaders('')).toThrow(/bot token/)
+  })
+
+  it('登录前后的公共头集合不同：未鉴权请求不携带 Authorization / X-WECHAT-UIN', () => {
+    const anonymous = { ...buildCommonHeaders(), ...buildQrStatusHeaders() }
+    expect(anonymous.Authorization).toBeUndefined()
+    expect(anonymous.AuthorizationType).toBeUndefined()
+    expect(anonymous['X-WECHAT-UIN']).toBeUndefined()
+    expect(anonymous['iLink-App-Id']).toBe('bot')
+  })
+
+  it('base_info 携带当前协议版本与合法 bot_agent', () => {
+    const baseInfo = buildBaseInfo()
+    expect(baseInfo.channel_version).toBe('2.4.6')
+    expect(baseInfo.bot_agent).toMatch(/^[\x21-\x7e]+$/)
+  })
+
+  it('bot_agent 清洗非法 token 并在为空时回退', () => {
+    expect(sanitizeBotAgent('TraceMemo/2.4.0')).toBe('TraceMemo/2.4.0')
+    expect(sanitizeBotAgent('TraceMemo/2.4.0 (mac)')).toBe('TraceMemo/2.4.0 (mac)')
+    expect(sanitizeBotAgent('中文非法/1.0')).toBe('OpenClaw')
+    expect(sanitizeBotAgent('')).toBe('OpenClaw')
+    expect(sanitizeBotAgent(undefined)).toBe('OpenClaw')
+    // 混合时丢弃非法 token，保留合法部分。
+    expect(sanitizeBotAgent('好坏的 Agent/1.0')).toBe('Agent/1.0')
+  })
+
+  it('bot_agent 总长度受 256 字节限制', () => {
+    const agent = Array.from({ length: 200 }, (_, index) => `Agent${index}/1.0`).join(' ')
+    expect(Buffer.byteLength(sanitizeBotAgent(agent), 'utf8')).toBeLessThanOrEqual(256)
+  })
+})
+
+describe('wechat-ilink markdown 降级', () => {
+  it('把常见 Markdown 降级成微信可读纯文本', () => {
+    const input = [
+      '# 群聊总结',
+      '',
+      '**重点**：`code` 与 ~~删除线~~',
+      '',
+      '- 第一项',
+      '- 第二项',
+      '',
+      '> 引用',
+      '',
+      '| A | B |',
+      '| --- | --- |',
+      '| 1 | 2 |',
+      '',
+      '链接：[官网](https://example.com)',
+      '图片：![图](https://example.com/a.png)',
+      '',
+      '```',
+      'const a = 1',
+      '```'
+    ].join('\n')
+
+    const output = markdownToPlainText(input)
+
+    expect(output).toContain('群聊总结')
+    expect(output).not.toContain('#')
+    expect(output).toContain('重点')
+    expect(output).not.toContain('**')
+    expect(output).toContain('code')
+    expect(output).not.toContain('`')
+    expect(output).toContain('删除线')
+    expect(output).not.toContain('~~')
+    expect(output).toContain('• 第一项')
+    expect(output).not.toContain('> 引用')
+    expect(output).toContain('A  B')
+    expect(output).not.toContain('|')
+    expect(output).toContain('官网')
+    expect(output).not.toContain('https://example.com/a.png')
+    expect(output).toContain('const a = 1')
+    expect(output).not.toContain('```')
+  })
+
+  it('折叠多余空行并去掉首尾空白', () => {
+    expect(markdownToPlainText('\n\n\n内容\n\n\n')).toBe('内容')
+  })
+
+  it('提取文本中内嵌的 http(s) 图片地址', () => {
+    const text = '看图 ![a](https://cdn.example.com/1.png) 和 ![b](/local/2.png)'
+    expect(extractMarkdownImageUrls(text)).toEqual(['https://cdn.example.com/1.png'])
+  })
+})
+
+describe('wechat-ilink 入站消息归一化', () => {
+  it('完整透传 context_token 与会话标识', () => {
+    const message = normalizeInboundMessage('bot-1', {
+      message_id: 12345,
+      seq: 3,
+      from_user_id: 'user@im.wechat',
+      message_type: 1,
+      item_list: [{ type: 1, text_item: { text: '你好' } }],
+      context_token: 'ctx-token',
+      session_id: 'session-1',
+      group_id: 'group-1'
+    })
+
+    expect(message).toMatchObject({
+      accountId: 'bot-1',
+      fromUserId: 'user@im.wechat',
+      messageId: '12345',
+      seq: 3,
+      sessionId: 'session-1',
+      groupId: 'group-1',
+      contextToken: 'ctx-token'
+    })
+    expect(message?.items).toEqual([{ type: 1, text: '你好' }])
+  })
+
+  it('缺少 from_user_id 的消息无法投递', () => {
+    expect(normalizeInboundMessage('bot-1', { message_id: 1 })).toBeUndefined()
+    expect(normalizeInboundMessage('bot-1', { from_user_id: '   ' })).toBeUndefined()
+  })
+
+  it('语音条目保留微信侧转写文本，但不算作文本条目', () => {
+    const message = normalizeInboundMessage('bot-1', {
+      from_user_id: 'user@im.wechat',
+      item_list: [{ type: 3, voice_item: { text: '语音转写内容' } }]
+    })
+
+    expect(message?.items).toEqual([{ type: 3, text: '语音转写内容' }])
+    // 与旧 webhook 行为一致：只有 type=1 参与文本提取。
+    expect(extractInboundText(message?.items)).toBe('')
+  })
+
+  it('多条文本条目按空格拼接', () => {
+    expect(
+      extractInboundText([{ type: 1, text: ' 第一段 ' }, { type: 2 }, { type: 1, text: '第二段' }])
+    ).toBe('第一段 第二段')
+  })
+})