@xmanrui/dsh-im 4.28.1 → 4.29.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xmanrui/dsh-im",
-  "version": "4.28.1",
+  "version": "4.29.0",
   "description": "把十二种 IM 渠道和公网 AI Office 接入本机 DeepSeek Harness。 Connect twelve IM channels and a public AI Office to a local DeepSeek Harness.",
   "keywords": [
     "deepseek-harness",
@@ -85,6 +85,10 @@
     {
       "name": "Zhe (Phil) Yang",
       "url": "https://github.com/yangzhe1991"
+    },
+    {
+      "name": "Biertolovisk",
+      "url": "https://github.com/rosalynthompson32-a11y"
     }
   ],
   "license": "MIT",

package/plugin-src/client/channels/feishu/api.js

@@ -14,6 +14,7 @@ import { normalizeLastMessageError } from "../../last-message-error.js";
 import { normalizeAccessPolicy } from "../../../../src/channels/shared/access-policy.mjs";
 import { normalizeContextEnhancementConfig } from "../../../../src/channels/shared/context-enhancement.mjs";
 import { normalizeFeishuStepPushMode } from "../../../../src/channels/feishu/step-push-mode.mjs";
+import { normalizeFeishuVoiceConfig } from "../../../../src/channels/feishu/voice-config.mjs";
 
 export const FEISHU_RPC_CHANNEL = "/feishu";
 
@@ -38,6 +39,7 @@ export const FEISHU_ENDPOINTS = Object.freeze({
   setGroupTopicReply: "bot.group-topic-reply.set",
   setStepPush: "bot.step-push.set",
   setStepPushMode: "bot.step-push-mode.set",
+  setVoice: "bot.voice.set",
   // Kept for rolling upgrades. The multi-bot UI never calls these endpoints.
   testConnection: "connection.test",
   disconnect: "connection.disconnect",
@@ -227,6 +229,7 @@ export function normalizeBotConnection(value, fallbackBotId) {
     groupTopicReply: value.groupTopicReply === true,
     stepPush: value.stepPush === true,
     stepPushMode: normalizeFeishuStepPushMode(value.stepPushMode),
+    voice: normalizeFeishuVoiceConfig(value.voice),
     groupMessagePermissionGranted: value.groupMessagePermissionGranted === true,
     bot: normalizeBot(value.bot),
     health: normalizeHealth(value.health, connected),

package/plugin-src/client/channels/feishu/voice-editor.js

@@ -0,0 +1,115 @@
+import * as React from 'react';
+
+import { h } from '../../i18n.js';
+
+// 语音交互编辑器:开关 + 凭据引用 + 模型/音色参数。
+// 凭据值由插件凭据库管理,这里只保存 secretRef 名称——密钥本身不进入配置,
+// 也不经本表单传输。开启后:收到语音消息时转写为文字走正常流水线,并用
+// 语音回复同回合答案(需 DashScope qwen3 系列密钥与可用 ffmpeg)。
+export function VoiceEditor({ value = null, disabled = false, onSave }) {
+  const [enabled, setEnabled] = React.useState(value != null);
+  const [secretRef, setSecretRef] = React.useState(value?.secretRef ?? 'DASHSCOPE_API_KEY');
+  const [asrModel, setAsrModel] = React.useState(value?.asrModel ?? 'qwen3-asr-flash');
+  const [ttsModel, setTtsModel] = React.useState(value?.ttsModel ?? 'qwen3-tts-flash');
+  const [ttsVoice, setTtsVoice] = React.useState(value?.ttsVoice ?? 'Momo');
+  const [saving, setSaving] = React.useState(false);
+  const [error, setError] = React.useState(null);
+  const titleId = React.useId();
+  const helpId = `${titleId}-help`;
+
+  React.useEffect(() => {
+    setEnabled(value != null);
+    if (value) {
+      setSecretRef(value.secretRef ?? 'DASHSCOPE_API_KEY');
+      setAsrModel(value.asrModel ?? 'qwen3-asr-flash');
+      setTtsModel(value.ttsModel ?? 'qwen3-tts-flash');
+      setTtsVoice(value.ttsVoice ?? 'Momo');
+    }
+  }, [value]);
+
+  const persist = async (payload) => {
+    if (saving || disabled) return;
+    setSaving(true);
+    setError(null);
+    try {
+      await onSave?.(payload);
+    } catch (cause) {
+      setError(cause?.message ?? '语音设置保存失败，请重试。');
+    } finally {
+      setSaving(false);
+    }
+  };
+
+  const save = () => persist(enabled ? { enabled: true, secretRef, asrModel, ttsModel, ttsVoice } : null);
+
+  // 切到“关闭”立即持久化 null:关闭态没有保存按钮,若不在这里落盘,
+  // 后台会一直保持原配置,用户以为语音已停而实际仍在处理语音。
+  const toggle = (next) => {
+    setEnabled(next);
+    if (!next) void persist(null);
+  };
+
+  const input = (label, state, setState, placeholder) => h('label', { className: 'dim-feishuVoiceField' },
+    h('span', { className: 'dim-feishuVoiceLabel' }, label),
+    h('input', {
+      type: 'text',
+      className: 'dim-feishuVoiceInput',
+      value: state,
+      placeholder,
+      disabled: disabled || saving,
+      onChange: (event) => setState(event.target.value),
+    }),
+  );
+
+  return h('section', {
+    className: 'dim-feishuGroupControl',
+    "aria-labelledby": titleId,
+  },
+  h('div', { className: 'dim-feishuGroupControlHeader' },
+    h('div', { className: 'dim-presetTitle' },
+      h('h3', { id: titleId }, '语音交互'),
+      h('span', { className: 'dim-presetHelp' },
+        h('button', {
+          type: 'button',
+          className: 'dim-presetHelpButton',
+          "aria-label": '查看语音交互说明',
+          "aria-describedby": helpId,
+        }, h('span', { "aria-hidden": 'true' }, '?')),
+        h('span', {
+          id: helpId,
+          className: 'dim-presetTooltip',
+          role: 'tooltip',
+        }, '收到语音消息时转写为文字并按文字流水线处理,答案同时以语音回复;转写失败时降级为仅支持文字/图片/文件的提示。凭据值在插件凭据库里管理,这里只引用环境变量名。'))),
+    saving
+      ? h('span', { className: 'dim-feishuGroupControlStatus', role: 'status' }, '保存中…')
+      : null),
+  h('select', {
+    className: 'dim-feishuGroupSelect',
+    value: enabled ? 'on' : 'off',
+    disabled: disabled || saving,
+    "aria-label": '语音交互开关',
+    onChange: (event) => toggle(event.target.value === 'on'),
+  },
+  h('option', { value: 'off' }, '关闭'),
+  h('option', { value: 'on' }, '开启(语音提问与语音回复)')),
+  enabled
+    ? h('div', { className: 'dim-feishuVoiceForm' },
+        input('凭据引用(环境变量名)', secretRef, setSecretRef, 'DASHSCOPE_API_KEY'),
+        input('转写模型(ASR)', asrModel, setAsrModel, 'qwen3-asr-flash'),
+        input('合成模型(TTS)', ttsModel, setTtsModel, 'qwen3-tts-flash'),
+        input('回复音色', ttsVoice, setTtsVoice, 'Momo'),
+        h('p', { className: 'dim-feishuGroupHelp' },
+          '需要 DashScope 密钥(凭据库中以"凭据引用"为名的环境变量)与可用 ffmpeg;音色支持 Cherry/Serena/Bella/Chelsie/Momo/Katerina 等 qwen3-tts-flash 官方音色。'),
+        h('div', { className: 'dim-feishuVoiceActions' },
+          h('button', {
+            type: 'button',
+            className: 'dim-feishuVoiceSave',
+            onClick: () => void save(),
+            disabled: disabled || saving || !secretRef.trim(),
+          }, '保存语音设置')))
+    : h('p', { className: 'dim-feishuGroupHelp' }, '默认关闭;开启后机器人可听懂语音消息并以语音回复。'),
+  error ? h('p', {
+    className: 'dim-feishuGroupError',
+    role: 'alert',
+  }, error) : null);
+}

package/plugin-src/client/channels/feishu/voice-settings.js

@@ -0,0 +1,71 @@
+import * as React from 'react';
+
+import { h } from '../../i18n.js';
+import { FEISHU_ENDPOINTS, normalizeBotsSnapshot, unwrapRpcResult } from './api.js';
+import { VoiceEditor } from './voice-editor.js';
+
+export function FeishuVoiceSettingsPage({ account, rpcCall }) {
+  const [voice, setVoice] = React.useState(null);
+  const [phase, setPhase] = React.useState('loading');
+  const [error, setError] = React.useState(null);
+  const mounted = React.useRef(true);
+
+  const invoke = React.useCallback(async (endpoint, payload, signal) => {
+    if (typeof rpcCall !== 'function') throw new Error('飞书语音设置暂不可用。');
+    return unwrapRpcResult(await rpcCall(endpoint, payload, signal));
+  }, [rpcCall]);
+
+  const applySnapshot = React.useCallback((value) => {
+    const bot = normalizeBotsSnapshot(value).bots.find((entry) => entry.botId === account.botId);
+    if (!bot) throw new Error('未找到当前飞书机器人，请返回列表后重试。');
+    if (mounted.current) setVoice(bot.voice);
+  }, [account.botId]);
+
+  const loadSettings = React.useCallback(async (signal) => {
+    setPhase('loading');
+    setError(null);
+    try {
+      const value = await invoke(FEISHU_ENDPOINTS.status, {}, signal);
+      if (signal?.aborted || !mounted.current) return;
+      applySnapshot(value);
+      setPhase('ready');
+    } catch (cause) {
+      if (signal?.aborted || !mounted.current) return;
+      setError(cause?.message ?? '飞书语音设置暂不可用。');
+      setPhase('error');
+    }
+  }, [applySnapshot, invoke]);
+
+  React.useEffect(() => {
+    mounted.current = true;
+    const controller = new AbortController();
+    void loadSettings(controller.signal);
+    return () => {
+      mounted.current = false;
+      controller.abort();
+    };
+  }, [loadSettings]);
+
+  const saveVoice = async (settings) => {
+    const value = await invoke(FEISHU_ENDPOINTS.setVoice, {
+      botId: account.botId,
+      voice: settings === null ? null : { ...voice, ...settings },
+    });
+    if (mounted.current) applySnapshot(value);
+  };
+
+  if (phase === 'loading') {
+    return h('div', { className: 'dim-deliveryState', role: 'status', 'aria-busy': 'true' },
+      '正在读取语音设置…');
+  }
+  if (phase === 'error') {
+    return h('div', { className: 'dim-deliveryState', role: 'alert' },
+      h('p', null, error),
+      h('button', {
+        type: 'button',
+        className: 'dim-deliveryButton',
+        onClick: () => void loadSettings(),
+      }, '重新读取'));
+  }
+  return h(VoiceEditor, { value: voice, onSave: saveVoice });
+}

package/plugin-src/client/delivery-settings.js

@@ -2,6 +2,7 @@ import * as React from 'react';
 
 import { AccessPolicySettingsPage } from './access-policy-settings.js';
 import { FeishuGroupSettingsPage } from './channels/feishu/group-settings.js';
+import { FeishuVoiceSettingsPage } from './channels/feishu/voice-settings.js';
 import { h, isEnglish, localizeText } from './i18n.js';
 
 export const DELIVERY_RPC_CHANNEL = '/dsh-im-delivery';
@@ -29,6 +30,7 @@ export const BOT_SETTINGS_TABS = Object.freeze([
 export const FEISHU_BOT_SETTINGS_TABS = Object.freeze([
   ...BOT_SETTINGS_TABS,
   Object.freeze({ id: 'group', label: '群聊' }),
+  Object.freeze({ id: 'voice', label: '语音交互' }),
 ]);
 
 export function botSettingsTabsForChannel(channel) {
@@ -792,6 +794,12 @@ export function DeliveryTargetSettingsPage({
           account,
           rpcCall: accessRpcCall,
         })
+    : activeTab.id === 'voice' && channel === 'feishu'
+      ? h(FeishuVoiceSettingsPage, {
+          key: account.botId,
+          account,
+          rpcCall: accessRpcCall,
+        })
     : h(React.Fragment, null,
   h('section', { className: 'dim-deliveryIdentity', 'aria-labelledby': 'dim-delivery-bot-title' },
     h('div', { className: 'dim-deliveryIdentityHeading' },

package/plugin-src/client/i18n.js

@@ -282,6 +282,22 @@ const EN = Object.freeze({
   '推荐长任务使用：过程与最终答案都在同一张卡片里实时更新，不刷屏': 'Recommended for long tasks: the process and the final answer update live in one card without flooding the chat',
   '每一步都单独发一条消息（含工具调用和过程说明）；注意长任务会连续发送较多消息': 'Every step is sent as its own message (including tool calls and notes); long tasks may send many messages in a row',
   '分步直推设置保存失败，请重试。': 'Could not save the step push setting. Try again.',
+  '语音交互': 'Voice',
+  '正在读取语音设置…': 'Loading voice settings…',
+  '飞书语音设置暂不可用。': 'Feishu voice settings are unavailable.',
+  '未找到当前飞书机器人，请返回列表后重试。': 'This Feishu bot was not found. Return to the bot list and try again.',
+  '查看语音交互说明': 'View voice help',
+  '语音交互开关': 'Voice on/off',
+  '开启(语音提问与语音回复)': 'On (voice questions and voice replies)',
+  '收到语音消息时转写为文字并按文字流水线处理,答案同时以语音回复;转写失败时降级为仅支持文字/图片/文件的提示。凭据值在插件凭据库里管理,这里只引用环境变量名。': 'Voice messages are transcribed to text and processed by the normal text pipeline; the answer is also sent as a voice reply. If transcription fails, the bot falls back to the text/image/file-only notice. The credential value lives in the plugin credential store; only the environment variable name is referenced here.',
+  '凭据引用(环境变量名)': 'Credential reference (environment variable name)',
+  '转写模型(ASR)': 'Transcription model (ASR)',
+  '合成模型(TTS)': 'Synthesis model (TTS)',
+  '回复音色': 'Reply voice',
+  '需要 DashScope 密钥(凭据库中以"凭据引用"为名的环境变量)与可用 ffmpeg;音色支持 Cherry/Serena/Bella/Chelsie/Momo/Katerina 等 qwen3-tts-flash 官方音色。': 'Requires a DashScope API key (the credential-store environment variable named by the credential reference above) and a working ffmpeg. Supported voices include the official qwen3-tts-flash voices Cherry/Serena/Bella/Chelsie/Momo/Katerina.',
+  '保存语音设置': 'Save voice settings',
+  '默认关闭;开启后机器人可听懂语音消息并以语音回复。': 'Off by default. When enabled, the bot understands voice messages and replies by voice.',
+  '语音设置保存失败，请重试。': 'Could not save the voice settings. Try again.',
   'zcode过程卡模式': 'ZCode process-card mode',
   '过程说明、工具摘要与最终答案都在同一张卡片中原地刷新':
     'Process notes, the tool summary, and the final answer refresh in place inside one card',

package/plugin-src/client/styles.js

@@ -479,6 +479,17 @@ const CSS = String.raw`
 .dim-feishuGroupSelect:focus-visible { outline: none; border-color: var(--dsw-alias-state-business-primary, #3370ff); box-shadow: 0 0 0 2px color-mix(in srgb, var(--dsw-alias-state-business-primary, #3370ff) 15%, transparent); }
 .dim-feishuGroupSelect:disabled { cursor: not-allowed; opacity: .55; }
 .dim-feishuGroupHelp { margin: -2px 0 0; color: var(--dsw-alias-label-tertiary, #8f959e); font-size: 12px; line-height: 19px; }
+.dim-feishuVoiceForm { min-width: 0; display: grid; gap: 10px; }
+.dim-feishuVoiceField { min-width: 0; display: grid; gap: 5px; }
+.dim-feishuVoiceLabel { color: var(--dsw-alias-label-secondary, #646a73); font-size: 12px; line-height: 18px; }
+.dim-feishuVoiceInput { min-width: 0; width: 100%; height: 36px; padding: 0 11px; border: 1px solid var(--dsw-alias-border-l2, #dfe1e5); border-radius: 8px; color: var(--dsw-alias-label-primary, #1f2329); background: var(--dsw-alias-bg-layer-1, #fff); font: inherit; font-size: 13px; line-height: 36px; transition: border-color .15s ease, box-shadow .15s ease; }
+.dim-feishuVoiceInput:hover:not(:disabled) { border-color: color-mix(in srgb, var(--dsw-alias-state-business-primary, #3370ff) 42%, var(--dsw-alias-border-l2, #dfe1e5)); }
+.dim-feishuVoiceInput:focus-visible { outline: none; border-color: var(--dsw-alias-state-business-primary, #3370ff); box-shadow: 0 0 0 2px color-mix(in srgb, var(--dsw-alias-state-business-primary, #3370ff) 15%, transparent); }
+.dim-feishuVoiceInput:disabled { cursor: not-allowed; opacity: .55; }
+.dim-feishuVoiceActions { display: flex; justify-content: flex-start; }
+.dim-feishuVoiceSave { height: 34px; padding: 0 15px; border: 1px solid var(--dsw-alias-state-business-primary, #3370ff); border-radius: 8px; color: var(--dsw-alias-state-business-primary, #3370ff); background: var(--dsw-alias-bg-layer-1, #fff); font: inherit; font-size: 13px; line-height: 32px; cursor: pointer; transition: border-color .15s ease, color .15s ease, background .15s ease; }
+.dim-feishuVoiceSave:hover:not(:disabled) { color: #0f5fce; border-color: #0f5fce; background: color-mix(in srgb, #3370ff 6%, var(--dsw-alias-bg-layer-1, #fff)); }
+.dim-feishuVoiceSave:disabled { cursor: not-allowed; opacity: .55; }
 .dim-feishuGroupPermissionAction { display: flex; justify-content: flex-start; }
 .dim-feishuGroupPermissionAction .dim-deliveryButton { color: var(--dsw-alias-state-business-primary, #3370ff); border-color: color-mix(in srgb, var(--dsw-alias-state-business-primary, #3370ff) 30%, var(--dsw-alias-border-l2, #dfe1e5)); }
 .dim-feishuGroupError, .dim-feishuGroupRefreshError { margin: 0; padding: 9px 11px; border-radius: 8px; font-size: 12px; line-height: 18px; }

package/plugin-src/host/channels/feishu/production.mjs

@@ -234,6 +234,16 @@ export async function createProductionController(ctx, config = {}, internals = {
       const workspaceScope = createBotWorkspaceScope(harness, {
         botId: id, workspaces, state, agentPresetCatalog,
       });
+      // 语音是可选渠道能力:凭据读取失败时静默禁用,不阻断机器人连接。
+      let voiceSource = null;
+      if (botConfig.voice) {
+        try {
+          const resolved = await ctx.credentials.resolve(botConfig.voice.secretRef);
+          if (resolved?.value) voiceSource = { config: botConfig.voice, secret: resolved.value };
+        } catch (error) {
+          logger.warn?.('[dsh-feishu-voice] 语音凭据读取失败,语音已禁用:', error?.message ?? String(error));
+        }
+      }
       return new Runtime({
         lark,
         botId: id,
@@ -246,6 +256,7 @@ export async function createProductionController(ctx, config = {}, internals = {
         groupTopicReply: botConfig.groupTopicReply,
         stepPush: botConfig.stepPush,
         stepPushMode: botConfig.stepPushMode,
+        voice: voiceSource,
         sessionSyncTargetsFor: sessionSyncTargetsFor,
         ownerOpenIds: botConfig.ownerOpenIds ?? [botConfig.ownerOpenId],
         harness: workspaceScope.harness,

package/plugin-src/host/channels/feishu/rpc.mjs

@@ -30,6 +30,7 @@ import {
   isFeishuStepPushMode,
   normalizeFeishuStepPushMode,
 } from '../../../../src/channels/feishu/step-push-mode.mjs';
+import { normalizeFeishuVoiceConfig } from '../../../../src/channels/feishu/voice-config.mjs';
 import {
   FEISHU_ENDPOINTS as FEISHU_CLIENT_ENDPOINTS,
   FEISHU_RPC_CHANNEL,
@@ -307,6 +308,7 @@ function publicBotEntry(entry) {
     groupTopicReply: source.groupTopicReply === true,
     stepPush: source.stepPush === true,
     stepPushMode: normalizeFeishuStepPushMode(source.stepPushMode),
+    voice: normalizeFeishuVoiceConfig(source.voice),
     groupMessagePermissionGranted: source.groupMessagePermissionGranted === true,
     bot: publicBot(source.bot),
     health: publicHealth(source, connected),
@@ -490,6 +492,17 @@ function validPayload(endpoint, payload) {
       ? null
       : '请选择分步直推的呈现方式。';
   }
+  if (endpoint === FEISHU_ENDPOINTS.setVoice) {
+    const voice = payload?.voice;
+    const validVoice = voice === null
+      || (typeof voice === 'object' && !Array.isArray(voice)
+        && (voice.enabled === false || normalizeFeishuVoiceConfig(voice) !== null));
+    return hasOnlyKeys(payload, new Set(['botId', 'voice']))
+      && safeOpaqueId(payload.botId)
+      && validVoice
+      ? null
+      : '语音配置无效。';
+  }
   return 'Unknown Feishu endpoint.';
 }
 
@@ -797,6 +810,14 @@ export function createFeishuRpcHandler(controller, { encodeQr = qrCodeDataUrl }
           await controller.updateStepPushMode(payload.botId, payload.stepPushMode),
           { encodeQr: cachedEncodeQr },
         );
+      } else if (endpoint === FEISHU_ENDPOINTS.setVoice) {
+        if (typeof controller.updateVoice !== 'function') {
+          throw new Error('Voice update is unavailable');
+        }
+        value = await toPublicFeishuStatus(
+          await controller.updateVoice(payload.botId, payload.voice),
+          { encodeQr: cachedEncodeQr },
+        );
       } else {
         if (typeof controller.deleteBot !== 'function') throw new Error('Multi-bot delete is unavailable');
         value = await toPublicFeishuStatus(await controller.deleteBot(payload.botId), { encodeQr: cachedEncodeQr });