feat(chat): add voice input to message composer with dual-mode recognition

.github/workflows/build-windows.yml

@@ -0,0 +1,62 @@
+name: Build Windows Native
+
+on:
+  push:
+    branches:
+      - feature/voice-input
+  workflow_dispatch:
+
+env:
+  RUST_BACKTRACE: 1
+  HUSKY: 0
+
+jobs:
+  build:
+    name: Build Windows App
+    runs-on: windows-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          version: 11.9.0
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: 22
+          cache: "pnpm"
+
+      - name: Install Rust
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: x86_64-pc-windows-msvc
+
+      - name: Rust Cache
+        uses: swatinem/rust-cache@v2
+        with:
+          workspaces: "./src-tauri -> target"
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      - name: Build codeg-server
+        run: pnpm server:build
+
+      - name: Build Tauri App
+        run: pnpm tauri build --bundles nsis
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Upload Build Artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: codeg-windows-voice-build
+          path: |
+            src-tauri/target/release/codeg.exe
+            src-tauri/target/release/codeg-server.exe
+            src-tauri/target/release/bundle/nsis/*.exe
+          if-no-files-found: error

src-tauri/tauri.conf.json

@@ -18,7 +18,7 @@
   },
   "bundle": {
     "active": true,
-    "createUpdaterArtifacts": true,
+    "createUpdaterArtifacts": false,
     "targets": "all",
     "icon": [
       "icons/32x32.png",

src/components/chat/composer/composer-voice-button.test.tsx

@@ -0,0 +1,77 @@
+import { describe, it, expect, vi } from "vitest"
+import { render, screen, fireEvent } from "@testing-library/react"
+import { ComposerVoiceButton } from "./composer-voice-button"
+import type { UseVoiceInputReturn } from "./use-voice-input"
+
+function createMockVoice(overrides: Partial<UseVoiceInputReturn> = {}): UseVoiceInputReturn {
+  return {
+    status: "idle",
+    mode: "web-speech",
+    isSupported: true,
+    volume: 0,
+    interimText: "",
+    errorMessage: null,
+    startListening: vi.fn(),
+    stopListening: vi.fn(),
+    cancel: vi.fn(),
+    ...overrides,
+  }
+}
+
+describe("ComposerVoiceButton", () => {
+  it("renders idle button with microphone icon", () => {
+    const voice = createMockVoice()
+    render(<ComposerVoiceButton voice={voice} />)
+
+    const button = screen.getByTestId("composer-voice-button")
+    expect(button).toBeInTheDocument()
+    expect(button).not.toBeDisabled()
+  })
+
+  it("calls startListening on click when idle", () => {
+    const voice = createMockVoice()
+    render(<ComposerVoiceButton voice={voice} />)
+
+    fireEvent.click(screen.getByTestId("composer-voice-button"))
+    expect(voice.startListening).toHaveBeenCalled()
+  })
+
+  it("renders active state and calls stopListening on click when listening", () => {
+    const voice = createMockVoice({ status: "listening" })
+    render(<ComposerVoiceButton voice={voice} />)
+
+    const button = screen.getByTestId("composer-voice-button")
+    expect(button.className).toContain("text-red-500")
+
+    fireEvent.click(button)
+    expect(voice.stopListening).toHaveBeenCalled()
+  })
+
+  it("shows interim text bubble when speaking", () => {
+    const voice = createMockVoice({
+      status: "listening",
+      interimText: "测试语音内容",
+    })
+    render(<ComposerVoiceButton voice={voice} />)
+
+    const bubble = screen.getByTestId("voice-interim-bubble")
+    expect(bubble).toBeInTheDocument()
+    expect(bubble).toHaveTextContent("测试语音内容")
+  })
+
+  it("cancels on Escape key while listening", () => {
+    const voice = createMockVoice({ status: "listening" })
+    render(<ComposerVoiceButton voice={voice} />)
+
+    fireEvent.keyDown(window, { key: "Escape" })
+    expect(voice.cancel).toHaveBeenCalled()
+  })
+
+  it("disables button when transcribing", () => {
+    const voice = createMockVoice({ status: "transcribing" })
+    render(<ComposerVoiceButton voice={voice} />)
+
+    const button = screen.getByTestId("composer-voice-button")
+    expect(button).toBeDisabled()
+  })
+})

src/components/chat/composer/composer-voice-button.tsx

@@ -0,0 +1,118 @@
+import { useEffect } from "react"
+import { Button } from "@/components/ui/button"
+import { Mic, Loader2 } from "lucide-react"
+import { cn } from "@/lib/utils"
+import type { UseVoiceInputReturn } from "./use-voice-input"
+import { toast } from "sonner"
+
+export interface ComposerVoiceButtonProps {
+  voice: UseVoiceInputReturn
+  disabled?: boolean
+  className?: string
+  /** Localized string for default tooltip */
+  label?: string
+  /** Localized string for listening state tooltip */
+  listeningLabel?: string
+  /** Localized string for transcribing state tooltip */
+  transcribingLabel?: string
+  /** Localized string for permission denied */
+  permissionDeniedLabel?: string
+}
+
+export function ComposerVoiceButton({
+  voice,
+  disabled = false,
+  className,
+  label = "语音输入 (点击开始)",
+  listeningLabel = "正在录音... 点击完成 (Esc 取消)",
+  transcribingLabel = "正在转写...",
+  permissionDeniedLabel = "麦克风权限被拒绝，请在设置中允许访问",
+}: ComposerVoiceButtonProps) {
+  const isListening = voice.status === "listening"
+  const isTranscribing = voice.status === "transcribing"
+
+  // Cancel on Escape key while listening
+  useEffect(() => {
+    if (!isListening) return
+
+    const handleKeyDown = (e: KeyboardEvent) => {
+      if (e.key === "Escape") {
+        e.preventDefault()
+        voice.cancel()
+      }
+    }
+
+    window.addEventListener("keydown", handleKeyDown)
+    return () => window.removeEventListener("keydown", handleKeyDown)
+  }, [isListening, voice])
+
+  // Toast on permission error
+  useEffect(() => {
+    if (voice.status === "error" && voice.errorMessage === "micPermissionDenied") {
+      toast.error(permissionDeniedLabel)
+    }
+  }, [voice.status, voice.errorMessage, permissionDeniedLabel])
+
+  const handleClick = (e: React.MouseEvent) => {
+    e.preventDefault()
+    e.stopPropagation()
+    if (disabled || isTranscribing) return
+
+    if (isListening) {
+      voice.stopListening()
+    } else {
+      void voice.startListening()
+    }
+  }
+
+  const title = isTranscribing ? transcribingLabel : isListening ? listeningLabel : label
+
+  return (
+    <div className="relative inline-flex items-center">
+      {/* Live speech interim preview bubble while user is speaking */}
+      {isListening && voice.interimText && (
+        <div
+          data-testid="voice-interim-bubble"
+          className="absolute bottom-full right-0 mb-2 z-50 max-w-xs truncate rounded-md bg-popover/95 px-2.5 py-1 text-xs text-popover-foreground shadow-md border border-border/80 backdrop-blur-sm animate-in fade-in zoom-in-95 duration-150"
+        >
+          <span className="inline-block size-1.5 rounded-full bg-red-500 mr-1.5 animate-pulse" />
+          {voice.interimText}
+        </div>
+      )}
+
+      <Button
+        type="button"
+        variant={isListening ? "outline" : "ghost"}
+        size="icon"
+        onClick={handleClick}
+        disabled={disabled || isTranscribing || !voice.isSupported}
+        title={title}
+        aria-label={title}
+        data-testid="composer-voice-button"
+        className={cn(
+          "relative h-8 w-8 transition-colors",
+          isListening && [
+            "border-red-500/40 bg-red-500/10 text-red-500 hover:bg-red-500/20 hover:text-red-600",
+            "dark:border-red-500/50 dark:bg-red-950/30 dark:text-red-400",
+          ],
+          className
+        )}
+      >
+        {isListening && (
+          <span
+            className="absolute inset-0 rounded-md bg-red-500/20 animate-ping opacity-75 pointer-events-none"
+            style={{
+              animationDuration: "1.5s",
+              transform: `scale(${1 + voice.volume * 0.3})`,
+            }}
+          />
+        )}
+        {isTranscribing ? (
+          <Loader2 className="size-4 animate-spin text-primary" />
+        ) : (
+          <Mic className={cn("size-4", isListening ? "animate-pulse text-red-500" : "text-muted-foreground")} />
+        )}
+      </Button>
+    </div>
+  )
+}