fix(macos): prevent Talk Mode audio hang

apps/macos/Sources/Clawdis/TalkAudioPlayer.swift

@@ -8,43 +8,142 @@ final class TalkAudioPlayer: NSObject, @preconcurrency AVAudioPlayerDelegate {
 
     private let logger = Logger(subsystem: "com.steipete.clawdis", category: "talk.tts")
     private var player: AVAudioPlayer?
-    private var continuation: CheckedContinuation<TalkPlaybackResult, Never>?
+    private var playback: Playback?
+
+    private final class Playback: @unchecked Sendable {
+        private let lock = NSLock()
+        private var finished = false
+        private var continuation: CheckedContinuation<TalkPlaybackResult, Never>?
+        private var watchdog: Task<Void, Never>?
+
+        func setContinuation(_ continuation: CheckedContinuation<TalkPlaybackResult, Never>) {
+            self.lock.lock()
+            defer { self.lock.unlock() }
+            self.continuation = continuation
+        }
+
+        func setWatchdog(_ task: Task<Void, Never>?) {
+            self.lock.lock()
+            let old = self.watchdog
+            self.watchdog = task
+            self.lock.unlock()
+            old?.cancel()
+        }
+
+        func cancelWatchdog() {
+            self.setWatchdog(nil)
+        }
+
+        func finish(_ result: TalkPlaybackResult) {
+            let continuation: CheckedContinuation<TalkPlaybackResult, Never>?
+            self.lock.lock()
+            if self.finished {
+                continuation = nil
+            } else {
+                self.finished = true
+                continuation = self.continuation
+                self.continuation = nil
+            }
+            self.lock.unlock()
+            continuation?.resume(returning: result)
+        }
+    }
 
     func play(data: Data) async -> TalkPlaybackResult {
-        self.stopInternal(interrupted: true)
+        self.stopInternal()
-        do {
+
-            let player = try AVAudioPlayer(data: data)
+        let playback = Playback()
-            self.player = player
+        self.playback = playback
-            player.delegate = self
+
-            player.prepareToPlay()
+        return await withCheckedContinuation { continuation in
-            player.play()
+            playback.setContinuation(continuation)
-            return await withCheckedContinuation { continuation in
+            do {
-                self.continuation = continuation
+                let player = try AVAudioPlayer(data: data)
+                self.player = player
+
+                player.delegate = self
+                player.prepareToPlay()
+
+                self.armWatchdog(playback: playback)
+
+                let ok = player.play()
+                if !ok {
+                    self.logger.error("talk audio player refused to play")
+                    self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
+                }
+            } catch {
+                self.logger.error("talk audio player failed: \(error.localizedDescription, privacy: .public)")
+                self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
             }
-        } catch {
-            self.logger.error("talk audio player failed: \(error.localizedDescription, privacy: .public)")
-            return TalkPlaybackResult(finished: false, interruptedAt: nil)
         }
     }
 
     func stop() -> Double? {
         guard let player else { return nil }
         let time = player.currentTime
-        self.stopInternal(interrupted: true, interruptedAt: time)
+        self.stopInternal(interruptedAt: time)
         return time
     }
 
     func audioPlayerDidFinishPlaying(_: AVAudioPlayer, successfully flag: Bool) {
-        self.stopInternal(interrupted: !flag)
+        self.stopInternal(finished: flag)
     }
 
-    private func stopInternal(interrupted: Bool, interruptedAt: Double? = nil) {
+    private func stopInternal(finished: Bool = false, interruptedAt: Double? = nil) {
+        guard let playback else { return }
+        let result = TalkPlaybackResult(finished: finished, interruptedAt: interruptedAt)
+        self.finish(playback: playback, result: result)
+    }
+
+    private func finish(playback: Playback, result: TalkPlaybackResult) {
+        playback.cancelWatchdog()
+        playback.finish(result)
+
+        guard self.playback === playback else { return }
+        self.playback = nil
         self.player?.stop()
         self.player = nil
-        if let continuation {
+    }
-            self.continuation = nil
+
-            continuation.resume(returning: TalkPlaybackResult(finished: !interrupted, interruptedAt: interruptedAt))
+    private func stopInternal() {
-        }
+        self.playback?.cancelWatchdog()
+        self.playback = nil
+        self.player?.stop()
+        self.player = nil
+    }
+
+    private func armWatchdog(playback: Playback) {
+        playback.setWatchdog(Task { @MainActor [weak self] in
+            guard let self else { return }
+
+            do {
+                try await Task.sleep(nanoseconds: 650_000_000)
+            } catch {
+                return
+            }
+            if Task.isCancelled { return }
+
+            guard self.playback === playback else { return }
+            if self.player?.isPlaying != true {
+                self.logger.error("talk audio player did not start playing")
+                self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
+                return
+            }
+
+            let duration = self.player?.duration ?? 0
+            let timeoutSeconds = min(max(2.0, duration + 2.0), 5 * 60.0)
+            do {
+                try await Task.sleep(nanoseconds: UInt64(timeoutSeconds * 1_000_000_000))
+            } catch {
+                return
+            }
+            if Task.isCancelled { return }
+
+            guard self.playback === playback else { return }
+            guard self.player?.isPlaying == true else { return }
+            self.logger.error("talk audio player watchdog fired")
+            self.finish(playback: playback, result: TalkPlaybackResult(finished: false, interruptedAt: nil))
+        })
     }
 }
 

apps/macos/Tests/ClawdisIPCTests/TalkAudioPlayerTests.swift

@@ -0,0 +1,79 @@
+import Foundation
+import Testing
+@testable import Clawdis
+
+@Suite(.serialized) struct TalkAudioPlayerTests {
+    @MainActor
+    @Test func playDoesNotHangWhenPlaybackEndsOrFails() async throws {
+        let wav = makeWav16Mono(sampleRate: 8000, samples: 80)
+        defer { _ = TalkAudioPlayer.shared.stop() }
+
+        _ = try await withTimeout(seconds: 2.0) {
+            await TalkAudioPlayer.shared.play(data: wav)
+        }
+
+        #expect(true)
+    }
+}
+
+private struct TimeoutError: Error {}
+
+private func withTimeout<T: Sendable>(
+    seconds: Double,
+    _ work: @escaping @Sendable () async throws -> T) async throws -> T
+{
+    try await withThrowingTaskGroup(of: T.self) { group in
+        group.addTask {
+            try await work()
+        }
+        group.addTask {
+            try await Task.sleep(nanoseconds: UInt64(seconds * 1_000_000_000))
+            throw TimeoutError()
+        }
+        let result = try await group.next()
+        group.cancelAll()
+        guard let result else { throw TimeoutError() }
+        return result
+    }
+}
+
+private func makeWav16Mono(sampleRate: UInt32, samples: Int) -> Data {
+    let channels: UInt16 = 1
+    let bitsPerSample: UInt16 = 16
+    let blockAlign = channels * (bitsPerSample / 8)
+    let byteRate = sampleRate * UInt32(blockAlign)
+    let dataSize = UInt32(samples) * UInt32(blockAlign)
+
+    var data = Data()
+    data.append(contentsOf: [0x52, 0x49, 0x46, 0x46]) // RIFF
+    data.appendLEUInt32(36 + dataSize)
+    data.append(contentsOf: [0x57, 0x41, 0x56, 0x45]) // WAVE
+
+    data.append(contentsOf: [0x66, 0x6D, 0x74, 0x20]) // fmt
+    data.appendLEUInt32(16) // PCM
+    data.appendLEUInt16(1) // audioFormat
+    data.appendLEUInt16(channels)
+    data.appendLEUInt32(sampleRate)
+    data.appendLEUInt32(byteRate)
+    data.appendLEUInt16(blockAlign)
+    data.appendLEUInt16(bitsPerSample)
+
+    data.append(contentsOf: [0x64, 0x61, 0x74, 0x61]) // data
+    data.appendLEUInt32(dataSize)
+
+    // Silence samples.
+    data.append(Data(repeating: 0, count: Int(dataSize)))
+    return data
+}
+
+private extension Data {
+    mutating func appendLEUInt16(_ value: UInt16) {
+        var v = value.littleEndian
+        Swift.withUnsafeBytes(of: &v) { append(contentsOf: $0) }
+    }
+
+    mutating func appendLEUInt32(_ value: UInt32) {
+        var v = value.littleEndian
+        Swift.withUnsafeBytes(of: &v) { append(contentsOf: $0) }
+    }
+}