fix(android): prevent duplicate talk playback

This commit is contained in:
Peter Steinberger
2026-04-25 20:43:01 +01:00
parent 346a72ddb9
commit fb7b798f96
3 changed files with 92 additions and 13 deletions

View File

@@ -226,14 +226,15 @@ class TalkModeManager(
// If this is a response we initiated, handle normally below.
// Otherwise, if ttsOnAllResponses, finish streaming TTS on terminal events.
val pending = pendingRunId
if (pending == null || runId != pending) {
val knownRun = pending == runId || hasRunCompletion(runId)
if (!knownRun) {
if (ttsOnAllResponses && state == "final") {
val text = extractTextFromChatEventMessage(obj["message"])
if (!text.isNullOrBlank()) {
playTtsForText(text)
}
}
if (pending == null || runId != pending) return
return
}
Log.d(tag, "chat event arrived runId=$runId state=$state pendingRunId=$pendingRunId")
val terminal =
@@ -539,6 +540,7 @@ class TalkModeManager(
private suspend fun sendChat(message: String, session: GatewaySession): String {
val runId = UUID.randomUUID().toString()
armPendingRun(runId)
val params =
buildJsonObject {
put("sessionKey", JsonPrimitive(mainSessionKey.ifBlank { "main" }))
@@ -547,19 +549,29 @@ class TalkModeManager(
put("timeoutMs", JsonPrimitive(30_000))
put("idempotencyKey", JsonPrimitive(runId))
}
val res = session.request("chat.send", params.toString())
val parsed = parseRunId(res) ?: runId
if (parsed != runId) {
pendingRunId = parsed
try {
val res = session.request("chat.send", params.toString())
val parsed = parseRunId(res) ?: runId
if (parsed != runId) {
pendingRunId = parsed
}
return parsed
} catch (err: Throwable) {
clearPendingRun(runId)
throw err
}
return parsed
}
private suspend fun waitForChatFinal(runId: String): Boolean {
pendingFinal?.cancel()
val deferred = CompletableDeferred<Boolean>()
pendingRunId = runId
pendingFinal = deferred
consumeRunCompletion(runId)?.let { return it }
val deferred =
if (pendingRunId == runId) {
pendingFinal ?: armPendingRun(runId)
} else {
armPendingRun(runId)
}
consumeRunCompletion(runId)?.let { return it }
val result =
withContext(Dispatchers.IO) {
@@ -570,11 +582,25 @@ class TalkModeManager(
}
}
if (!result) {
if (!result && pendingRunId == runId) {
clearPendingRun(runId)
}
return result
}
private fun armPendingRun(runId: String): CompletableDeferred<Boolean> {
pendingFinal?.cancel()
val deferred = CompletableDeferred<Boolean>()
pendingRunId = runId
pendingFinal = deferred
return deferred
}
private fun clearPendingRun(runId: String) {
if (pendingRunId == runId) {
pendingFinal = null
pendingRunId = null
}
return result
}
private fun cacheRunCompletion(runId: String, isFinal: Boolean) {
@@ -593,6 +619,12 @@ class TalkModeManager(
}
}
private fun hasRunCompletion(runId: String): Boolean {
synchronized(completedRunsLock) {
return completedRunStates.containsKey(runId)
}
}
private fun consumeRunText(runId: String): String? {
synchronized(completedRunsLock) {
return completedRunTexts.remove(runId)

View File

@@ -5,6 +5,7 @@ import ai.openclaw.app.gateway.DeviceAuthTokenStore
import ai.openclaw.app.gateway.DeviceIdentityStore
import ai.openclaw.app.gateway.GatewaySession
import java.util.concurrent.atomic.AtomicLong
import kotlinx.coroutines.CompletableDeferred
import kotlinx.coroutines.CoroutineScope
import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.Job
@@ -49,6 +50,34 @@ class TalkModeManagerTest {
assertEquals(12L, playbackGeneration(manager).get())
}
@Test
fun duplicateFinalForPendingTalkRunDoesNotStartAllResponseTts() {
val manager = createManager()
val final = CompletableDeferred<Boolean>()
manager.ttsOnAllResponses = true
setPrivateField(manager, "pendingRunId", "run-talk")
setPrivateField(manager, "pendingFinal", final)
manager.handleGatewayEvent("chat", chatFinalPayload(runId = "run-talk", text = "spoken once"))
assertTrue(final.isCompleted)
assertEquals(0L, playbackGeneration(manager).get())
manager.handleGatewayEvent("chat", chatFinalPayload(runId = "run-talk", text = "spoken once"))
assertEquals(0L, playbackGeneration(manager).get())
}
@Test
fun nonPendingFinalStillUsesAllResponseTts() {
val manager = createManager()
manager.ttsOnAllResponses = true
manager.handleGatewayEvent("chat", chatFinalPayload(runId = "run-other", text = "speak this"))
assertEquals(1L, playbackGeneration(manager).get())
}
private fun createManager(): TalkModeManager {
val app = RuntimeEnvironment.getApplication()
val sessionJob = SupervisorJob()
@@ -86,6 +115,22 @@ class TalkModeManagerTest {
field.isAccessible = true
return field.get(target)
}
private fun chatFinalPayload(runId: String, text: String): String {
return """
{
"runId": "$runId",
"sessionKey": "main",
"state": "final",
"message": {
"role": "assistant",
"content": [
{ "type": "text", "text": "$text" }
]
}
}
""".trimIndent()
}
}
private class InMemoryDeviceAuthStore : DeviceAuthTokenStore {