Quelle  nsSynthVoiceRegistry.cpp   Sprache: C

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim:set ts=2 sw=2 sts=2 et cindent: */
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#include "nsISpeechService.h"
#include "nsServiceManagerUtils.h"
#include "nsCategoryManagerUtils.h"

#include "SpeechSynthesisUtterance.h"
#include "SpeechSynthesisVoice.h"
#include "nsContentUtils.h"
#include "nsSynthVoiceRegistry.h"
#include "nsSpeechTask.h"
#include "AudioChannelService.h"

#include "nsString.h"
#include "mozilla/ClearOnShutdown.h"
#include "mozilla/dom/ContentChild.h"
#include "mozilla/dom/ContentParent.h"
#include "mozilla/dom/Document.h"
#include "mozilla/intl/LocaleService.h"
#include "mozilla/StaticPrefs_media.h"
#include "mozilla/StaticPtr.h"
#include "mozilla/Unused.h"

#include "SpeechSynthesisChild.h"
#include "SpeechSynthesisParent.h"

using mozilla::intl::LocaleService;

#undef LOG
extern mozilla::LogModule* GetSpeechSynthLog();
#define LOG(type, msg) MOZ_LOG(GetSpeechSynthLog(), type, msg)

namespace {

void GetAllSpeechSynthActors(
    nsTArray<mozilla::dom::SpeechSynthesisParent*>& aActors) {
  MOZ_ASSERT(NS_IsMainThread());
  MOZ_ASSERT(aActors.IsEmpty());

  AutoTArray<mozilla::dom::ContentParent*, 20> contentActors;
  mozilla::dom::ContentParent::GetAll(contentActors);

  for (uint32_t contentIndex = 0; contentIndex < contentActors.Length();
       ++contentIndex) {
    MOZ_ASSERT(contentActors[contentIndex]);

    AutoTArray<mozilla::dom::PSpeechSynthesisParent*, 5> speechsynthActors;
    contentActors[contentIndex]->ManagedPSpeechSynthesisParent(
        speechsynthActors);

    for (uint32_t speechsynthIndex = 0;
         speechsynthIndex < speechsynthActors.Length(); ++speechsynthIndex) {
      MOZ_ASSERT(speechsynthActors[speechsynthIndex]);

      mozilla::dom::SpeechSynthesisParent* actor =
          static_cast<mozilla::dom::SpeechSynthesisParent*>(
              speechsynthActors[speechsynthIndex]);
      aActors.AppendElement(actor);
    }
  }
}

}  // namespace

namespace mozilla::dom {

// VoiceData

class VoiceData final {
 private:
  // Private destructor, to discourage deletion outside of Release():
  ~VoiceData() = default;

 public:
  VoiceData(nsISpeechService* aService, const nsAString& aUri,
            const nsAString& aName, const nsAString& aLang, bool aIsLocal,
            bool aQueuesUtterances)
      : mService(aService),
        mUri(aUri),
        mName(aName),
        mLang(aLang),
        mIsLocal(aIsLocal),
        mIsQueued(aQueuesUtterances) {}

  NS_INLINE_DECL_REFCOUNTING(VoiceData)

  nsCOMPtr<nsISpeechService> mService;

  nsString mUri;

  nsString mName;

  nsString mLang;

  bool mIsLocal;

  bool mIsQueued;
};

// GlobalQueueItem

class GlobalQueueItem final {
 private:
  // Private destructor, to discourage deletion outside of Release():
  ~GlobalQueueItem() = default;

 public:
  GlobalQueueItem(VoiceData* aVoice, nsSpeechTask* aTask,
                  const nsAString& aText, const float& aVolume,
                  const float& aRate, const float& aPitch)
      : mVoice(aVoice),
        mTask(aTask),
        mText(aText),
        mVolume(aVolume),
        mRate(aRate),
        mPitch(aPitch),
        mIsLocal(false) {}

  NS_INLINE_DECL_REFCOUNTING(GlobalQueueItem)

  RefPtr<VoiceData> mVoice;

  RefPtr<nsSpeechTask> mTask;

  nsString mText;

  float mVolume;

  float mRate;

  float mPitch;

  bool mIsLocal;
};

// nsSynthVoiceRegistry

static StaticRefPtr<nsSynthVoiceRegistry> gSynthVoiceRegistry;

NS_IMPL_ISUPPORTS(nsSynthVoiceRegistry, nsISynthVoiceRegistry)

nsSynthVoiceRegistry::nsSynthVoiceRegistry()
    : mSpeechSynthChild(nullptr), mUseGlobalQueue(false), mIsSpeaking(false) {
  if (XRE_IsContentProcess()) {
    RefPtr<SpeechSynthesisChild> actor = new SpeechSynthesisChild();
    if (ContentChild::GetSingleton()->SendPSpeechSynthesisConstructor(actor)) {
      mSpeechSynthChild = actor;
    }
  }
}

nsSynthVoiceRegistry::~nsSynthVoiceRegistry() {
  LOG(LogLevel::Debug, ("~nsSynthVoiceRegistry"));

  mUriVoiceMap.Clear();
}

nsSynthVoiceRegistry* nsSynthVoiceRegistry::GetInstance() {
  MOZ_ASSERT(NS_IsMainThread());

  if (!gSynthVoiceRegistry) {
    gSynthVoiceRegistry = new nsSynthVoiceRegistry();
    ClearOnShutdown(&gSynthVoiceRegistry);
    if (XRE_IsParentProcess()) {
      // Start up all speech synth services.
      NS_CreateServicesFromCategory(NS_SPEECH_SYNTH_STARTED, nullptr,
                                    NS_SPEECH_SYNTH_STARTED);
    }
  }

  return gSynthVoiceRegistry;
}

already_AddRefed<nsSynthVoiceRegistry>
nsSynthVoiceRegistry::GetInstanceForService() {
  RefPtr<nsSynthVoiceRegistry> registry = GetInstance();

  return registry.forget();
}

bool nsSynthVoiceRegistry::SendInitialVoicesAndState(
    SpeechSynthesisParent* aParent) {
  MOZ_ASSERT(XRE_IsParentProcess());

  nsTArray<RemoteVoice> voices;
  nsTArray<nsString> defaults;

  for (uint32_t i = 0; i < mVoices.Length(); ++i) {
    RefPtr<VoiceData> voice = mVoices[i];

    voices.AppendElement(RemoteVoice(voice->mUri, voice->mName, voice->mLang,
                                     voice->mIsLocal, voice->mIsQueued));
  }

  for (uint32_t i = 0; i < mDefaultVoices.Length(); ++i) {
    defaults.AppendElement(mDefaultVoices[i]->mUri);
  }

  return aParent->SendInitialVoicesAndState(voices, defaults, IsSpeaking());
}

void nsSynthVoiceRegistry::RecvInitialVoicesAndState(
    const nsTArray<RemoteVoice>& aVoices, const nsTArray<nsString>& aDefaults,
    const bool& aIsSpeaking) {
  // We really should have a local instance since this is a directed response to
  // an Init() call.
  MOZ_ASSERT(gSynthVoiceRegistry);

  for (uint32_t i = 0; i < aVoices.Length(); ++i) {
    RemoteVoice voice = aVoices[i];
    gSynthVoiceRegistry->AddVoiceImpl(nullptr, voice.voiceURI(), voice.name(),
                                      voice.lang(), voice.localService(),
                                      voice.queued());
  }

  for (uint32_t i = 0; i < aDefaults.Length(); ++i) {
    gSynthVoiceRegistry->SetDefaultVoice(aDefaults[i], true);
  }

  gSynthVoiceRegistry->mIsSpeaking = aIsSpeaking;

  if (aVoices.Length()) {
    gSynthVoiceRegistry->NotifyVoicesChanged();
  }
}

void nsSynthVoiceRegistry::RecvRemoveVoice(const nsAString& aUri) {
  // If we dont have a local instance of the registry yet, we will recieve
  // current voices at contruction time.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->RemoveVoice(nullptr, aUri);
}

void nsSynthVoiceRegistry::RecvAddVoice(const RemoteVoice& aVoice) {
  // If we dont have a local instance of the registry yet, we will recieve
  // current voices at contruction time.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->AddVoiceImpl(nullptr, aVoice.voiceURI(), aVoice.name(),
                                    aVoice.lang(), aVoice.localService(),
                                    aVoice.queued());
}

void nsSynthVoiceRegistry::RecvSetDefaultVoice(const nsAString& aUri,
                                               bool aIsDefault) {
  // If we dont have a local instance of the registry yet, we will recieve
  // current voices at contruction time.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->SetDefaultVoice(aUri, aIsDefault);
}

void nsSynthVoiceRegistry::RecvIsSpeakingChanged(bool aIsSpeaking) {
  // If we dont have a local instance of the registry yet, we will get the
  // speaking state on construction.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->mIsSpeaking = aIsSpeaking;
}

void nsSynthVoiceRegistry::RecvNotifyVoicesChanged() {
  // If we dont have a local instance of the registry yet, we don't care.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->NotifyVoicesChanged();
}

void nsSynthVoiceRegistry::RecvNotifyVoicesError(const nsAString& aError) {
  // If we dont have a local instance of the registry yet, we don't care.
  if (!gSynthVoiceRegistry) {
    return;
  }

  gSynthVoiceRegistry->NotifyVoicesError(aError);
}

NS_IMETHODIMP
nsSynthVoiceRegistry::AddVoice(nsISpeechService* aService,
                               const nsAString& aUri, const nsAString& aName,
                               const nsAString& aLang, bool aLocalService,
                               bool aQueuesUtterances) {
  LOG(LogLevel::Debug,
      ("nsSynthVoiceRegistry::AddVoice uri='%s' name='%s' lang='%s' local=%s "
       "queued=%s",
       NS_ConvertUTF16toUTF8(aUri).get(), NS_ConvertUTF16toUTF8(aName).get(),
       NS_ConvertUTF16toUTF8(aLang).get(), aLocalService ? "true" : "false",
       aQueuesUtterances ? "true" : "false"));

  if (NS_WARN_IF(XRE_IsContentProcess())) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  return AddVoiceImpl(aService, aUri, aName, aLang, aLocalService,
                      aQueuesUtterances);
}

NS_IMETHODIMP
nsSynthVoiceRegistry::RemoveVoice(nsISpeechService* aService,
                                  const nsAString& aUri) {
  LOG(LogLevel::Debug, ("nsSynthVoiceRegistry::RemoveVoice uri='%s' (%s)",
                        NS_ConvertUTF16toUTF8(aUri).get(),
                        (XRE_IsContentProcess()) ? "child" : "parent"));

  bool found = false;
  VoiceData* retval = mUriVoiceMap.GetWeak(aUri, &found);

  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }
  if (NS_WARN_IF(!(aService == retval->mService))) {
    return NS_ERROR_INVALID_ARG;
  }

  mVoices.RemoveElement(retval);
  mDefaultVoices.RemoveElement(retval);
  mUriVoiceMap.Remove(aUri);

  if (retval->mIsQueued &&
      !StaticPrefs::media_webspeech_synth_force_global_queue()) {
    // Check if this is the last queued voice, and disable the global queue if
    // it is.
    bool queued = false;
    for (uint32_t i = 0; i < mVoices.Length(); i++) {
      VoiceData* voice = mVoices[i];
      if (voice->mIsQueued) {
        queued = true;
        break;
      }
    }
    if (!queued) {
      mUseGlobalQueue = false;
    }
  }

  nsTArray<SpeechSynthesisParent*> ssplist;
  GetAllSpeechSynthActors(ssplist);

  for (uint32_t i = 0; i < ssplist.Length(); ++i)
    Unused << ssplist[i]->SendVoiceRemoved(aUri);

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::NotifyVoicesChanged() {
  if (XRE_IsParentProcess()) {
    nsTArray<SpeechSynthesisParent*> ssplist;
    GetAllSpeechSynthActors(ssplist);

    for (uint32_t i = 0; i < ssplist.Length(); ++i)
      Unused << ssplist[i]->SendNotifyVoicesChanged();
  }

  nsCOMPtr<nsIObserverService> obs = mozilla::services::GetObserverService();
  if (NS_WARN_IF(!(obs))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  obs->NotifyObservers(nullptr, "synth-voices-changed", nullptr);

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::NotifyVoicesError(const nsAString& aError) {
  if (XRE_IsParentProcess()) {
    nsTArray<SpeechSynthesisParent*> ssplist;
    GetAllSpeechSynthActors(ssplist);

    for (uint32_t i = 0; i < ssplist.Length(); ++i) {
      Unused << ssplist[i]->SendNotifyVoicesError(aError);
    }
  }

  nsCOMPtr<nsIObserverService> obs = mozilla::services::GetObserverService();
  if (NS_WARN_IF(!(obs))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  obs->NotifyObservers(nullptr, "synth-voices-error", aError.BeginReading());

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::SetDefaultVoice(const nsAString& aUri, bool aIsDefault) {
  bool found = false;
  VoiceData* retval = mUriVoiceMap.GetWeak(aUri, &found);
  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  mDefaultVoices.RemoveElement(retval);

  LOG(LogLevel::Debug,
      ("nsSynthVoiceRegistry::SetDefaultVoice %s %s",
       NS_ConvertUTF16toUTF8(aUri).get(), aIsDefault ? "true" : "false"));

  if (aIsDefault) {
    mDefaultVoices.AppendElement(retval);
  }

  if (XRE_IsParentProcess()) {
    nsTArray<SpeechSynthesisParent*> ssplist;
    GetAllSpeechSynthActors(ssplist);

    for (uint32_t i = 0; i < ssplist.Length(); ++i) {
      Unused << ssplist[i]->SendSetDefaultVoice(aUri, aIsDefault);
    }
  }

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::GetVoiceCount(uint32_t* aRetval) {
  *aRetval = mVoices.Length();

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::GetVoice(uint32_t aIndex, nsAString& aRetval) {
  if (NS_WARN_IF(!(aIndex < mVoices.Length()))) {
    return NS_ERROR_INVALID_ARG;
  }

  aRetval = mVoices[aIndex]->mUri;

  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::IsDefaultVoice(const nsAString& aUri, bool* aRetval) {
  bool found;
  VoiceData* voice = mUriVoiceMap.GetWeak(aUri, &found);
  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  for (int32_t i = mDefaultVoices.Length(); i > 0;) {
    VoiceData* defaultVoice = mDefaultVoices[--i];

    if (voice->mLang.Equals(defaultVoice->mLang)) {
      *aRetval = voice == defaultVoice;
      return NS_OK;
    }
  }

  *aRetval = false;
  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::IsLocalVoice(const nsAString& aUri, bool* aRetval) {
  bool found;
  VoiceData* voice = mUriVoiceMap.GetWeak(aUri, &found);
  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  *aRetval = voice->mIsLocal;
  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::GetVoiceLang(const nsAString& aUri, nsAString& aRetval) {
  bool found;
  VoiceData* voice = mUriVoiceMap.GetWeak(aUri, &found);
  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  aRetval = voice->mLang;
  return NS_OK;
}

NS_IMETHODIMP
nsSynthVoiceRegistry::GetVoiceName(const nsAString& aUri, nsAString& aRetval) {
  bool found;
  VoiceData* voice = mUriVoiceMap.GetWeak(aUri, &found);
  if (NS_WARN_IF(!(found))) {
    return NS_ERROR_NOT_AVAILABLE;
  }

  aRetval = voice->mName;
  return NS_OK;
}

nsresult nsSynthVoiceRegistry::AddVoiceImpl(
    nsISpeechService* aService, const nsAString& aUri, const nsAString& aName,
    const nsAString& aLang, bool aLocalService, bool aQueuesUtterances) {
  const bool found = mUriVoiceMap.Contains(aUri);
  if (NS_WARN_IF(found)) {
    return NS_ERROR_INVALID_ARG;
  }

  RefPtr<VoiceData> voice = new VoiceData(aService, aUri, aName, aLang,
                                          aLocalService, aQueuesUtterances);

  mVoices.AppendElement(voice);
  mUriVoiceMap.InsertOrUpdate(aUri, std::move(voice));
  mUseGlobalQueue |= aQueuesUtterances;

  nsTArray<SpeechSynthesisParent*> ssplist;
  GetAllSpeechSynthActors(ssplist);

  if (!ssplist.IsEmpty()) {
    mozilla::dom::RemoteVoice ssvoice(nsString(aUri), nsString(aName),
                                      nsString(aLang), aLocalService,
                                      aQueuesUtterances);

    for (uint32_t i = 0; i < ssplist.Length(); ++i) {
      Unused << ssplist[i]->SendVoiceAdded(ssvoice);
    }
  }

  return NS_OK;
}

bool nsSynthVoiceRegistry::FindVoiceByLang(const nsAString& aLang,
                                           VoiceData** aRetval) {
  nsAString::const_iterator dashPos, start, end;
  aLang.BeginReading(start);
  aLang.EndReading(end);

  while (true) {
    nsAutoString langPrefix(Substring(start, end));

    for (int32_t i = mDefaultVoices.Length(); i > 0;) {
      VoiceData* voice = mDefaultVoices[--i];

      if (StringBeginsWith(voice->mLang, langPrefix)) {
        *aRetval = voice;
        return true;
      }
    }

    for (int32_t i = mVoices.Length(); i > 0;) {
      VoiceData* voice = mVoices[--i];

      if (StringBeginsWith(voice->mLang, langPrefix)) {
        *aRetval = voice;
        return true;
      }
    }

    dashPos = end;
    end = start;

    if (!RFindInReadable(u"-"_ns, end, dashPos)) {
      break;
    }
  }

  return false;
}

VoiceData* nsSynthVoiceRegistry::FindBestMatch(const nsAString& aUri,
                                               const nsAString& aLang) {
  if (mVoices.IsEmpty()) {
    return nullptr;
  }

  bool found = false;
  VoiceData* retval = mUriVoiceMap.GetWeak(aUri, &found);

  if (found) {
    LOG(LogLevel::Debug, ("nsSynthVoiceRegistry::FindBestMatch - Matched URI"));
    return retval;
  }

  // Try finding a match for given voice.
  if (!aLang.IsVoid() && !aLang.IsEmpty()) {
    if (FindVoiceByLang(aLang, &retval)) {
      LOG(LogLevel::Debug,
          ("nsSynthVoiceRegistry::FindBestMatch - Matched language (%s ~= %s)",
           NS_ConvertUTF16toUTF8(aLang).get(),
           NS_ConvertUTF16toUTF8(retval->mLang).get()));

      return retval;
    }
  }

  // Try UI language.
  nsAutoCString uiLang;
  LocaleService::GetInstance()->GetAppLocaleAsBCP47(uiLang);

  if (FindVoiceByLang(NS_ConvertASCIItoUTF16(uiLang), &retval)) {
    LOG(LogLevel::Debug,
        ("nsSynthVoiceRegistry::FindBestMatch - Matched UI language (%s ~= %s)",
         uiLang.get(), NS_ConvertUTF16toUTF8(retval->mLang).get()));

    return retval;
  }

  // Try en-US, the language of locale "C"
  if (FindVoiceByLang(u"en-US"_ns, &retval)) {
    LOG(LogLevel::Debug, ("nsSynthVoiceRegistry::FindBestMatch - Matched C "
                          "locale language (en-US ~= %s)",
                          NS_ConvertUTF16toUTF8(retval->mLang).get()));

    return retval;
  }

  // The top default voice is better than nothing...
  if (!mDefaultVoices.IsEmpty()) {
    return mDefaultVoices.LastElement();
  }

  return nullptr;
}

already_AddRefed<nsSpeechTask> nsSynthVoiceRegistry::SpeakUtterance(
    SpeechSynthesisUtterance& aUtterance, const nsAString& aDocLang) {
  nsString lang =
      nsString(aUtterance.mLang.IsEmpty() ? aDocLang : aUtterance.mLang);
  nsAutoString uri;

  if (aUtterance.mVoice) {
    aUtterance.mVoice->GetVoiceURI(uri);
  }

  // Get current audio volume to apply speech call
  float volume = aUtterance.Volume();
  RefPtr<AudioChannelService> service = AudioChannelService::GetOrCreate();
  if (service) {
    if (nsCOMPtr<nsPIDOMWindowInner> topWindow = aUtterance.GetOwnerWindow()) {
      // TODO : use audio channel agent, open new bug to fix it.
      AudioPlaybackConfig config =
          service->GetMediaConfig(topWindow->GetOuterWindow());
      volume = config.mMuted ? 0.0f : config.mVolume * volume;
    }
  }

  RefPtr<nsSpeechTask> task;
  if (XRE_IsContentProcess()) {
    task = new SpeechTaskChild(&aUtterance,
                               aUtterance.ShouldResistFingerprinting());
    SpeechSynthesisRequestChild* actor = new SpeechSynthesisRequestChild(
        static_cast<SpeechTaskChild*>(task.get()));
    mSpeechSynthChild->SendPSpeechSynthesisRequestConstructor(
        actor, aUtterance.mText, lang, uri, volume, aUtterance.Rate(),
        aUtterance.Pitch(), aUtterance.ShouldResistFingerprinting());
  } else {
    task =
        new nsSpeechTask(&aUtterance, aUtterance.ShouldResistFingerprinting());
    Speak(aUtterance.mText, lang, uri, volume, aUtterance.Rate(),
          aUtterance.Pitch(), task);
  }

  return task.forget();
}

void nsSynthVoiceRegistry::Speak(const nsAString& aText, const nsAString& aLang,
                                 const nsAString& aUri, const float& aVolume,
                                 const float& aRate, const float& aPitch,
                                 nsSpeechTask* aTask) {
  MOZ_ASSERT(XRE_IsParentProcess());

  if (aTask->ShouldResistFingerprinting()) {
    aTask->ForceError(0, 0);
    return;
  }

  VoiceData* voice = FindBestMatch(aUri, aLang);

  if (!voice) {
    NS_WARNING("No voices found.");
    aTask->ForceError(0, 0);
    return;
  }

  aTask->SetChosenVoiceURI(voice->mUri);

  if (mUseGlobalQueue ||
      StaticPrefs::media_webspeech_synth_force_global_queue()) {
    LOG(LogLevel::Debug,
        ("nsSynthVoiceRegistry::Speak queueing text='%s' lang='%s' uri='%s' "
         "rate=%f pitch=%f",
         NS_ConvertUTF16toUTF8(aText).get(), NS_ConvertUTF16toUTF8(aLang).get(),
         NS_ConvertUTF16toUTF8(aUri).get(), aRate, aPitch));
    RefPtr<GlobalQueueItem> item =
        new GlobalQueueItem(voice, aTask, aText, aVolume, aRate, aPitch);
    mGlobalQueue.AppendElement(item);

    if (mGlobalQueue.Length() == 1) {
      SpeakImpl(item->mVoice, item->mTask, item->mText, item->mVolume,
                item->mRate, item->mPitch);
    }
  } else {
    SpeakImpl(voice, aTask, aText, aVolume, aRate, aPitch);
  }
}

void nsSynthVoiceRegistry::SpeakNext() {
  MOZ_ASSERT(XRE_IsParentProcess());

  LOG(LogLevel::Debug,
      ("nsSynthVoiceRegistry::SpeakNext %d", mGlobalQueue.IsEmpty()));

  SetIsSpeaking(false);

  if (mGlobalQueue.IsEmpty()) {
    return;
  }

  mGlobalQueue.RemoveElementAt(0);

  while (!mGlobalQueue.IsEmpty()) {
    RefPtr<GlobalQueueItem> item = mGlobalQueue.ElementAt(0);
    if (item->mTask->IsPreCanceled()) {
      mGlobalQueue.RemoveElementAt(0);
      continue;
    }
    if (!item->mTask->IsPrePaused()) {
      SpeakImpl(item->mVoice, item->mTask, item->mText, item->mVolume,
                item->mRate, item->mPitch);
    }
    break;
  }
}

void nsSynthVoiceRegistry::ResumeQueue() {
  MOZ_ASSERT(XRE_IsParentProcess());
  LOG(LogLevel::Debug,
      ("nsSynthVoiceRegistry::ResumeQueue %d", mGlobalQueue.IsEmpty()));

  if (mGlobalQueue.IsEmpty()) {
    return;
  }

  RefPtr<GlobalQueueItem> item = mGlobalQueue.ElementAt(0);
  if (!item->mTask->IsPrePaused()) {
    SpeakImpl(item->mVoice, item->mTask, item->mText, item->mVolume,
              item->mRate, item->mPitch);
  }
}

bool nsSynthVoiceRegistry::IsSpeaking() { return mIsSpeaking; }

void nsSynthVoiceRegistry::SetIsSpeaking(bool aIsSpeaking) {
  MOZ_ASSERT(XRE_IsParentProcess());

  // Only set to 'true' if global queue is enabled.
  mIsSpeaking =
      aIsSpeaking && (mUseGlobalQueue ||
                      StaticPrefs::media_webspeech_synth_force_global_queue());

  nsTArray<SpeechSynthesisParent*> ssplist;
  GetAllSpeechSynthActors(ssplist);
  for (uint32_t i = 0; i < ssplist.Length(); ++i) {
    Unused << ssplist[i]->SendIsSpeakingChanged(aIsSpeaking);
  }
}

void nsSynthVoiceRegistry::SpeakImpl(VoiceData* aVoice, nsSpeechTask* aTask,
                                     const nsAString& aText,
                                     const float& aVolume, const float& aRate,
                                     const float& aPitch) {
  LOG(LogLevel::Debug,
      ("nsSynthVoiceRegistry::SpeakImpl queueing text='%s' uri='%s' rate=%f "
       "pitch=%f",
       NS_ConvertUTF16toUTF8(aText).get(),
       NS_ConvertUTF16toUTF8(aVoice->mUri).get(), aRate, aPitch));

  aTask->Init();

  if (NS_FAILED(aVoice->mService->Speak(aText, aVoice->mUri, aVolume, aRate,
                                        aPitch, aTask))) {
    aTask->DispatchError(0, 0);
  }
}

}  // namespace mozilla::dom