Stop TtsUtterance if its associated WebContents is destroyed

In
https://chromium-review.googlesource.com/c/chromium/src/+/2261098

TtsUtteranceImpl started taking an optional WebContents in its
constructor.

When the WebContents is destroyed, the utterance is cleared and stopped
from triggering tts.

This change exposes this behavior up to c/p/b and cleans up
AccessibilityManager which approximated this behavior by calling
TtsController::Stop().

Some notable examples that benefit from this:
- extensions that trigger tts utterances, when destroyed, stop those
  utterances
- the tts settings webui, if closed, will stop tts utterances in
  progress

Change-Id: I7169b7ea3093e24e093c6c478040d943f34353f5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2541002
Commit-Queue: David Tseng <dtseng@chromium.org>
Reviewed-by: Avi Drissman <avi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#828529}

chrome/browser/chromeos/accessibility/DEPS

@@ -3,7 +3,6 @@ specific_include_rules = {
     "+ash/root_window_controller.h",
     "+ash/shell.h",
     "+ash/sticky_keys/sticky_keys_controller.h",
-    "+content/public/browser/tts_controller.h",
   ],
   "magnification_manager\.cc": [
     "+ash/magnifier/magnification_controller.h",

chrome/browser/chromeos/accessibility/accessibility_manager.cc

@@ -71,7 +71,6 @@
 #include "content/public/browser/notification_details.h"
 #include "content/public/browser/notification_service.h"
 #include "content/public/browser/notification_source.h"
-#include "content/public/browser/tts_controller.h"
 #include "content/public/browser/web_ui.h"
 #include "content/public/common/content_switches.h"
 #include "extensions/browser/api/virtual_keyboard_private/virtual_keyboard_delegate.h"
@@ -1399,9 +1398,6 @@ void AccessibilityManager::PostUnloadChromeVox() {
   // In case the user darkened the screen, undarken it now.
   SetDarkenScreen(false);
 
-  // Stop speech.
-  content::TtsController::GetInstance()->Stop();
-
   audio_focus_manager_->SetEnforcementMode(
       media_session::mojom::EnforcementMode::kDefault);
 }
@@ -1434,9 +1430,6 @@ void AccessibilityManager::PostUnloadSelectToSpeak() {
   // Clear the accessibility focus ring and highlight.
   RemoveFocusRings(extension_misc::kSelectToSpeakExtensionId);
   HideHighlights();
-
-  // Stop speech.
-  content::TtsController::GetInstance()->Stop();
 }
 
 void AccessibilityManager::PostLoadSwitchAccess() {

chrome/browser/profiles/profile_activity_metrics_recorder.cc

@@ -35,7 +35,8 @@ int GetMetricsBucketIndex(const Profile* profile) {
     return 0;
 
   ProfileAttributesEntry* entry;
-  if (!g_browser_process->profile_manager()
+  if (!g_browser_process->profile_manager() ||
+      !g_browser_process->profile_manager()
            ->GetProfileAttributesStorage()
            .GetProfileAttributesWithPath(profile->GetPath(), &entry)) {
     // This can happen if the profile is deleted.

chrome/browser/speech/extension_api/tts_extension_api.cc

@@ -20,6 +20,8 @@
 #include "content/public/browser/tts_controller.h"
 #include "extensions/browser/event_router.h"
 #include "extensions/browser/extension_function_registry.h"
+#include "extensions/browser/extension_host.h"
+#include "extensions/browser/process_manager.h"
 #include "third_party/blink/public/mojom/speech/speech_synthesis.mojom.h"
 #include "ui/base/l10n/l10n_util.h"
 
@@ -270,8 +272,16 @@ ExtensionFunction::ResponseAction TtsSpeakFunction::Run() {
   // the behavior more predictable and easier to write unit tests for too.
   Respond(NoArguments());
 
-  std::unique_ptr<content::TtsUtterance> utterance =
-      content::TtsUtterance::Create(browser_context());
+  std::unique_ptr<content::TtsUtterance> utterance;
+  if (extension()) {
+    extensions::ExtensionHost* host =
+        extensions::ProcessManager::Get(browser_context())
+            ->GetBackgroundHostForExtension(extension()->id());
+    utterance = content::TtsUtterance::Create(host->host_contents());
+  } else {
+    utterance = content::TtsUtterance::Create(browser_context());
+  }
+
   utterance->SetText(text);
   utterance->SetVoiceName(voice_name);
   utterance->SetSrcId(src_id);

chrome/browser/speech/extension_api/tts_extension_apitest.cc

@@ -312,6 +312,10 @@ IN_PROC_BROWSER_TEST_F(TtsApiTest, PlatformSpeakOptionalArgs) {
       .WillOnce(Return(true));
   EXPECT_CALL(mock_platform_impl_, DoSpeak(_, "Echo", _, _, _))
       .WillOnce(Return());
+
+  // Called when the extension unloads.
+  EXPECT_CALL(mock_platform_impl_, StopSpeaking()).WillOnce(Return(false));
+
   ASSERT_TRUE(RunExtensionTest("tts/optional_args")) << message_;
 }
 

chrome/browser/speech/tts_controller_delegate_impl_unittest.cc

@@ -40,7 +40,7 @@ TEST(TtsControllerDelegateImplTest, TestTtsControllerUtteranceDefaults) {
   double volume = blink::mojom::kSpeechSynthesisDoublePrefNotSet;
 
   std::unique_ptr<content::TtsUtterance> utterance1 =
-      content::TtsUtterance::Create(nullptr);
+      content::TtsUtterance::Create();
   tts_controller_delegate->UpdateUtteranceDefaultsFromPrefs(
       utterance1.get(), &rate, &pitch, &volume);
   // Updated to global defaults.
@@ -59,7 +59,7 @@ TEST(TtsControllerDelegateImplTest, TestTtsControllerUtteranceDefaults) {
   tts_controller_delegate->pref_service_ = &pref_service_;
 
   std::unique_ptr<content::TtsUtterance> utterance2 =
-      content::TtsUtterance::Create(nullptr);
+      content::TtsUtterance::Create();
   tts_controller_delegate->UpdateUtteranceDefaultsFromPrefs(
       utterance2.get(), &rate, &pitch, &volume);
   // Updated to pref values.
@@ -73,7 +73,7 @@ TEST(TtsControllerDelegateImplTest, TestTtsControllerUtteranceDefaults) {
   volume = 1.3f;
 
   std::unique_ptr<content::TtsUtterance> utterance3 =
-      content::TtsUtterance::Create(nullptr);
+      content::TtsUtterance::Create();
   tts_controller_delegate->UpdateUtteranceDefaultsFromPrefs(
       utterance3.get(), &rate, &pitch, &volume);
   // Updated to pref values.
@@ -85,7 +85,7 @@ TEST(TtsControllerDelegateImplTest, TestTtsControllerUtteranceDefaults) {
 TEST(TtsControllerDelegateImplTest, GetPreferredVoiceIdsForUtterance) {
   MockTtsControllerDelegate delegate;
   std::unique_ptr<content::TtsUtterance> utterance =
-      content::TtsUtterance::Create(nullptr);
+      content::TtsUtterance::Create();
   auto ids = delegate.GetPreferredVoiceIdsForUtterance(utterance.get());
   EXPECT_EQ(nullptr, ids.get());
 

chrome/browser/ui/webui/settings/chromeos/tts_handler.cc

@@ -141,7 +141,7 @@ void TtsHandler::HandlePreviewTtsVoice(const base::ListValue* args) {
   json->GetString("extension", &extension_id);
 
   std::unique_ptr<content::TtsUtterance> utterance =
-      content::TtsUtterance::Create((Profile::FromWebUI(web_ui())));
+      content::TtsUtterance::Create(web_ui()->GetWebContents());
   utterance->SetText(text);
   utterance->SetVoiceName(name);
   utterance->SetEngineId(extension_id);

content/browser/speech/tts_controller_unittest.cc

@@ -231,12 +231,12 @@ class TtsControllerTest : public testing::Test {
 };
 
 TEST_F(TtsControllerTest, TestTtsControllerShutdown) {
-  std::unique_ptr<TtsUtterance> utterance1 = TtsUtterance::Create(nullptr);
+  std::unique_ptr<TtsUtterance> utterance1 = TtsUtterance::Create();
   utterance1->SetCanEnqueue(true);
   utterance1->SetSrcId(1);
   controller()->SpeakOrEnqueue(std::move(utterance1));
 
-  std::unique_ptr<TtsUtterance> utterance2 = TtsUtterance::Create(nullptr);
+  std::unique_ptr<TtsUtterance> utterance2 = TtsUtterance::Create();
   utterance2->SetCanEnqueue(true);
   utterance2->SetSrcId(2);
   controller()->SpeakOrEnqueue(std::move(utterance2));
@@ -286,8 +286,7 @@ TEST_F(TtsControllerTest, TestBrowserContextRemoved) {
 }
 #else
 TEST_F(TtsControllerTest, TestTtsControllerUtteranceDefaults) {
-  std::unique_ptr<TtsUtterance> utterance1 =
-      content::TtsUtterance::Create(nullptr);
+  std::unique_ptr<TtsUtterance> utterance1 = content::TtsUtterance::Create();
   // Initialized to default (unset constant) values.
   EXPECT_EQ(blink::mojom::kSpeechSynthesisDoublePrefNotSet,
             utterance1->GetContinuousParameters().rate);
@@ -312,7 +311,7 @@ TEST_F(TtsControllerTest, TestGetMatchingVoice) {
 
   {
     // Calling GetMatchingVoice with no voices returns -1.
-    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create(nullptr));
+    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create());
     std::vector<VoiceData> voices;
     EXPECT_EQ(-1, controller()->GetMatchingVoice(utterance.get(), voices));
   }
@@ -320,7 +319,7 @@ TEST_F(TtsControllerTest, TestGetMatchingVoice) {
   {
     // Calling GetMatchingVoice with any voices returns the first one
     // even if there are no criteria that match.
-    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create(nullptr));
+    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create());
     std::vector<VoiceData> voices(2);
     EXPECT_EQ(0, controller()->GetMatchingVoice(utterance.get(), voices));
   }
@@ -328,7 +327,7 @@ TEST_F(TtsControllerTest, TestGetMatchingVoice) {
   {
     // If nothing else matches, the English voice is returned.
     // (In tests the language will always be English.)
-    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create(nullptr));
+    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create());
     std::vector<VoiceData> voices;
     VoiceData fr_voice;
     fr_voice.lang = "fr";
@@ -379,7 +378,7 @@ TEST_F(TtsControllerTest, TestGetMatchingVoice) {
     voice8.native = true;
     voices.push_back(voice8);
 
-    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create(nullptr));
+    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create());
     EXPECT_EQ(0, controller()->GetMatchingVoice(utterance.get(), voices));
 
     std::set<TtsEventType> types;
@@ -443,7 +442,7 @@ TEST_F(TtsControllerTest, TestGetMatchingVoice) {
     voice1.name = "voice1";
     voice1.lang = "en-US";
     voices.push_back(voice1);
-    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create(nullptr));
+    std::unique_ptr<TtsUtterance> utterance(TtsUtterance::Create());
 
     // voice1 is matched against the exact default system language.
     TestContentBrowserClient::GetInstance()->set_application_locale("en-US");

content/browser/speech/tts_ssml_browsertest.cc

@@ -31,7 +31,7 @@ class TtsSsmlBrowserTest : public ContentBrowserTest {
     std::unique_ptr<MockTtsControllerImpl> controller =
         std::make_unique<MockTtsControllerImpl>();
 
-    std::unique_ptr<TtsUtterance> utterance = TtsUtterance::Create(nullptr);
+    std::unique_ptr<TtsUtterance> utterance = TtsUtterance::Create();
     utterance->SetText(input);
 
     base::RunLoop run_loop;

content/browser/speech/tts_utterance_impl.cc

@@ -5,6 +5,7 @@
 #include "content/browser/speech/tts_utterance_impl.h"
 
 #include "base/values.h"
+#include "content/public/browser/web_contents.h"
 #include "third_party/blink/public/mojom/speech/speech_synthesis.mojom.h"
 
 namespace content {
@@ -36,11 +37,25 @@ UtteranceContinuousParameters::UtteranceContinuousParameters()
 // static
 int TtsUtteranceImpl::next_utterance_id_ = 0;
 
+// static
+std::unique_ptr<TtsUtterance> TtsUtterance::Create(WebContents* web_contents) {
+  DCHECK(web_contents);
+  return std::make_unique<TtsUtteranceImpl>(web_contents->GetBrowserContext(),
+                                            web_contents);
+}
+
+// static
 std::unique_ptr<TtsUtterance> TtsUtterance::Create(
     BrowserContext* browser_context) {
+  DCHECK(browser_context);
   return std::make_unique<TtsUtteranceImpl>(browser_context, nullptr);
 }
 
+// static
+std::unique_ptr<TtsUtterance> TtsUtterance::Create() {
+  return std::make_unique<TtsUtteranceImpl>(nullptr, nullptr);
+}
+
 TtsUtteranceImpl::TtsUtteranceImpl(BrowserContext* browser_context,
                                    WebContents* web_contents)
     : WebContentsObserver(web_contents),

content/public/browser/tts_utterance.h

@@ -18,6 +18,7 @@ class Value;
 namespace content {
 class BrowserContext;
 class TtsUtterance;
+class WebContents;
 
 // Events sent back from the TTS engine indicating the progress.
 enum TtsEventType {
@@ -59,10 +60,15 @@ class CONTENT_EXPORT UtteranceEventDelegate {
 // One speech utterance.
 class CONTENT_EXPORT TtsUtterance {
  public:
-  // Construct an utterance given a profile and a completion task to call
-  // when the utterance is done speaking. Before speaking this utterance,
-  // its other parameters like text, rate, pitch, etc. should all be set.
+  // Construct an utterance given a WebContents, BrowserContext, or no backing
+  // context. Prefer the more specific WebContents variant if possible. The
+  // utterance's speaking lifetime is tied to their lifetime.
+  // Before speaking this utterance, its other parameters like text, rate,
+  // pitch, etc. should all be set.
+  static std::unique_ptr<TtsUtterance> Create(WebContents* web_contents);
   static std::unique_ptr<TtsUtterance> Create(BrowserContext* browser_context);
+  static std::unique_ptr<TtsUtterance> Create();
+
   virtual ~TtsUtterance() = default;
 
   // Sends an event to the delegate. If the event type is TTS_EVENT_END