SpeechInputExtensionManager now interface (exclusively) with...

SpeechInputExtensionManager now interface (exclusively) with SpeechRecognitionManagerDelegate (Speech CL1.11).

- The tray icon and balloon handing has been moved from speech_input_extensions_manager to chrome_speech_recognition_manager_delegate, since that code (tray icon) will be used also for continuous recognition.
- Removed the SpeechRecognizer interface from /content/public (thus de-virtualized and refcounted the SpeechRecognizerImpl)

BUG=116954
TEST=none

Review URL: https://chromiumcodereview.appspot.com/10377082

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@138498 0039d316-1c4b-4281-b951-d872f2087c98

chrome/browser/speech/chrome_speech_recognition_manager_delegate.h

@@ -13,6 +13,8 @@
 #include "content/public/browser/speech_recognition_manager_delegate.h"
 #include "content/public/browser/speech_recognition_session_config.h"
 
+class SpeechRecognitionTrayIconController;
+
 namespace speech {
 
 // This is Chrome's implementation of the SpeechRecognitionManagerDelegate
@@ -69,7 +71,12 @@ class ChromeSpeechRecognitionManagerDelegate
   // (which is copied into |last_session_config_|). Used for "try again".
   void RestartLastSession();
 
+  // Lazy initializers for bubble and tray icon controller.
+  SpeechRecognitionBubbleController* GetBubbleController();
+  SpeechRecognitionTrayIconController* GetTrayIconController();
+
   scoped_refptr<SpeechRecognitionBubbleController> bubble_controller_;
+  scoped_refptr<SpeechRecognitionTrayIconController> tray_icon_controller_;
   scoped_refptr<OptionalRequestInfo> optional_request_info_;
   scoped_ptr<content::SpeechRecognitionSessionConfig> last_session_config_;
 

chrome/browser/speech/speech_input_extension_apitest.cc

@@ -23,6 +23,10 @@ namespace net {
 class URLRequestContextGetter;
 }
 
+namespace {
+const int kSessionIDForTests = 0;
+}
+
 // Mock class used to test the extension speech input API.
 class SpeechInputExtensionApiTest : public ExtensionApiTest,
                                     public SpeechInputExtensionInterface {
@@ -77,10 +81,11 @@ class SpeechInputExtensionApiTest : public ExtensionApiTest,
   virtual void StartRecording(
       content::SpeechRecognitionEventListener* listener,
       net::URLRequestContextGetter* context_getter,
-      int session_id,
+      const string16& extension_name,
       const std::string& language,
       const std::string& grammar,
-      bool filter_profanities) OVERRIDE;
+      bool filter_profanities,
+      bool show_notification) OVERRIDE;
 
   virtual void StopRecording(bool recognition_failed) OVERRIDE;
 
@@ -107,7 +112,7 @@ class SpeechInputExtensionApiTest : public ExtensionApiTest,
   };
 
  private:
-  void ProvideResults(int session_id);
+  void ProvideResults();
 
   bool recording_devices_available_;
   bool recognizer_is_valid_;
@@ -131,10 +136,11 @@ SpeechInputExtensionApiTest::~SpeechInputExtensionApiTest() {
 void SpeechInputExtensionApiTest::StartRecording(
       content::SpeechRecognitionEventListener* listener,
       net::URLRequestContextGetter* context_getter,
-      int session_id,
+      const string16& extension_name,
       const std::string& language,
       const std::string& grammar,
-      bool filter_profanities) {
+      bool filter_profanities,
+      bool show_notification) {
   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
   recognizer_is_valid_ = true;
 
@@ -143,7 +149,7 @@ void SpeechInputExtensionApiTest::StartRecording(
       FROM_HERE,
       base::Bind(&SpeechInputExtensionManager::OnAudioStart,
                  GetManager(),
-                 session_id),
+                 kSessionIDForTests),
       base::TimeDelta());
 
   // Notify sound start in the input device.
@@ -151,16 +157,14 @@ void SpeechInputExtensionApiTest::StartRecording(
       FROM_HERE,
       base::Bind(&SpeechInputExtensionManager::OnSoundStart,
                  GetManager(),
-                 session_id),
+                 kSessionIDForTests),
       base::TimeDelta());
 
   if (result_delay_ms_ != kDontDispatchCall) {
     // Dispatch the recognition results.
     MessageLoop::current()->PostDelayedTask(
         FROM_HERE,
-        base::Bind(&SpeechInputExtensionApiTest::ProvideResults,
-                   this,
-                   session_id),
+        base::Bind(&SpeechInputExtensionApiTest::ProvideResults, this),
         base::TimeDelta::FromMilliseconds(result_delay_ms_));
   }
 }
@@ -170,18 +174,19 @@ void SpeechInputExtensionApiTest::StopRecording(bool recognition_failed) {
   recognizer_is_valid_ = false;
 }
 
-void SpeechInputExtensionApiTest::ProvideResults(int session_id) {
+void SpeechInputExtensionApiTest::ProvideResults() {
   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 
   if (next_error_ != content::SPEECH_RECOGNITION_ERROR_NONE) {
     GetManager()->OnRecognitionError(
-        session_id, content::SpeechRecognitionError(next_error_));
+        kSessionIDForTests, content::SpeechRecognitionError(next_error_));
     return;
   }
 
-  GetManager()->OnSoundEnd(session_id);
-  GetManager()->OnAudioEnd(session_id);
-  GetManager()->OnRecognitionResult(session_id, next_result_);
+  GetManager()->OnSoundEnd(kSessionIDForTests);
+  GetManager()->OnAudioEnd(kSessionIDForTests);
+  GetManager()->OnRecognitionResult(kSessionIDForTests, next_result_);
+  GetManager()->OnRecognitionEnd(kSessionIDForTests);
 }
 
 // Every test should leave the manager in the idle state when finished.

chrome/browser/speech/speech_input_extension_manager.h

@@ -11,18 +11,18 @@
 #include "base/callback_forward.h"
 #include "base/memory/ref_counted.h"
 #include "base/memory/scoped_ptr.h"
+#include "base/string16.h"
 #include "base/synchronization/lock.h"
 #include "content/public/browser/notification_observer.h"
 #include "content/public/browser/speech_recognition_event_listener.h"
 
 class Profile;
-class SpeechRecognitionTrayIconController;
 
 namespace content {
 class NotificationRegistrar;
 struct SpeechRecognitionError;
+class SpeechRecognitionManager;
 struct SpeechRecognitionResult;
-class SpeechRecognizer;
 }
 
 namespace net {
@@ -39,10 +39,11 @@ class SpeechInputExtensionInterface {
   virtual void StartRecording(
       content::SpeechRecognitionEventListener* listener,
       net::URLRequestContextGetter* context_getter,
-      int session_id,
+      const string16& extension_name,
       const std::string& language,
       const std::string& grammar,
-      bool filter_profanities) = 0;
+      bool filter_profanities,
+      bool show_notification) = 0;
 
   virtual void StopRecording(bool recognition_failed) = 0;
   virtual bool HasAudioInputDevices() = 0;
@@ -143,19 +144,22 @@ class SpeechInputExtensionManager
   virtual void StartRecording(
       content::SpeechRecognitionEventListener* listener,
       net::URLRequestContextGetter* context_getter,
-      int session_id,
+      const string16& extension_name,
       const std::string& language,
       const std::string& grammar,
-      bool filter_profanities) OVERRIDE;
+      bool filter_profanities,
+      bool show_notification) OVERRIDE;
 
   virtual void StopRecording(bool recognition_failed) OVERRIDE;
 
   // Internal methods.
   void StartOnIOThread(
-      net::URLRequestContextGetter* context_getter,
+      scoped_refptr<net::URLRequestContextGetter> context_getter,
+      const string16& extension_name,
       const std::string& language,
       const std::string& grammar,
-      bool filter_profanities);
+      bool filter_profanities,
+      bool show_notification);
   void ForceStopOnIOThread();
   void IsRecordingOnIOThread(const IsRecordingCallback& callback);
 
@@ -172,7 +176,6 @@ class SpeechInputExtensionManager
                                 const std::string& json_args);
   void ExtensionUnloaded(const std::string& extension_id);
 
-  void SetInputVolumeOnUIThread(float volume);
   void ResetToIdleState();
 
   virtual ~SpeechInputExtensionManager();
@@ -196,10 +199,10 @@ class SpeechInputExtensionManager
   // Used in the UI thread.
   scoped_ptr<content::NotificationRegistrar> registrar_;
   SpeechInputExtensionInterface* speech_interface_;
-  scoped_refptr<SpeechRecognitionTrayIconController> notification_;
 
   // Used in the IO thread.
-  scoped_refptr<content::SpeechRecognizer> recognizer_;
+  bool is_recognition_in_progress_;
+  int speech_recognition_session_id_;
 };
 
 #endif  // CHROME_BROWSER_SPEECH_SPEECH_INPUT_EXTENSION_MANAGER_H_

content/browser/speech/speech_recognition_manager_impl.cc

@@ -327,6 +327,12 @@ void SpeechRecognitionManagerImpl::AbortAllSessionsForListener(
 void SpeechRecognitionManagerImpl::DispatchEvent(int session_id,
                                                  FSMEvent event) {
   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
+
+  // There are some corner cases in which the session might be deleted (due to
+  // an EndRecognition event) between a request (e.g. Abort) and its dispatch.
+  if (!SessionExists(session_id))
+    return;
+
   const Session& session = GetSession(session_id);
   FSMState session_state = GetSessionState(session_id);
   DCHECK_LE(session_state, SESSION_STATE_MAX_VALUE);

content/browser/speech/speech_recognizer_impl.cc

@@ -12,7 +12,6 @@
 #include "content/browser/speech/google_one_shot_remote_engine.h"
 #include "content/public/browser/browser_thread.h"
 #include "content/public/browser/speech_recognition_event_listener.h"
-#include "content/public/browser/speech_recognizer.h"
 #include "content/public/common/speech_recognition_error.h"
 #include "content/public/common/speech_recognition_grammar.h"
 #include "content/public/common/speech_recognition_result.h"
@@ -24,7 +23,6 @@ using content::SpeechRecognitionError;
 using content::SpeechRecognitionEventListener;
 using content::SpeechRecognitionGrammar;
 using content::SpeechRecognitionResult;
-using content::SpeechRecognizer;
 using media::AudioInputController;
 using media::AudioManager;
 using media::AudioParameters;
@@ -68,41 +66,6 @@ void KeepAudioControllerRefcountedForDtor(scoped_refptr<AudioInputController>) {
 
 }  // namespace
 
-// TODO(primiano) Create(...) is transitional (until we fix speech input
-// extensions) and should be removed soon. The manager should be the only one
-// knowing the existence of SpeechRecognizer(Impl), thus the only one in charge
-// of instantiating it.
-SpeechRecognizer* SpeechRecognizer::Create(
-    SpeechRecognitionEventListener* listener,
-    int session_id,
-    const std::string& language,
-    const std::string& grammar,
-    net::URLRequestContextGetter* context_getter,
-    bool filter_profanities,
-    const std::string& hardware_info,
-    const std::string& origin_url) {
-  speech::SpeechRecognitionEngineConfig remote_engine_config;
-  remote_engine_config.language = language;
-  if (!grammar.empty())
-    remote_engine_config.grammars.push_back(SpeechRecognitionGrammar(grammar));
-  remote_engine_config.audio_sample_rate =
-      speech::SpeechRecognizerImpl::kAudioSampleRate;
-  remote_engine_config.audio_num_bits_per_sample =
-      speech::SpeechRecognizerImpl::kNumBitsPerAudioSample;
-  remote_engine_config.filter_profanities = filter_profanities;
-  remote_engine_config.hardware_info = hardware_info;
-  remote_engine_config.origin_url = origin_url;
-
-  // SpeechRecognizerImpl takes ownership of google_remote_engine.
-  speech::GoogleOneShotRemoteEngine* google_remote_engine =
-      new speech::GoogleOneShotRemoteEngine(context_getter);
-  google_remote_engine->SetConfig(remote_engine_config);
-
-  return new speech::SpeechRecognizerImpl(listener,
-                                          session_id,
-                                          google_remote_engine);
-}
-
 namespace speech {
 
 const int SpeechRecognizerImpl::kAudioSampleRate = 16000;

content/browser/speech/speech_recognizer_impl.h

@@ -11,7 +11,6 @@
 #include "base/memory/scoped_ptr.h"
 #include "content/browser/speech/endpointer/endpointer.h"
 #include "content/browser/speech/speech_recognition_engine.h"
-#include "content/public/browser/speech_recognizer.h"
 #include "content/public/common/speech_recognition_error.h"
 #include "content/public/common/speech_recognition_result.h"
 #include "media/audio/audio_input_controller.h"
@@ -27,15 +26,13 @@ class AudioManager;
 
 namespace speech {
 
-// TODO(primiano) Next CL: Remove the Impl suffix and the exported
-// /content/public/browser/speech_recognizer.h interface since this class should
-// not be visible outside (currently we need it for speech input extension API).
+// TODO(primiano) Next CL: Remove the Impl suffix.
 
 // Handles speech recognition for a session (identified by |session_id|), taking
 // care of audio capture, silence detection/endpointer and interaction with the
 // SpeechRecognitionEngine.
 class CONTENT_EXPORT SpeechRecognizerImpl
-    : public NON_EXPORTED_BASE(content::SpeechRecognizer),
+    : public base::RefCountedThreadSafe<SpeechRecognizerImpl>,
       public media::AudioInputController::EventHandler,
       public NON_EXPORTED_BASE(SpeechRecognitionEngineDelegate) {
  public:
@@ -50,18 +47,15 @@ class CONTENT_EXPORT SpeechRecognizerImpl
       int session_id,
       SpeechRecognitionEngine* engine);
 
-  // content::SpeechRecognizer methods.
-  virtual void StartRecognition() OVERRIDE;
-  virtual void AbortRecognition() OVERRIDE;
-  virtual void StopAudioCapture() OVERRIDE;
-  virtual bool IsActive() const OVERRIDE;
-  virtual bool IsCapturingAudio() const OVERRIDE;
+  void StartRecognition();
+  void AbortRecognition();
+  void StopAudioCapture();
+  bool IsActive() const;
+  bool IsCapturingAudio() const;
   const SpeechRecognitionEngine& recognition_engine() const;
 
- protected:
-  virtual ~SpeechRecognizerImpl();
-
  private:
+  friend class base::RefCountedThreadSafe<SpeechRecognizerImpl>;
   friend class SpeechRecognizerImplTest;
 
   enum FSMState {
@@ -96,6 +90,8 @@ class CONTENT_EXPORT SpeechRecognizerImpl
     content::SpeechRecognitionError engine_error;
   };
 
+  virtual ~SpeechRecognizerImpl();
+
   // Entry point for pushing any new external event into the recognizer FSM.
   void DispatchEvent(const FSMEventArgs& event_args);
 

content/content_browser.gypi

@@ -139,8 +139,8 @@
     'public/browser/speech_recognition_preferences.h',
     'public/browser/speech_recognition_session_config.cc',
     'public/browser/speech_recognition_session_config.h',
+    'public/browser/speech_recognition_session_context.cc',
     'public/browser/speech_recognition_session_context.h',
-    'public/browser/speech_recognizer.h',
     'public/browser/trace_controller.h',
     'public/browser/trace_subscriber.h',
     'public/browser/user_metrics.h',

content/public/browser/speech_recognition_session_context.cc

+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "content/public/browser/speech_recognition_session_context.h"
+
+namespace content {
+
+SpeechRecognitionSessionContext::SpeechRecognitionSessionContext()
+    : render_process_id(0),
+      render_view_id(0),
+      render_request_id(0),
+      requested_by_page_element(true),
+      is_first_request_for_context(false) {
+}
+
+SpeechRecognitionSessionContext::~SpeechRecognitionSessionContext() {
+}
+
+}  // namespace content

content/public/browser/speech_recognition_session_context.h

@@ -6,6 +6,7 @@
 #define CONTENT_PUBLIC_BROWSER_SPEECH_RECOGNITION_SESSION_CONTEXT_H_
 #pragma once
 
+#include "base/string16.h"
 #include "content/common/content_export.h"
 #include "ui/gfx/rect.h"
 
@@ -21,18 +22,32 @@ namespace content {
 // SpeechRecognitionManager::GetSessionContext and
 // SpeechRecognitionManager::LookupSessionByContext methods).
 struct CONTENT_EXPORT SpeechRecognitionSessionContext {
-  SpeechRecognitionSessionContext()
-      : render_process_id(0),
-        render_view_id(0),
-        render_request_id(0),
-        js_handle_id(0) {}
-  ~SpeechRecognitionSessionContext() {}
+  SpeechRecognitionSessionContext();
+  ~SpeechRecognitionSessionContext();
 
   int render_process_id;
   int render_view_id;
   int render_request_id;
   int js_handle_id;
+
+  // Determines whether recognition was requested by a page element (in which
+  // case its coordinates are passed in |element_rect|).
+  bool requested_by_page_element;
+
+  // The coordinates of the page element for placing the bubble (valid only when
+  // |requested_by_page_element| = true).
   gfx::Rect element_rect;
+
+  // Determines whether this is the first time that this context (identified by
+  // |context_name|) is requesting a recognition.
+  // TODO(primiano) This is really temporary, remove after CL1.12 which will
+  // refactor SpeechRecognitionPreferences and move this check entirely whithin
+  // chrome, without involving content.
+  bool is_first_request_for_context;
+
+  // A texual description of the context (website, extension name) that is
+  // requesting recognition, for prompting security notifications to the user.
+  string16 context_name;
 };
 
 }  // namespace content

content/public/browser/speech_recognizer.h

-// Copyright (c) 2012 The Chromium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-#ifndef CONTENT_PUBLIC_BROWSER_SPEECH_RECOGNIZER_H_
-#define CONTENT_PUBLIC_BROWSER_SPEECH_RECOGNIZER_H_
-#pragma once
-
-#include <string>
-
-#include "base/memory/ref_counted.h"
-#include "content/common/content_export.h"
-
-namespace net {
-class URLRequestContextGetter;
-}
-
-namespace content {
-
-class SpeechRecognitionEventListener;
-
-// Records audio, sends recorded audio to server and translates server response
-// to recognition result.
-// TODO(primiano) remove this public interface as soon as we fix speech input
-// extensions.
-class SpeechRecognizer : public base::RefCountedThreadSafe<SpeechRecognizer> {
- public:
-  CONTENT_EXPORT static SpeechRecognizer* Create(
-      SpeechRecognitionEventListener* event_listener,
-      int session_id,
-      const std::string& language,
-      const std::string& grammar,
-      net::URLRequestContextGetter* context_getter,
-      bool filter_profanities,
-      const std::string& hardware_info,
-      const std::string& origin_url);
-
-  // Starts audio recording and the recognition process. The same
-  // SpeechRecognizer instance can be used multiple times for speech recognition
-  // though each recognition request can be made only after the previous one
-  // completes (i.e. after receiving
-  // SpeechRecognitionEventListener::OnRecognitionEnd).
-  virtual void StartRecognition() = 0;
-
-  // Stops recording audio and cancels recognition. Any audio recorded so far
-  // gets discarded.
-  virtual void AbortRecognition() = 0;
-
-  // Stops recording audio and finalizes recognition, possibly getting results.
-  virtual void StopAudioCapture() = 0;
-
-  // Checks wether the recognizer is active, that is, either capturing audio
-  // or waiting for a result.
-  virtual bool IsActive() const = 0;
-
-  // Checks whether the recognizer is capturing audio.
-  virtual bool IsCapturingAudio() const = 0;
-
- protected:
-  friend class base::RefCountedThreadSafe<SpeechRecognizer>;
-  virtual ~SpeechRecognizer() {}
-};
-
-}  // namespace content
-
-#endif  // CONTENT_PUBLIC_BROWSER_SPEECH_RECOGNIZER_H_