1 /*
2 * Copyright (C) 2013 Apple Inc. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
6 * are met:
7 * 1. Redistributions of source code must retain the above copyright
8 * notice, this list of conditions and the following disclaimer.
9 * 2. Redistributions in binary form must reproduce the above copyright
10 * notice, this list of conditions and the following disclaimer in the
11 * documentation and/or other materials provided with the distribution.
12 *
13 * THIS SOFTWARE IS PROVIDED BY APPLE COMPUTER, INC. ``AS IS'' AND ANY
14 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
15 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
16 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE COMPUTER, INC. OR
17 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
18 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
19 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
20 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
21 * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
23 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24 */
25
26 #include "third_party/blink/renderer/modules/speech/speech_synthesis.h"
27
28 #include "build/build_config.h"
29 #include "third_party/blink/public/common/browser_interface_broker_proxy.h"
30 #include "third_party/blink/public/common/thread_safe_browser_interface_broker_proxy.h"
31 #include "third_party/blink/public/platform/platform.h"
32 #include "third_party/blink/renderer/bindings/modules/v8/v8_speech_synthesis_error_event_init.h"
33 #include "third_party/blink/renderer/bindings/modules/v8/v8_speech_synthesis_event_init.h"
34 #include "third_party/blink/renderer/core/dom/document.h"
35 #include "third_party/blink/renderer/core/execution_context/execution_context.h"
36 #include "third_party/blink/renderer/core/frame/deprecation.h"
37 #include "third_party/blink/renderer/core/html/media/autoplay_policy.h"
38 #include "third_party/blink/renderer/core/timing/dom_window_performance.h"
39 #include "third_party/blink/renderer/core/timing/performance.h"
40 #include "third_party/blink/renderer/modules/speech/speech_synthesis_error_event.h"
41 #include "third_party/blink/renderer/modules/speech/speech_synthesis_event.h"
42 #include "third_party/blink/renderer/platform/instrumentation/use_counter.h"
43
44 namespace blink {
45
Create(ExecutionContext * context)46 SpeechSynthesis* SpeechSynthesis::Create(ExecutionContext* context) {
47 SpeechSynthesis* synthesis = MakeGarbageCollected<SpeechSynthesis>(context);
48 #if defined(OS_ANDROID)
49 // On Android devices we lazily initialize |mojom_synthesis_| to avoid
50 // needlessly binding to the TTS service, see https://crbug.com/811929.
51 // TODO(crbug/811929): Consider moving this logic into the Android-
52 // specific backend implementation.
53 #else
54 synthesis->InitializeMojomSynthesis();
55 #endif
56 return synthesis;
57 }
58
CreateForTesting(ExecutionContext * context,mojo::PendingRemote<mojom::blink::SpeechSynthesis> mojom_synthesis)59 SpeechSynthesis* SpeechSynthesis::CreateForTesting(
60 ExecutionContext* context,
61 mojo::PendingRemote<mojom::blink::SpeechSynthesis> mojom_synthesis) {
62 SpeechSynthesis* synthesis = MakeGarbageCollected<SpeechSynthesis>(context);
63 synthesis->SetMojomSynthesisForTesting(std::move(mojom_synthesis));
64 return synthesis;
65 }
66
SpeechSynthesis(ExecutionContext * context)67 SpeechSynthesis::SpeechSynthesis(ExecutionContext* context)
68 : ExecutionContextClient(context),
69 receiver_(this, context),
70 mojom_synthesis_(context) {
71 DCHECK(!GetExecutionContext() || GetExecutionContext()->IsDocument());
72 }
73
OnSetVoiceList(Vector<mojom::blink::SpeechSynthesisVoicePtr> mojom_voices)74 void SpeechSynthesis::OnSetVoiceList(
75 Vector<mojom::blink::SpeechSynthesisVoicePtr> mojom_voices) {
76 voice_list_.clear();
77 for (auto& mojom_voice : mojom_voices) {
78 voice_list_.push_back(
79 MakeGarbageCollected<SpeechSynthesisVoice>(std::move(mojom_voice)));
80 }
81 VoicesDidChange();
82 }
83
getVoices()84 const HeapVector<Member<SpeechSynthesisVoice>>& SpeechSynthesis::getVoices() {
85 // Kick off initialization here to ensure voice list gets populated.
86 InitializeMojomSynthesisIfNeeded();
87 return voice_list_;
88 }
89
speaking() const90 bool SpeechSynthesis::speaking() const {
91 // If we have a current speech utterance, then that means we're assumed to be
92 // in a speaking state. This state is independent of whether the utterance
93 // happens to be paused.
94 return CurrentSpeechUtterance();
95 }
96
pending() const97 bool SpeechSynthesis::pending() const {
98 // This is true if there are any utterances that have not started.
99 // That means there will be more than one in the queue.
100 return utterance_queue_.size() > 1;
101 }
102
paused() const103 bool SpeechSynthesis::paused() const {
104 return is_paused_;
105 }
106
speak(SpeechSynthesisUtterance * utterance)107 void SpeechSynthesis::speak(SpeechSynthesisUtterance* utterance) {
108 DCHECK(utterance);
109 Document* document = Document::From(GetExecutionContext());
110 if (!document)
111 return;
112
113 // Note: Non-UseCounter based TTS metrics are of the form TextToSpeech.* and
114 // are generally global, whereas these are scoped to a single page load.
115 UseCounter::Count(document, WebFeature::kTextToSpeech_Speak);
116 document->CountUseOnlyInCrossOriginIframe(
117 WebFeature::kTextToSpeech_SpeakCrossOrigin);
118 if (!IsAllowedToStartByAutoplay()) {
119 Deprecation::CountDeprecation(
120 document, WebFeature::kTextToSpeech_SpeakDisallowedByAutoplay);
121 FireErrorEvent(utterance, 0 /* char_index */, "not-allowed");
122 return;
123 }
124
125 utterance_queue_.push_back(utterance);
126
127 // If the queue was empty, speak this immediately.
128 if (utterance_queue_.size() == 1)
129 StartSpeakingImmediately();
130 }
131
cancel()132 void SpeechSynthesis::cancel() {
133 // Remove all the items from the utterance queue. The platform
134 // may still have references to some of these utterances and may
135 // fire events on them asynchronously.
136 utterance_queue_.clear();
137
138 InitializeMojomSynthesisIfNeeded();
139 mojom_synthesis_->Cancel();
140 }
141
pause()142 void SpeechSynthesis::pause() {
143 if (is_paused_)
144 return;
145
146 InitializeMojomSynthesisIfNeeded();
147 mojom_synthesis_->Pause();
148 }
149
resume()150 void SpeechSynthesis::resume() {
151 if (!CurrentSpeechUtterance())
152 return;
153
154 InitializeMojomSynthesisIfNeeded();
155 mojom_synthesis_->Resume();
156 }
157
DidStartSpeaking(SpeechSynthesisUtterance * utterance)158 void SpeechSynthesis::DidStartSpeaking(SpeechSynthesisUtterance* utterance) {
159 FireEvent(event_type_names::kStart, utterance, 0, 0, String());
160 }
161
DidPauseSpeaking(SpeechSynthesisUtterance * utterance)162 void SpeechSynthesis::DidPauseSpeaking(SpeechSynthesisUtterance* utterance) {
163 is_paused_ = true;
164 FireEvent(event_type_names::kPause, utterance, 0, 0, String());
165 }
166
DidResumeSpeaking(SpeechSynthesisUtterance * utterance)167 void SpeechSynthesis::DidResumeSpeaking(SpeechSynthesisUtterance* utterance) {
168 is_paused_ = false;
169 FireEvent(event_type_names::kResume, utterance, 0, 0, String());
170 }
171
DidFinishSpeaking(SpeechSynthesisUtterance * utterance)172 void SpeechSynthesis::DidFinishSpeaking(SpeechSynthesisUtterance* utterance) {
173 HandleSpeakingCompleted(utterance, false);
174 }
175
SpeakingErrorOccurred(SpeechSynthesisUtterance * utterance)176 void SpeechSynthesis::SpeakingErrorOccurred(
177 SpeechSynthesisUtterance* utterance) {
178 HandleSpeakingCompleted(utterance, true);
179 }
180
WordBoundaryEventOccurred(SpeechSynthesisUtterance * utterance,unsigned char_index,unsigned char_length)181 void SpeechSynthesis::WordBoundaryEventOccurred(
182 SpeechSynthesisUtterance* utterance,
183 unsigned char_index,
184 unsigned char_length) {
185 DEFINE_STATIC_LOCAL(const String, word_boundary_string, ("word"));
186 FireEvent(event_type_names::kBoundary, utterance, char_index, char_length,
187 word_boundary_string);
188 }
189
SentenceBoundaryEventOccurred(SpeechSynthesisUtterance * utterance,unsigned char_index,unsigned char_length)190 void SpeechSynthesis::SentenceBoundaryEventOccurred(
191 SpeechSynthesisUtterance* utterance,
192 unsigned char_index,
193 unsigned char_length) {
194 DEFINE_STATIC_LOCAL(const String, sentence_boundary_string, ("sentence"));
195 FireEvent(event_type_names::kBoundary, utterance, char_index, char_length,
196 sentence_boundary_string);
197 }
198
VoicesDidChange()199 void SpeechSynthesis::VoicesDidChange() {
200 if (GetExecutionContext())
201 DispatchEvent(*Event::Create(event_type_names::kVoiceschanged));
202 }
203
StartSpeakingImmediately()204 void SpeechSynthesis::StartSpeakingImmediately() {
205 SpeechSynthesisUtterance* utterance = CurrentSpeechUtterance();
206 DCHECK(utterance);
207
208 double millis;
209 if (!GetElapsedTimeMillis(&millis))
210 return;
211
212 utterance->SetStartTime(millis / 1000.0);
213 is_paused_ = false;
214
215 InitializeMojomSynthesisIfNeeded();
216 utterance->Start(this);
217 }
218
HandleSpeakingCompleted(SpeechSynthesisUtterance * utterance,bool error_occurred)219 void SpeechSynthesis::HandleSpeakingCompleted(
220 SpeechSynthesisUtterance* utterance,
221 bool error_occurred) {
222 DCHECK(utterance);
223
224 bool should_start_speaking = false;
225 // If the utterance that completed was the one we're currently speaking,
226 // remove it from the queue and start speaking the next one.
227 if (utterance == CurrentSpeechUtterance()) {
228 utterance_queue_.pop_front();
229 should_start_speaking = !utterance_queue_.empty();
230 }
231
232 // Always fire the event, because the platform may have asynchronously
233 // sent an event on an utterance before it got the message that we
234 // canceled it, and we should always report to the user what actually
235 // happened.
236 if (error_occurred) {
237 // TODO(csharrison): Actually pass the correct message. For now just use a
238 // generic error.
239 FireErrorEvent(utterance, 0, "synthesis-failed");
240 } else {
241 FireEvent(event_type_names::kEnd, utterance, 0, 0, String());
242 }
243
244 // Start the next utterance if we just finished one and one was pending.
245 if (should_start_speaking && !utterance_queue_.IsEmpty())
246 StartSpeakingImmediately();
247 }
248
FireEvent(const AtomicString & type,SpeechSynthesisUtterance * utterance,uint32_t char_index,uint32_t char_length,const String & name)249 void SpeechSynthesis::FireEvent(const AtomicString& type,
250 SpeechSynthesisUtterance* utterance,
251 uint32_t char_index,
252 uint32_t char_length,
253 const String& name) {
254 double millis;
255 if (!GetElapsedTimeMillis(&millis))
256 return;
257
258 SpeechSynthesisEventInit* init = SpeechSynthesisEventInit::Create();
259 init->setUtterance(utterance);
260 init->setCharIndex(char_index);
261 init->setCharLength(char_length);
262 init->setElapsedTime(millis - (utterance->StartTime() * 1000.0));
263 init->setName(name);
264 utterance->DispatchEvent(*SpeechSynthesisEvent::Create(type, init));
265 }
266
FireErrorEvent(SpeechSynthesisUtterance * utterance,uint32_t char_index,const String & error)267 void SpeechSynthesis::FireErrorEvent(SpeechSynthesisUtterance* utterance,
268 uint32_t char_index,
269 const String& error) {
270 double millis;
271 if (!GetElapsedTimeMillis(&millis))
272 return;
273
274 SpeechSynthesisErrorEventInit* init = SpeechSynthesisErrorEventInit::Create();
275 init->setUtterance(utterance);
276 init->setCharIndex(char_index);
277 init->setElapsedTime(millis - (utterance->StartTime() * 1000.0));
278 init->setError(error);
279 utterance->DispatchEvent(
280 *SpeechSynthesisErrorEvent::Create(event_type_names::kError, init));
281 }
282
CurrentSpeechUtterance() const283 SpeechSynthesisUtterance* SpeechSynthesis::CurrentSpeechUtterance() const {
284 if (utterance_queue_.IsEmpty())
285 return nullptr;
286
287 return utterance_queue_.front();
288 }
289
Trace(Visitor * visitor)290 void SpeechSynthesis::Trace(Visitor* visitor) {
291 visitor->Trace(receiver_);
292 visitor->Trace(mojom_synthesis_);
293 visitor->Trace(voice_list_);
294 visitor->Trace(utterance_queue_);
295 ExecutionContextClient::Trace(visitor);
296 EventTargetWithInlineData::Trace(visitor);
297 }
298
GetElapsedTimeMillis(double * millis)299 bool SpeechSynthesis::GetElapsedTimeMillis(double* millis) {
300 if (!GetExecutionContext())
301 return false;
302 Document* delegate_document = Document::From(GetExecutionContext());
303 if (!delegate_document || delegate_document->IsStopped())
304 return false;
305 LocalDOMWindow* delegate_dom_window = delegate_document->domWindow();
306 if (!delegate_dom_window)
307 return false;
308
309 *millis = DOMWindowPerformance::performance(*delegate_dom_window)->now();
310 return true;
311 }
312
IsAllowedToStartByAutoplay() const313 bool SpeechSynthesis::IsAllowedToStartByAutoplay() const {
314 Document* document = Document::From(GetExecutionContext());
315 DCHECK(document);
316
317 // Note: could check the utterance->volume here, but that could be overriden
318 // in the case of SSML.
319 if (AutoplayPolicy::GetAutoplayPolicyForDocument(*document) !=
320 AutoplayPolicy::Type::kDocumentUserActivationRequired) {
321 return true;
322 }
323 return AutoplayPolicy::IsDocumentAllowedToPlay(*document);
324 }
325
SetMojomSynthesisForTesting(mojo::PendingRemote<mojom::blink::SpeechSynthesis> mojom_synthesis)326 void SpeechSynthesis::SetMojomSynthesisForTesting(
327 mojo::PendingRemote<mojom::blink::SpeechSynthesis> mojom_synthesis) {
328 mojom_synthesis_.Bind(
329 std::move(mojom_synthesis),
330 GetExecutionContext()->GetTaskRunner(TaskType::kMiscPlatformAPI));
331 receiver_.reset();
332 mojom_synthesis_->AddVoiceListObserver(receiver_.BindNewPipeAndPassRemote(
333 GetExecutionContext()->GetTaskRunner(TaskType::kMiscPlatformAPI)));
334 }
335
InitializeMojomSynthesis()336 void SpeechSynthesis::InitializeMojomSynthesis() {
337 DCHECK(!mojom_synthesis_.is_bound());
338
339 // The frame could be detached. In that case, calls on mojom_synthesis_ will
340 // just get dropped. That's okay and is simpler than having to null-check
341 // mojom_synthesis_ before each use.
342 ExecutionContext* context = GetExecutionContext();
343
344 if (!context)
345 return;
346
347 auto receiver = mojom_synthesis_.BindNewPipeAndPassReceiver(
348 context->GetTaskRunner(TaskType::kMiscPlatformAPI));
349
350 context->GetBrowserInterfaceBroker().GetInterface(std::move(receiver));
351
352 mojom_synthesis_->AddVoiceListObserver(receiver_.BindNewPipeAndPassRemote(
353 context->GetTaskRunner(TaskType::kMiscPlatformAPI)));
354 }
355
InitializeMojomSynthesisIfNeeded()356 void SpeechSynthesis::InitializeMojomSynthesisIfNeeded() {
357 if (!mojom_synthesis_.is_bound())
358 InitializeMojomSynthesis();
359 }
360
InterfaceName() const361 const AtomicString& SpeechSynthesis::InterfaceName() const {
362 return event_target_names::kSpeechSynthesis;
363 }
364
365 } // namespace blink
366