Update the Speech Api to support array(s) of result items

content/browser/speech/google_streaming_remote_engine_unittest.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 <queue>
 
 #include "base/memory/scoped_ptr.h"
 #include "base/message_loop.h"
 #include "base/utf_string_conversions.h"
 #include "content/browser/speech/audio_buffer.h"
@@ skipping 20 matching lines @@
  public:
   GoogleStreamingRemoteEngineTest()
       : last_number_of_upstream_chunks_seen_(0U),
         error_(SPEECH_RECOGNITION_ERROR_NONE) { }
 
   // Creates a speech recognition request and invokes its URL fetcher delegate
   // with the given test data.
   void CreateAndTestRequest(bool success, const std::string& http_response);
 
   // SpeechRecognitionRequestDelegate methods.
-  virtual void OnSpeechRecognitionEngineResult(
-      const SpeechRecognitionResult& result) OVERRIDE {
-    results_.push(result);
+  virtual void OnSpeechRecognitionEngineResults(
+      const SpeechRecognitionResults& results) OVERRIDE {
+    results_.push(results);
   }
   virtual void OnSpeechRecognitionEngineError(
       const SpeechRecognitionError& error) OVERRIDE {
     error_ = error.code;
   }
 
   // testing::Test methods.
   virtual void SetUp() OVERRIDE;
   virtual void TearDown() OVERRIDE;
 
  protected:
   enum DownstreamError {
     DOWNSTREAM_ERROR_NONE,
     DOWNSTREAM_ERROR_HTTP500,
     DOWNSTREAM_ERROR_NETWORK,
     DOWNSTREAM_ERROR_WEBSERVICE_NO_MATCH
   };
-  static bool ResultsAreEqual(const SpeechRecognitionResult& a,
-                              const SpeechRecognitionResult& b);
+  static bool ResultsAreEqual(const SpeechRecognitionResults& a,
+                              const SpeechRecognitionResults& b);
   static std::string SerializeProtobufResponse(
       const proto::SpeechRecognitionEvent& msg);
   static std::string ToBigEndian32(uint32 value);
 
   TestURLFetcher* GetUpstreamFetcher();
   TestURLFetcher* GetDownstreamFetcher();
   void StartMockRecognition();
   void EndMockRecognition();
   void InjectDummyAudioChunk();
   size_t UpstreamChunksUploadedFromLastCall();
   void ProvideMockProtoResultDownstream(
       const proto::SpeechRecognitionEvent& result);
   void ProvideMockResultDownstream(const SpeechRecognitionResult& result);
-  void ExpectResultReceived(const SpeechRecognitionResult& result);
+  void ExpectResultsReceived(const SpeechRecognitionResults& result);
   void CloseMockDownstream(DownstreamError error);
 
   scoped_ptr<GoogleStreamingRemoteEngine> engine_under_test_;
   TestURLFetcherFactory url_fetcher_factory_;
   size_t last_number_of_upstream_chunks_seen_;
   MessageLoop message_loop_;
   std::string response_buffer_;
   SpeechRecognitionErrorCode error_;
-  std::queue<SpeechRecognitionResult> results_;
+  std::queue<SpeechRecognitionResults> results_;
 };
 
 TEST_F(GoogleStreamingRemoteEngineTest, SingleDefinitiveResult) {
   StartMockRecognition();
   ASSERT_TRUE(GetUpstreamFetcher());
   ASSERT_EQ(0U, UpstreamChunksUploadedFromLastCall());
 
   // Inject some dummy audio chunks and check a corresponding chunked upload
   // is performed every time on the server.
   for (int i = 0; i < 3; ++i) {
     InjectDummyAudioChunk();
     ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
   }
 
   // Ensure that a final (empty) audio chunk is uploaded on chunks end.
   engine_under_test_->AudioChunksEnded();
   ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Simulate a protobuf message streamed from the server containing a single
   // result with two hypotheses.
-  SpeechRecognitionResult result;
+  SpeechRecognitionResults results;
+  results.push_back(SpeechRecognitionResult());
+  SpeechRecognitionResult& result = results.back();
   result.is_provisional = false;
   result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("hypothesis 1"), 0.1F));
   result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("hypothesis 2"), 0.2F));
 
   ProvideMockResultDownstream(result);
-  ExpectResultReceived(result);
+  ExpectResultsReceived(results);
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Ensure everything is closed cleanly after the downstream is closed.
   CloseMockDownstream(DOWNSTREAM_ERROR_NONE);
   ASSERT_FALSE(engine_under_test_->IsRecognitionPending());
   EndMockRecognition();
   ASSERT_EQ(SPEECH_RECOGNITION_ERROR_NONE, error_);
   ASSERT_EQ(0U, results_.size());
 }
 
 TEST_F(GoogleStreamingRemoteEngineTest, SeveralStreamingResults) {
   StartMockRecognition();
   ASSERT_TRUE(GetUpstreamFetcher());
   ASSERT_EQ(0U, UpstreamChunksUploadedFromLastCall());
 
   for (int i = 0; i < 4; ++i) {
     InjectDummyAudioChunk();
     ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
 
-    SpeechRecognitionResult result;
+    SpeechRecognitionResults results;
+    results.push_back(SpeechRecognitionResult());
+    SpeechRecognitionResult& result = results.back();
     result.is_provisional = (i % 2 == 0);  // Alternate result types.
     float confidence = result.is_provisional ? 0.0F : (i * 0.1F);
     result.hypotheses.push_back(
         SpeechRecognitionHypothesis(UTF8ToUTF16("hypothesis"), confidence));
 
     ProvideMockResultDownstream(result);
-    ExpectResultReceived(result);
+    ExpectResultsReceived(results);
     ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
   }
 
   // Ensure that a final (empty) audio chunk is uploaded on chunks end.
   engine_under_test_->AudioChunksEnded();
   ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Simulate a final definitive result.
-  SpeechRecognitionResult result;
+  SpeechRecognitionResults results;
+  results.push_back(SpeechRecognitionResult());
+  SpeechRecognitionResult& result = results.back();
   result.is_provisional = false;
   result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("The final result"), 1.0F));
   ProvideMockResultDownstream(result);
-  ExpectResultReceived(result);
+  ExpectResultsReceived(results);
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Ensure everything is closed cleanly after the downstream is closed.
   CloseMockDownstream(DOWNSTREAM_ERROR_NONE);
   ASSERT_FALSE(engine_under_test_->IsRecognitionPending());
   EndMockRecognition();
   ASSERT_EQ(SPEECH_RECOGNITION_ERROR_NONE, error_);
   ASSERT_EQ(0U, results_.size());
 }
 
 TEST_F(GoogleStreamingRemoteEngineTest, NoFinalResultAfterAudioChunksEnded) {
   StartMockRecognition();
   ASSERT_TRUE(GetUpstreamFetcher());
   ASSERT_EQ(0U, UpstreamChunksUploadedFromLastCall());
 
   // Simulate one pushed audio chunk.
   InjectDummyAudioChunk();
   ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
 
   // Simulate the corresponding definitive result.
-  SpeechRecognitionResult result;
+  SpeechRecognitionResults results;
+  results.push_back(SpeechRecognitionResult());
+  SpeechRecognitionResult& result = results.back();
   result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("hypothesis"), 1.0F));
   ProvideMockResultDownstream(result);
-  ExpectResultReceived(result);
+  ExpectResultsReceived(results);
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Simulate a silent downstream closure after |AudioChunksEnded|.
   engine_under_test_->AudioChunksEnded();
   ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
   CloseMockDownstream(DOWNSTREAM_ERROR_NONE);
 
   // Expect an empty result, aimed at notifying recognition ended with no
   // actual results nor errors.
-  SpeechRecognitionResult empty_result;
-  ExpectResultReceived(empty_result);
+  SpeechRecognitionResults empty_results;
+  ExpectResultsReceived(empty_results);
 
   // Ensure everything is closed cleanly after the downstream is closed.
   ASSERT_FALSE(engine_under_test_->IsRecognitionPending());
   EndMockRecognition();
   ASSERT_EQ(SPEECH_RECOGNITION_ERROR_NONE, error_);
   ASSERT_EQ(0U, results_.size());
 }
 
 TEST_F(GoogleStreamingRemoteEngineTest, NoMatchError) {
   StartMockRecognition();
   ASSERT_TRUE(GetUpstreamFetcher());
   ASSERT_EQ(0U, UpstreamChunksUploadedFromLastCall());
 
   for (int i = 0; i < 3; ++i)
     InjectDummyAudioChunk();
   engine_under_test_->AudioChunksEnded();
   ASSERT_EQ(4U, UpstreamChunksUploadedFromLastCall());
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Simulate only a provisional result.
-  SpeechRecognitionResult result;
+  SpeechRecognitionResults results;
+  results.push_back(SpeechRecognitionResult());
+  SpeechRecognitionResult& result = results.back();
   result.is_provisional = true;
   result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("The final result"), 0.0F));
   ProvideMockResultDownstream(result);
-  ExpectResultReceived(result);
+  ExpectResultsReceived(results);
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   CloseMockDownstream(DOWNSTREAM_ERROR_WEBSERVICE_NO_MATCH);
 
   // Expect an empty result.
   ASSERT_FALSE(engine_under_test_->IsRecognitionPending());
   EndMockRecognition();
-  SpeechRecognitionResult empty_result;
-  ExpectResultReceived(empty_result);
+  SpeechRecognitionResults empty_result;
+  ExpectResultsReceived(empty_result);
 }
 
 TEST_F(GoogleStreamingRemoteEngineTest, HTTPError) {
   StartMockRecognition();
   ASSERT_TRUE(GetUpstreamFetcher());
   ASSERT_EQ(0U, UpstreamChunksUploadedFromLastCall());
 
   InjectDummyAudioChunk();
   ASSERT_EQ(1U, UpstreamChunksUploadedFromLastCall());
 
@@ skipping 40 matching lines @@
   proto::SpeechRecognitionEvent proto_event;
   proto_event.set_status(proto::SpeechRecognitionEvent::STATUS_SUCCESS);
   proto::SpeechRecognitionResult* proto_result = proto_event.add_result();
   proto_result->set_stability(0.5);
   proto::SpeechRecognitionAlternative *proto_alternative =
       proto_result->add_alternative();
   proto_alternative->set_transcript("foo");
   ProvideMockProtoResultDownstream(proto_event);
 
   // Set up expectations.
-  SpeechRecognitionResult expected;
-  expected.is_provisional = true;
-  expected.hypotheses.push_back(
+  SpeechRecognitionResults results;
+  results.push_back(SpeechRecognitionResult());
+  SpeechRecognitionResult& result = results.back();
+  result.is_provisional = true;
+  result.hypotheses.push_back(
       SpeechRecognitionHypothesis(UTF8ToUTF16("foo"), 0.5));
 
   // Check that the protobuf generated the expected result.
-  ExpectResultReceived(expected);
+  ExpectResultsReceived(results);
 
   // Since it was a provisional result, recognition is still pending.
   ASSERT_TRUE(engine_under_test_->IsRecognitionPending());
 
   // Shut down.
   CloseMockDownstream(DOWNSTREAM_ERROR_NONE);
   ASSERT_FALSE(engine_under_test_->IsRecognitionPending());
   EndMockRecognition();
 
   // Since there was no final result, we get an empty "no match" result.
-  SpeechRecognitionResult empty_result;
-  ExpectResultReceived(empty_result);
+  SpeechRecognitionResults empty_result;
+  ExpectResultsReceived(empty_result);
   ASSERT_EQ(SPEECH_RECOGNITION_ERROR_NONE, error_);
   ASSERT_EQ(0U, results_.size());
 }
 
 void GoogleStreamingRemoteEngineTest::SetUp() {
   engine_under_test_.reset(
       new  GoogleStreamingRemoteEngine(NULL /*URLRequestContextGetter*/));
   engine_under_test_->set_delegate(this);
 }
 
@@ skipping 110 matching lines @@
 
   if (error == DOWNSTREAM_ERROR_WEBSERVICE_NO_MATCH) {
     // Send empty response.
     proto::SpeechRecognitionEvent response;
     response_buffer_.append(SerializeProtobufResponse(response));
   }
   downstream_fetcher->SetResponseString(response_buffer_);
   downstream_fetcher->delegate()->OnURLFetchComplete(downstream_fetcher);
 }
 
-void GoogleStreamingRemoteEngineTest::ExpectResultReceived(
-    const SpeechRecognitionResult& result) {
+void GoogleStreamingRemoteEngineTest::ExpectResultsReceived(
+    const SpeechRecognitionResults& results) {
   ASSERT_GE(1U, results_.size());
-  ASSERT_TRUE(ResultsAreEqual(result, results_.front()));
+  ASSERT_TRUE(ResultsAreEqual(results, results_.front()));
   results_.pop();
 }
 
 bool GoogleStreamingRemoteEngineTest::ResultsAreEqual(
-    const SpeechRecognitionResult& a, const SpeechRecognitionResult& b) {
-  if (a.is_provisional != b.is_provisional ||
-      a.hypotheses.size() != b.hypotheses.size()) {
+    const SpeechRecognitionResults& a, const SpeechRecognitionResults& b) {
+  if (a.size() != b.size())
     return false;
-  }
-  for (size_t i = 0; i < a.hypotheses.size(); ++i) {
-    const SpeechRecognitionHypothesis& hyp_a = a.hypotheses[i];
-    const SpeechRecognitionHypothesis& hyp_b = b.hypotheses[i];
-    if (hyp_a.utterance != hyp_b.utterance ||
-        hyp_a.confidence != hyp_b.confidence) {
+
+  SpeechRecognitionResults::const_iterator it_a = a.begin();
+  SpeechRecognitionResults::const_iterator it_b = b.begin();
+  for (; it_a != a.end() && it_b != b.end(); ++it_a, ++it_b) {
+    if ((*it_a).is_provisional != (*it_b).is_provisional ||

[hans, 2012/11/28 14:15:35]

nit: instead if (*it_a).is_provisional, I'd have gone for it_a->is_provisional

It might even be worth doing
const SpeechRecognitionResult& res_a = *it_a;

to simplify the code below a bit.

[tommi (sloooow) - chröme, 2012/11/29 09:30:20]

Done.  I just changed (*foo). to foo-> since I figured that |it_a->| would be as
readable as |res_a.|

+        (*it_a).hypotheses.size() != (*it_b).hypotheses.size()) {
       return false;
     }
+    for (size_t i = 0; i < (*it_a).hypotheses.size(); ++i) {
+      const SpeechRecognitionHypothesis& hyp_a = (*it_a).hypotheses[i];
+      const SpeechRecognitionHypothesis& hyp_b = (*it_b).hypotheses[i];
+      if (hyp_a.utterance != hyp_b.utterance ||
+          hyp_a.confidence != hyp_b.confidence) {
+        return false;
+      }
+    }
   }
+
   return true;
 }
 
 std::string GoogleStreamingRemoteEngineTest::SerializeProtobufResponse(
     const proto::SpeechRecognitionEvent& msg) {
   std::string msg_string;
   msg.SerializeToString(&msg_string);
 
   // Prepend 4 byte prefix length indication to the protobuf message as
   // envisaged by the google streaming recognition webservice protocol.
   msg_string.insert(0, ToBigEndian32(msg_string.size()));
   return msg_string;
 }
 
 std::string GoogleStreamingRemoteEngineTest::ToBigEndian32(uint32 value) {
   char raw_data[4];
   raw_data[0] = static_cast<uint8>((value >> 24) & 0xFF);
   raw_data[1] = static_cast<uint8>((value >> 16) & 0xFF);
   raw_data[2] = static_cast<uint8>((value >> 8) & 0xFF);
   raw_data[3] = static_cast<uint8>(value & 0xFF);
   return std::string(raw_data, sizeof(raw_data));
 }
 
 }  // namespace content