Update libtgvoip for latest version

libtgvoip/.gitignore

@@ -18,9 +18,33 @@ bld/
 [Bb]in/
 [Oo]bj/
 DerivedData/
+*.o
 
 # Visual Studio 2015 cache/options directory
 .vs/
 
 xcuserdata/
-autom4te.cache/
+
+# autoconf build files
+/Makefile
+/Makefile.in
+/aclocal.m4
+/autom4te.cache/
+/compile
+/config.*
+/configure
+/depcomp
+/install-sh
+/libtool
+/ltmain.sh
+/m4/
+/missing
+/stamp-h?
+.deps/
+.dirstamp
+.libs/
+*.l[ao]
+*~
+
+# generated pkg-config file
+/tgvoip.pc

libtgvoip/.gitmodules

@@ -0,0 +1,3 @@
+[submodule "cmake"]
+	path = cmake
+	url = https://github.com/desktop-app/cmake_helpers.git

libtgvoip/Android.mk

@@ -47,6 +47,7 @@ LOCAL_SRC_FILES := \
 ./audio/AudioIO.cpp \
 ./video/VideoRenderer.cpp \
 ./video/VideoSource.cpp \
+./video/ScreamCongestionController.cpp \
 ./os/android/VideoSourceAndroid.cpp \
 ./os/android/VideoRendererAndroid.cpp \
 ./client/android/tg_voip_jni.cpp

libtgvoip/BlockingQueue.h

@@ -12,10 +12,10 @@
 #include "threading.h"
 #include "utils.h"
 
-using namespace std;
-
 namespace tgvoip{
 
+using namespace std;
+
 template<typename T>
 class BlockingQueue{
 public:

libtgvoip/Buffers.cpp

@@ -130,6 +130,8 @@ void BufferInputStream::EnsureEnoughRemaining(size_t need){
 
 BufferOutputStream::BufferOutputStream(size_t size){
 	buffer=(unsigned char*) malloc(size);
+	if(!buffer)
+		throw std::bad_alloc();
 	offset=0;
 	this->size=size;
 	bufferProvided=false;
@@ -217,6 +219,8 @@ void BufferOutputStream::ExpandBufferIfNeeded(size_t need){
 			buffer=(unsigned char *) realloc(buffer, size+need);
 			size+=need;
 		}
+		if(!buffer)
+			throw std::bad_alloc();
 	}
 }
 

libtgvoip/Buffers.h

@@ -66,7 +66,7 @@ namespace tgvoip{
 		size_t GetLength();
 		void Reset();
 		void Rewind(size_t numBytes);
-		
+
 		BufferOutputStream& operator=(BufferOutputStream&& other){
 			if(this!=&other){
 				if(!bufferProvided && buffer)
@@ -265,6 +265,10 @@ namespace tgvoip{
 				_i=size+_i;
 			return data[_i];
 		}
+
+		size_t Size(){
+			return size;
+		}
 	private:
 		std::array<T, size> data;
 		ptrdiff_t offset=0;

libtgvoip/CMakeLists.txt

@@ -26,6 +26,8 @@ add_library(libtgvoip STATIC
         OpusEncoder.cpp
         OpusEncoder.h
         threading.h
+        TgVoip.cpp
+        TgVoip.h
         VoIPController.cpp
         VoIPGroupController.cpp
         VoIPController.h
@@ -46,6 +48,8 @@ add_library(libtgvoip STATIC
         MessageThread.h
         audio/AudioIO.cpp
         audio/AudioIO.h
+        video/ScreamCongestionController.cpp
+        video/ScreamCongestionController.h
         video/VideoSource.cpp
         video/VideoSource.h
         video/VideoRenderer.cpp
@@ -648,7 +652,8 @@ add_library(libtgvoip STATIC
         audio/SoftwareAudioOutput.h
         audio/SoftwareAudioOutput.cpp)
 
-set_property(TARGET libtgvoip PROPERTY CXX_STANDARD 11)
+# FIXME: last libtgvoip looks adapted to C++20
+set_property(TARGET libtgvoip PROPERTY CXX_STANDARD 14)
 
 target_include_directories(libtgvoip PRIVATE
         webrtc_dsp
@@ -666,8 +671,10 @@ target_compile_definitions(libtgvoip PRIVATE
 
 target_compile_definitions(libtgvoip PUBLIC
         # Chose one of NO_DSP / DESKTOP_DSP
-        # TGVOIP_NO_DSP
-        TGVOIP_USE_DESKTOP_DSP
+        # FIXME: desktop DSP cannot be used since last libtgvoip update
+        #  since libtgvoip codebase is tighten to external webrtc
+        TGVOIP_NO_DSP
+        # TGVOIP_USE_DESKTOP_DSP
         TGVOIP_USE_SOFTWARE_AUDIO)
 
 if (${spdlog_FOUND})

libtgvoip/EchoCanceller.cpp

@@ -5,8 +5,9 @@
 //
 
 #ifndef TGVOIP_NO_DSP
-#include "webrtc_dsp/modules/audio_processing/include/audio_processing.h"
-#include "webrtc_dsp/api/audio/audio_frame.h"
+#include "modules/audio_processing/include/audio_processing.h"
+#include "modules/audio_processing/include/audio_frame_proxies.h"
+#include "api/audio/audio_frame.h"
 #endif
 
 #include "EchoCanceller.h"
@@ -27,12 +28,14 @@ EchoCanceller::EchoCanceller(bool enableAEC, bool enableNS, bool enableAGC){
 	this->enableNS=enableNS;
 	isOn=true;
 
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
 	webrtc::Config extraConfig;
-#ifdef TGVOIP_USE_DESKTOP_DSP
 	extraConfig.Set(new webrtc::DelayAgnostic(true));
-#endif
 
 	apm=webrtc::AudioProcessingBuilder().Create(extraConfig);
+#else
+	apm=webrtc::AudioProcessingBuilder().Create();
+#endif
 
 	webrtc::AudioProcessing::Config config;
 	config.echo_canceller.enabled = enableAEC;
@@ -43,28 +46,35 @@ EchoCanceller::EchoCanceller(bool enableAEC, bool enableNS, bool enableAGC){
 #endif
 	config.high_pass_filter.enabled = enableAEC;
 	config.gain_controller2.enabled = enableAGC;
+
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
 	apm->ApplyConfig(config);
-
-	webrtc::NoiseSuppression::Level nsLevel;
+
+	using Level = webrtc::NoiseSuppression::Level;
+#else
+	using Level = webrtc::AudioProcessing::Config::NoiseSuppression::Level;
+#endif
+	Level nsLevel;
 #ifdef __APPLE__
 	switch(ServerConfig::GetSharedInstance()->GetInt("webrtc_ns_level_vpio", 0)){
 #else
 	switch(ServerConfig::GetSharedInstance()->GetInt("webrtc_ns_level", 2)){
 #endif
 		case 0:
-			nsLevel=webrtc::NoiseSuppression::Level::kLow;
+			nsLevel=Level::kLow;
 			break;
 		case 1:
-			nsLevel=webrtc::NoiseSuppression::Level::kModerate;
+			nsLevel=Level::kModerate;
 			break;
 		case 3:
-			nsLevel=webrtc::NoiseSuppression::Level::kVeryHigh;
+			nsLevel=Level::kVeryHigh;
 			break;
 		case 2:
 		default:
-			nsLevel=webrtc::NoiseSuppression::Level::kHigh;
+			nsLevel=Level::kHigh;
 			break;
 	}
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
 	apm->noise_suppression()->set_level(nsLevel);
 	apm->noise_suppression()->Enable(enableNS);
 	if(enableAGC){
@@ -74,6 +84,19 @@ EchoCanceller::EchoCanceller(bool enableAEC, bool enableNS, bool enableAGC){
 		apm->gain_control()->set_compression_gain_db(ServerConfig::GetSharedInstance()->GetInt("webrtc_agc_compression_gain", 20));
 	}
 	apm->voice_detection()->set_likelihood(webrtc::VoiceDetection::Likelihood::kVeryLowLikelihood);
+#else
+	config.noise_suppression.level = nsLevel;
+	config.noise_suppression.enabled = enableNS;
+	if(enableAGC){
+		config.gain_controller1.mode = webrtc::AudioProcessing::Config::GainController1::kAdaptiveDigital;
+		config.gain_controller1.target_level_dbfs = ServerConfig::GetSharedInstance()->GetInt("webrtc_agc_target_level", 9);
+		config.gain_controller1.enable_limiter = ServerConfig::GetSharedInstance()->GetBoolean("webrtc_agc_enable_limiter", true);
+		config.gain_controller1.compression_gain_db = ServerConfig::GetSharedInstance()->GetInt("webrtc_agc_compression_gain", 20);
+	}
+	config.voice_detection.enabled = true;
+
+	apm->ApplyConfig(config);
+#endif
 
 	audioFrame=new webrtc::AudioFrame();
 	audioFrame->samples_per_channel_=480;
@@ -94,8 +117,9 @@ EchoCanceller::EchoCanceller(bool enableAEC, bool enableNS, bool enableAGC){
 
 EchoCanceller::~EchoCanceller(){
 #ifndef TGVOIP_NO_DSP
-	delete apm;
+	apm = nullptr;
 	delete audioFrame;
+	delete farendBufferPool;
 #endif
 }
 
@@ -109,7 +133,7 @@ void EchoCanceller::Stop(){
 
 
 void EchoCanceller::SpeakerOutCallback(unsigned char* data, size_t len){
-    if(len!=960*2 || !enableAEC || !isOn)
+	if(len!=960*2 || !enableAEC || !isOn)
 		return;
 #ifndef TGVOIP_NO_DSP
 	int16_t* buf=(int16_t*)farendBufferPool->Get();
@@ -130,9 +154,9 @@ void EchoCanceller::RunBufferFarendThread(){
 		int16_t* samplesIn=farendQueue->GetBlocking();
 		if(samplesIn){
 			memcpy(frame.mutable_data(), samplesIn, 480*2);
-			apm->ProcessReverseStream(&frame);
+			webrtc::ProcessReverseAudioFrame(apm, &frame);
 			memcpy(frame.mutable_data(), samplesIn+480, 480*2);
-			apm->ProcessReverseStream(&frame);
+			webrtc::ProcessReverseAudioFrame(apm, &frame);
 			didBufferFarend=true;
 			farendBufferPool->Reuse(reinterpret_cast<unsigned char*>(samplesIn));
 		}
@@ -154,17 +178,25 @@ void EchoCanceller::ProcessInput(int16_t* inOut, size_t numSamples, bool& hasVoi
 
 	memcpy(audioFrame->mutable_data(), inOut, 480*2);
 	if(enableAEC)
-    	apm->set_stream_delay_ms(delay);
-	apm->ProcessStream(audioFrame);
+		apm->set_stream_delay_ms(delay);
+	webrtc::ProcessAudioFrame(apm, audioFrame);
 	if(enableVAD)
-    	hasVoice=apm->voice_detection()->stream_has_voice();
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
+		hasVoice=apm->voice_detection()->stream_has_voice();
+#else
+		hasVoice= apm->GetStatistics().voice_detected.value_or(false);
+#endif
 	memcpy(inOut, audioFrame->data(), 480*2);
 	memcpy(audioFrame->mutable_data(), inOut+480, 480*2);
 	if(enableAEC)
-    	apm->set_stream_delay_ms(delay);
-	apm->ProcessStream(audioFrame);
+		apm->set_stream_delay_ms(delay);
+	webrtc::ProcessAudioFrame(apm, audioFrame);
 	if(enableVAD){
-    	hasVoice=hasVoice || apm->voice_detection()->stream_has_voice();
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
+		hasVoice=hasVoice || apm->voice_detection()->stream_has_voice();
+#else
+		hasVoice=hasVoice || apm->GetStatistics().voice_detected.value_or(false);
+#endif
 	}
 	memcpy(inOut+480, audioFrame->data(), 480*2);
 #endif
@@ -186,7 +218,13 @@ void EchoCanceller::SetAECStrength(int strength){
 void EchoCanceller::SetVoiceDetectionEnabled(bool enabled){
 	enableVAD=enabled;
 #ifndef TGVOIP_NO_DSP
+#ifdef TGVOIP_USE_DESKTOP_DSP_BUNDLED
 	apm->voice_detection()->Enable(enabled);
+#else
+	auto config = apm->GetConfig();
+	config.voice_detection.enabled = enabled;
+	apm->ApplyConfig(config);
+#endif
 #endif
 }
 

libtgvoip/EchoCanceller.h

@@ -13,6 +13,10 @@
 #include "MediaStreamItf.h"
 #include "utils.h"
 
+#ifndef TGVOIP_NO_DSP
+#include "api/scoped_refptr.h"
+#endif // TGVOIP_NO_DSP
+
 namespace webrtc{
 	class AudioProcessing;
 	class AudioFrame;
@@ -40,7 +44,7 @@ class EchoCanceller{
 	bool enableVAD=false;
 	bool isOn;
 #ifndef TGVOIP_NO_DSP
-	webrtc::AudioProcessing* apm=NULL;
+	rtc::scoped_refptr<webrtc::AudioProcessing> apm;
 	webrtc::AudioFrame* audioFrame=NULL;
 	void RunBufferFarendThread();
 	bool didBufferFarend;

libtgvoip/JitterBuffer.cpp

@@ -16,7 +16,6 @@ JitterBuffer::JitterBuffer(MediaStreamItf *out, uint32_t step):bufferPool(JITTER
 	if(out)
 		out->SetCallback(JitterBuffer::CallbackOut, this);
 	this->step=step;
-	memset(slots, 0, sizeof(jitter_packet_t)*JITTER_SLOT_COUNT);
 	if(step<30){
 		minMinDelay=(uint32_t) ServerConfig::GetSharedInstance()->GetInt("jitter_min_delay_20", 6);
 		maxMinDelay=(uint32_t) ServerConfig::GetSharedInstance()->GetInt("jitter_max_delay_20", 25);
@@ -246,7 +245,7 @@ void JitterBuffer::PutInternal(jitter_packet_t* pkt, bool overwriteExisting){
 		first=true;
 		LOGI("jitter: resyncing, next timestamp = %lld (step=%d, minDelay=%f)", (long long int)nextTimestamp, step, minDelay);
 	}
-	
+
 	for(i=0;i<JITTER_SLOT_COUNT;i++){
 		if(slots[i].buffer!=NULL){
 			if(slots[i].timestamp<nextTimestamp-1){

libtgvoip/JitterBuffer.h

@@ -42,11 +42,11 @@ class JitterBuffer{
 
 private:
 	struct jitter_packet_t{
-		unsigned char* buffer=NULL;
-		size_t size;
-		uint32_t timestamp;
-		bool isEC;
-		double recvTimeDiff;
+		unsigned char* buffer = nullptr;
+		size_t size = 0;
+		uint32_t timestamp = 0;
+		bool isEC = false;
+		double recvTimeDiff = 0.;
 	};
 	static size_t CallbackIn(unsigned char* data, size_t len, void* param);
 	static size_t CallbackOut(unsigned char* data, size_t len, void* param);