video-pc/AvRecorder/capturer/audio/audio_capturer.cpp

#include "audio_capturer.h"

#include "basic/basic.h"

#define DEFAULT_SAMPLE_RATE 48000        // 默认采样率:48kHz
#define DEFAULT_BITS_PER_SAMPLE 16       // 默认位深:16bit
#define DEFAULT_CHANNELS 1               // 默认音频通道数:1
#define DEFAULT_AUDIO_PACKET_INTERVAL 10 // 默认音频包发送间隔:10ms

bool AudioCapturer::Init(Type deviceType)
{
    Stop();
    _deviceType = deviceType;
    if (!_CreateDeviceEnumerator(&_pDeviceEnumerator)) { __DebugPrint("_CreateDeviceEnumerator failed"); return false; }
    if (!_CreateDevice(_pDeviceEnumerator, &_pDevice)) { __DebugPrint("_CreateDevice failed"); return false; }
    if (!_CreateAudioClient(_pDevice, &_pAudioClient)) { __DebugPrint("_CreateAudioClient failed"); return false; }

    if (!_IsFormatSupported(_pAudioClient)) {
        if (!_GetPreferFormat(_pAudioClient, &_formatex)) { __DebugPrint("_GetPreferFormat failed"); return false; }
    }
    if (!_InitAudioClient(_pAudioClient, &_formatex)) { __DebugPrint("_InitAudioClient failed"); return false; }
    if (!_CreateAudioCaptureClient(_pAudioClient, &_pAudioCaptureClient)) { __DebugPrint("_CreateAudioCaptureClient failed"); return false; }

    _format.sampleRate = _formatex.Format.nSamplesPerSec;
    _format.channels = _formatex.Format.nChannels;
    _format.bitsPerSample = _formatex.Format.wBitsPerSample;
    _format.avgBytesPerSec = _formatex.Format.nAvgBytesPerSec;
    _format.blockAlign = _formatex.Format.nBlockAlign;

    _isInit = true;
    return true;
}

bool AudioCapturer::Start()
{
    if (!_isInit) { return false; }
    
    // 如果是麦克风设备，启动静音播放器确保音频引擎活跃
    if (_deviceType == Microphone) {
        if (!_InitializeSilencePlayer()) {
            // 静音播放器初始化失败不影响主要功能，只记录日志
            // qDebug() << "Failed to initialize silence player for microphone";
        }
    }
    
    _loopFlag = true;
    // PlaySoundA("./rc/mute.wav", nullptr, SND_FILENAME | SND_ASYNC | SND_LOOP);
    _captureThread = new std::thread(
        [this] { _ThreadRun(_pAudioClient, _pAudioCaptureClient); });
    return true;
}

void AudioCapturer::Stop()
{
    // CoUninitialize();
    _isInit = false;
    _loopFlag = false;
    
    // 清理静音播放器
    _CleanupSilencePlayer();
    
    Free(_captureThread, [this] {
        _captureThread->join();
        delete _captureThread;
    });
    Free(_pAudioCaptureClient, [this] { _pAudioCaptureClient->Release(); });
    if (_pAudioClient != nullptr) {
        _pAudioClient->Stop();
    }
    //PlaySoundA(nullptr, nullptr, SND_FILENAME | SND_ASYNC | SND_LOOP);

    Free(_pAudioClient, [this] { _pAudioClient->Release(); });
    Free(_pDevice, [this] { _pDevice->Release(); });
    Free(_pDeviceEnumerator, [this] { _pDeviceEnumerator->Release(); });
}

bool AudioCapturer::_CreateDeviceEnumerator(IMMDeviceEnumerator** enumerator)
{
    // __CheckBool(SUCCEEDED(CoInitializeEx(nullptr, COINIT_MULTITHREADED)));
    // __CheckBool(SUCCEEDED(CoInitializeEx(nullptr, COINIT_APARTMENTTHREADED)));
    HRESULT hr = CoCreateInstance(__uuidof(MMDeviceEnumerator), NULL, CLSCTX_ALL,
        __uuidof(IMMDeviceEnumerator),
        reinterpret_cast<void**>(enumerator));
    if (FAILED(hr)) { __DebugPrint("CoCreateInstance for MMDeviceEnumerator failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    return true;
}
bool AudioCapturer::_CreateDevice(IMMDeviceEnumerator* enumerator, IMMDevice** device)
{
    EDataFlow enDataFlow = _deviceType == Microphone ? eCapture : eRender;
    ERole enRole = eConsole;
    HRESULT hr = enumerator->GetDefaultAudioEndpoint(enDataFlow, enRole, device);
    if (FAILED(hr)) { __DebugPrint("GetDefaultAudioEndpoint failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    return true;
}
bool AudioCapturer::_CreateAudioClient(IMMDevice* device, IAudioClient** audioClient)
{
    HRESULT hr = device->Activate(__uuidof(IAudioClient), CLSCTX_ALL, NULL,
        (void**)audioClient);
    if (FAILED(hr)) { __DebugPrint("IAudioClient Activate failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    return true;
}
bool AudioCapturer::_IsFormatSupported(IAudioClient* audioClient)
{
    memset(&_formatex, 0, sizeof(_formatex));
    WAVEFORMATEX* format = &_formatex.Format;
    format->nSamplesPerSec = DEFAULT_SAMPLE_RATE;
    format->wBitsPerSample = DEFAULT_BITS_PER_SAMPLE;
    format->nChannels = DEFAULT_CHANNELS;

    WAVEFORMATEX* closestMatch = nullptr;

    HRESULT hr = audioClient->IsFormatSupported(AUDCLNT_SHAREMODE_SHARED,
        format, &closestMatch);
    if (hr == AUDCLNT_E_UNSUPPORTED_FORMAT) // 0x88890008
    {
        if (closestMatch == nullptr) // 如果找不到最相近的格式，closestMatch可能为nullptr
        {
            return false;
        }

        format->nSamplesPerSec = closestMatch->nSamplesPerSec;
        format->wBitsPerSample = closestMatch->wBitsPerSample;
        format->nChannels = closestMatch->nChannels;

        return true;
    }

    return false;
}
bool AudioCapturer::_GetPreferFormat(IAudioClient* audioClient,
    WAVEFORMATEXTENSIBLE* formatex)
{
    WAVEFORMATEX* format = nullptr;
    HRESULT hr = audioClient->GetMixFormat(&format);
    if (FAILED(hr)) { __DebugPrint("GetMixFormat failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    formatex->Format.nSamplesPerSec = format->nSamplesPerSec;
    formatex->Format.wBitsPerSample = format->wBitsPerSample;
    formatex->Format.nChannels = format->nChannels;
    CoTaskMemFree(format);
    return true;
}
bool AudioCapturer::_InitAudioClient(IAudioClient* audioClient,
    WAVEFORMATEXTENSIBLE* formatex)
{
    AUDCLNT_SHAREMODE shareMode = AUDCLNT_SHAREMODE_SHARED; // share Audio Engine with other applications
    DWORD streamFlags = _deviceType == Microphone ? 0 : AUDCLNT_STREAMFLAGS_LOOPBACK;
    streamFlags |= AUDCLNT_STREAMFLAGS_AUTOCONVERTPCM;      // A channel matrixer and a sample
                                                            // rate converter are inserted
    streamFlags |= AUDCLNT_STREAMFLAGS_SRC_DEFAULT_QUALITY; // a sample rate converter
                                                            // with better quality than
                                                            // the default conversion but
                                                            // with a higher performance
                                                            // cost is used
    REFERENCE_TIME hnsBufferDuration = 0;
    WAVEFORMATEX* format = &formatex->Format;
    format->wFormatTag = WAVE_FORMAT_EXTENSIBLE;
    format->nBlockAlign = (format->wBitsPerSample >> 3) * format->nChannels;
    format->nAvgBytesPerSec = format->nBlockAlign * format->nSamplesPerSec;
    format->cbSize = sizeof(WAVEFORMATEXTENSIBLE) - sizeof(WAVEFORMATEX);
    formatex->Samples.wValidBitsPerSample = format->wBitsPerSample;
    formatex->dwChannelMask = format->nChannels == 1 ? KSAUDIO_SPEAKER_MONO : KSAUDIO_SPEAKER_STEREO;
    formatex->SubFormat = KSDATAFORMAT_SUBTYPE_PCM;

    HRESULT hr = audioClient->Initialize(shareMode, streamFlags, hnsBufferDuration, 0,
        format, nullptr);
    if (FAILED(hr)) { __DebugPrint("IAudioClient Initialize failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    return true;
}

bool AudioCapturer::_CreateAudioCaptureClient(IAudioClient* audioClient,
    IAudioCaptureClient** audioCaptureClient)
{
    HRESULT hr = audioClient->GetService(IID_PPV_ARGS(audioCaptureClient));
    if (FAILED(hr)) { __DebugPrint("GetService(IAudioCaptureClient) failed, hr=0x%08lx", static_cast<unsigned long>(hr)); return false; }
    return true;
}

bool AudioCapturer::_ThreadRun(IAudioClient* audio_client,
    IAudioCaptureClient* audio_capture_client)
{
    // 初始化COM到MTA，确保当前线程可安全调用COM接口
    HRESULT hrCoInit = CoInitializeEx(nullptr, COINIT_MULTITHREADED);
    if (FAILED(hrCoInit)) {
        __DebugPrint("CoInitializeEx failed, hr=0x%08lx", static_cast<unsigned long>(hrCoInit));
        return false;
    }

    UINT32 num_success = 0;
    BYTE* p_audio_data = nullptr;
    UINT32 num_frames_to_read = 0;
    DWORD dw_flag = 0;
    UINT32 num_frames_in_next_packet = 0;
    audio_client->Start();
    while (_loopFlag) {
        SleepMs(5);
        while (true) {
            HRESULT hr = audio_capture_client->GetNextPacketSize(&num_frames_in_next_packet);
            if (FAILED(hr)) { __DebugPrint("GetNextPacketSize failed, hr=0x%08lx", static_cast<unsigned long>(hr)); audio_client->Stop(); CoUninitialize(); return false; }
            if (num_frames_in_next_packet == 0) {
                break;
            }

            hr = audio_capture_client->GetBuffer(&p_audio_data, &num_frames_to_read,
                &dw_flag, nullptr, nullptr);
            if (FAILED(hr)) { __DebugPrint("GetBuffer failed, hr=0x%08lx", static_cast<unsigned long>(hr)); audio_client->Stop(); CoUninitialize(); return false; }

            size_t size = (_formatex.Format.wBitsPerSample >> 3) * _formatex.Format.nChannels * num_frames_to_read;
            {
                std::lock_guard<std::mutex> lock(_bufferMutex);
                size_t oldSize = _buffer.size();
                _buffer.resize(oldSize + size);
                memcpy(_buffer.data() + oldSize, p_audio_data, size);
            }
            hr = audio_capture_client->ReleaseBuffer(num_frames_to_read);
            if (FAILED(hr)) { __DebugPrint("ReleaseBuffer failed, hr=0x%08lx", static_cast<unsigned long>(hr)); audio_client->Stop(); CoUninitialize(); return false; }
        }
    }

    audio_client->Stop();
    CoUninitialize();
    return true;
}

// 静音播放器实现
bool AudioCapturer::_InitializeSilencePlayer()
{
    HRESULT hr;
    
    // 获取默认音频渲染设备
    hr = _pDeviceEnumerator->GetDefaultAudioEndpoint(eRender, eConsole, &_pSilenceDevice);
    if (FAILED(hr)) {
        return false;
    }
    
    // 激活音频客户端
    hr = _pSilenceDevice->Activate(__uuidof(IAudioClient), CLSCTX_ALL, nullptr, (void**)&_pSilenceAudioClient);
    if (FAILED(hr)) {
        return false;
    }
    
    // 获取混音格式
    WAVEFORMATEX* pSilenceFormat = nullptr;
    hr = _pSilenceAudioClient->GetMixFormat(&pSilenceFormat);
    if (FAILED(hr)) {
        return false;
    }
    
    // 初始化音频客户端（共享模式）
    hr = _pSilenceAudioClient->Initialize(
        AUDCLNT_SHAREMODE_SHARED,
        0,
        10000000,  // 1秒缓冲区
        0,
        pSilenceFormat,
        nullptr
    );
    
    CoTaskMemFree(pSilenceFormat);
    
    if (FAILED(hr)) {
        return false;
    }
    
    // 获取渲染客户端
    hr = _pSilenceAudioClient->GetService(__uuidof(IAudioRenderClient), (void**)&_pSilenceRenderClient);
    if (FAILED(hr)) {
        return false;
    }
    
    // 启动音频客户端
    hr = _pSilenceAudioClient->Start();
    if (FAILED(hr)) {
        return false;
    }
    
    // 启动静音播放线程
    _silencePlayerRunning = true;
    _silencePlayerThread = new std::thread([this] { _SilencePlayerThreadFunc(); });
    
    return true;
}

void AudioCapturer::_SilencePlayerThreadFunc()
{
    // 线程内初始化COM
    HRESULT hrCoInit = CoInitializeEx(nullptr, COINIT_MULTITHREADED);
    if (FAILED(hrCoInit)) {
        __DebugPrint("CoInitializeEx for silence thread failed, hr=0x%08lx", static_cast<unsigned long>(hrCoInit));
        return;
    }

    UINT32 bufferFrameCount;
    HRESULT hr = _pSilenceAudioClient->GetBufferSize(&bufferFrameCount);
    if (FAILED(hr)) {
        CoUninitialize();
        return;
    }

    // 计算每帧字节数（使用设备混音格式）
    WAVEFORMATEX* pMix = nullptr;
    UINT32 frameBytes = 0;
    hr = _pSilenceAudioClient->GetMixFormat(&pMix);
    if (SUCCEEDED(hr) && pMix) {
        frameBytes = pMix->nBlockAlign;
        CoTaskMemFree(pMix);
    } else {
        // 回退：猜测为立体声float（安全起见仍可为0填充任何格式）
        frameBytes = sizeof(float) * 2;
    }
    
    while (_silencePlayerRunning) {
        UINT32 numFramesPadding;
        hr = _pSilenceAudioClient->GetCurrentPadding(&numFramesPadding);
        if (FAILED(hr)) {
            break;
        }
        
        UINT32 numFramesAvailable = bufferFrameCount - numFramesPadding;
        
        if (numFramesAvailable > 0) {
            BYTE* pData;
            hr = _pSilenceRenderClient->GetBuffer(numFramesAvailable, &pData);
            if (SUCCEEDED(hr)) {
                // 填充静音数据（全零）
                memset(pData, 0, numFramesAvailable * frameBytes);
                _pSilenceRenderClient->ReleaseBuffer(numFramesAvailable, 0);
            }
        }
        
        Sleep(10); // 10ms 间隔
    }

    CoUninitialize();
}

int AudioCapturer::readAudioData(char* buf, int maxLen)
{
    std::lock_guard<std::mutex> lock(_bufferMutex);
    int toRead = std::min<int>(maxLen, _buffer.size());
    if (toRead > 0) {
        memcpy(buf, _buffer.data(), toRead);
        _buffer.erase(_buffer.begin(), _buffer.begin() + toRead);
    }
    return toRead;
}

void AudioCapturer::_CleanupSilencePlayer()
{
    // 停止静音播放线程
    if (_silencePlayerRunning) {
        _silencePlayerRunning = false;
        if (_silencePlayerThread && _silencePlayerThread->joinable()) {
            _silencePlayerThread->join();
            delete _silencePlayerThread;
            _silencePlayerThread = nullptr;
        }
    }
    
    // 清理 WASAPI 资源
    if (_pSilenceAudioClient) {
        _pSilenceAudioClient->Stop();
        _pSilenceAudioClient->Release();
        _pSilenceAudioClient = nullptr;
    }
    
    if (_pSilenceRenderClient) {
        _pSilenceRenderClient->Release();
        _pSilenceRenderClient = nullptr;
    }
    
    if (_pSilenceDevice) {
        _pSilenceDevice->Release();
        _pSilenceDevice = nullptr;
    }
}