1 files changed, 73 insertions, 3 deletions

diff --git a/source/frontend/StarVoice.hpp b/source/frontend/StarVoice.hpp
index 269adb4..e7ecd80 100644
--- a/source/frontend/StarVoice.hpp
+++ b/source/frontend/StarVoice.hpp
@@ -6,8 +6,11 @@
 #include "StarGameTypes.hpp"
 #include "StarMaybe.hpp"
 #include "StarThread.hpp"
+#include "StarDataStreamDevices.hpp"
 #include "StarApplicationController.hpp"
 
+#include <queue>
+
 struct OpusDecoder;
 typedef std::unique_ptr<OpusDecoder, void(*)(OpusDecoder*)> OpusDecoderPtr;
 struct OpusEncoder;
@@ -27,6 +30,36 @@ STAR_CLASS(Voice);
 STAR_CLASS(VoiceAudioStream);
 STAR_CLASS(ApplicationController);
 
+struct VoiceAudioChunk {
+  std::unique_ptr<int16_t[]> data;
+  size_t remaining;
+  size_t offset = 0;
+
+  VoiceAudioChunk(int16_t* ptr, size_t size) {
+    data.reset(ptr);
+    remaining = size;
+    offset = 0;
+  }
+
+  inline size_t takeSamples(std::vector<int16_t>& out, size_t count) {
+    size_t toRead = std::min<size_t>(count, remaining);
+    int16_t* start = data.get() + offset;
+    out.insert(out.end(), start, start + toRead);
+    offset += toRead;
+    remaining -= toRead;
+    return toRead;
+  }
+
+  //this one's unsafe
+  inline int16_t takeSample() {
+    --remaining;
+    return *(data.get() + offset++);
+  }
+
+  inline bool exhausted() { return remaining == 0; }
+};
+
+
 class Voice {
 public:
   // Individual speakers are represented by their connection ID.
@@ -45,6 +78,13 @@ public:
     VoiceAudioStreamPtr audioStream;
     Mutex mutex;
 
+    atomic<bool> muted = false;
+    atomic<bool> playing = false;
+    atomic<float> decibelLevel = 0.0f;
+    atomic<Array<float, 2>> channelVolumes = Array<float, 2>::filled(1.0f);
+
+    unsigned int minimumPlaySamples = 4096;
+
     Speaker(SpeakerId speakerId);
   };
 
@@ -77,7 +117,7 @@ public:
   SpeakerPtr speaker(SpeakerId speakerId);
 
   // Called when receiving input audio data from SDL, on its own thread.
-  void getAudioData(uint8_t* stream, int len);
+  void readAudioData(uint8_t* stream, int len);
 
   // Called to mix voice audio with the game.
   void mix(int16_t* buffer, size_t frames, unsigned channels);
@@ -87,6 +127,12 @@ public:
 
   void setDeviceName(Maybe<String> device);
 
+  int send(DataStreamBuffer& out, size_t budget);
+  bool receive(SpeakerPtr speaker, std::string_view view);
+
+  // Must be called every frame with input state, expires after 1s.
+  void setInput(bool input = true);
+
   inline int encoderChannels() const {
     return m_channelMode == VoiceChannelMode::Mono ? 1 : 2;
   }
@@ -99,10 +145,13 @@ private:
   void openDevice();
   void closeDevice();
 
+  bool playSpeaker(SpeakerPtr const& speaker, int channels);
+
   SpeakerId m_speakerId = 0;
   SpeakerPtr m_clientSpeaker;
   HashMap<SpeakerId, SpeakerPtr> m_speakers;
 
+  Mutex m_activeSpeakersMutex;
   HashSet<SpeakerPtr> m_activeSpeakers;
 
   OpusEncoderPtr m_encoder;
@@ -110,10 +159,15 @@ private:
   float m_outputVolume = 1.0f;
   float m_inputVolume = 1.0f;
   float m_threshold = -50.0f;
-
+  
+  int64_t m_lastSentTime = 0;
+  int64_t m_lastInputTime = 0;
+  int64_t m_lastThresholdTime = 0;
+  int64_t m_nextSaveTime = 0;
   bool m_enabled = true;
   bool m_inputEnabled = true;
 
+  int m_deviceChannels = 1;
   bool m_deviceOpen = false;
   Maybe<String> m_deviceName;
   VoiceInputMode m_inputMode;
@@ -121,7 +175,23 @@ private:
 
   ApplicationControllerPtr m_applicationController;
 
-  double nextSaveTime = 0.0f;
+  struct EncodedChunk {
+    std::unique_ptr<unsigned char[]> data;
+    size_t size;
+
+    EncodedChunk(unsigned char* _data, size_t len) {
+      data.reset(_data);
+      size = len;
+    }
+  };
+
+  std::vector<ByteArray> m_encodedChunks;
+  size_t m_encodedChunksLength = 0;
+
+  std::queue<VoiceAudioChunk> m_capturedChunks;
+  size_t m_capturedChunksFrames = 0;
+
+  Mutex m_captureMutex;
 };
   
 }