Collidoscope

#pragma once

#include "cinder/Cinder.h"

#include "cinder/audio/Node.h"

#include "cinder/audio/SampleRecorderNode.h"

#include "cinder/audio/dsp/RingBuffer.h"

#include "cinder/Filesystem.h"

#include "Messages.h"

typedef std::shared_ptr<class BufferToWaveRecorderNode>        BufferToWaveRecorderNodeRef;

typedef ci::audio::dsp::RingBufferT<RecordWaveMsg> RecordWaveMsgRingBuffer;

/**

 * A \a Node in the audio graph of the Cinder audio library that records input in a buffer.

 *

 * This class is similar to \a cinder::audio::BufferRecorderNode (it's a derivative work of this class indeed) but it has an additional feature.

 * When recording it uses the audio input samples to compute the size values of the visual chunks. 

 * The chunks values are stored in a ring buffer and fetched by the graphic thread to paint the wave as it gets recorded.

 *

 */

class BufferToWaveRecorderNode : public ci::audio::SampleRecorderNode {

public:

    static const float kRampTime;

    //! Constructor. numChunks is the total number of chunks this biffer has to be borken down in. 

    //! numSeconds lenght of the buffer in seconds 

    BufferToWaveRecorderNode( std::size_t numChunks, double numSeconds );

    //! Starts recording. Resets the write position to zero (call disable() to pause recording).

    void start();

    //! Stops recording. Same as calling disable().

    void stop();

    //! \brief Sets the length of the recording buffer in frames.

    //!

    //! If the write position is non-zero, the old contents will be preserved (by copying it to the newly allocated Buffer).

    //! If \a shrinkToFit is set to `true`, the internal Buffer will be down-sized if necessary, otherwise it will only re-allocate when growing while changing its dimensions to match \a numFrames (default shrinkToFit = false).

    void setNumFrames(size_t numFrames, bool shrinkToFit = false);

    //! Sets the length of the recording buffer in seconds. \see setNumFrames

    void setNumSeconds(double numSeconds, bool shrinkToFit = false);

    //! Returns the length of the recording buffer in frames.

    size_t                getNumFrames() const        { return mRecorderBuffer.getNumFrames(); }

    //! Returns the length of the recording buffer in seconds.

    double                getNumSeconds() const;

    //! \brief Returns a copy of the recored samples, up to the current write position.

    //!

    //! This method is non locking, and as such any resizing calls must be performed on the same thread or be otherwise synchronized.

    ci::audio::BufferRef        getRecordedCopy() const;

    //! \brief Writes the currently recorded samples to a file at \a filePath

    //!

    //! The encoding format is derived from \a filePath's extension and \a sampleType (default = SampleType::INT_16).

    //! \note throws AudioFileExc if the write request cannot be completed.

    void writeToFile(const ci::fs::path &filePath, ci::audio::SampleType sampleType = ci::audio::SampleType::INT_16);

    //! Returns the frame of the last buffer overrun or 0 if none since the last time this method was called. When this happens, it means the recorded buffer probably has skipped some frames.

    uint64_t getLastOverrun();

    //! returns a reference to the ring buffer when the size values of the chunks is stored, when a new wave is recorder

    RecordWaveMsgRingBuffer& getRingBuffer() { return mRingBuffer; }

    //!returns a pointer to the buffer where the audio is recorder. This is used by the PGranular to create the granular synthesis 

    ci::audio::Buffer* getRecorderBuffer() { return &mRecorderBuffer; }

protected:

    void initialize()                                override;

    void process(ci::audio::Buffer *buffer)        override;

    void initBuffers(size_t numFrames);

    static const float kMinAudioVal; 

    static const float kMaxAudioVal;

    ci::audio::BufferDynamic                mRecorderBuffer;

    ci::audio::BufferDynamicRef                mCopiedBuffer;

    std::atomic<uint64_t>        mLastOverrun;

    RecordWaveMsgRingBuffer mRingBuffer;

    const std::size_t mNumChunks;

    const double mNumSeconds;

    std::size_t mNumSamplesPerChunk;

    std::atomic<std::size_t> mChunkIndex;

    size_t mChunkSampleCounter;

    float mChunkMaxAudioVal;

    float mChunkMinAudioVal;

    float mEnvRamp;

    float mEnvRampRate;

    size_t mEnvRampLen;

    size_t mEnvDecayStart;

};