415 lines
16 KiB
C++
415 lines
16 KiB
C++
#pragma once
|
|
|
|
#include <functional>
|
|
|
|
PFC_DECLARE_EXCEPTION(exception_output_device_not_found, pfc::exception, "Audio device not found")
|
|
PFC_DECLARE_EXCEPTION(exception_output_module_not_found, exception_output_device_not_found, "Output module not found")
|
|
PFC_DECLARE_EXCEPTION(exception_output_invalidated, pfc::exception, "Audio device invalidated")
|
|
PFC_DECLARE_EXCEPTION(exception_output_device_in_use, pfc::exception, "Audio device in use")
|
|
PFC_DECLARE_EXCEPTION(exception_output_unsupported_stream_format, pfc::exception, "Unsupported audio stream format")
|
|
|
|
|
|
// =======================================================
|
|
// IDEA BIN
|
|
// ========
|
|
// Accurate timing info required! get_latency NOT safe to call from any thread while it should be
|
|
// There should be a legitimate way ( as in other than matching get_latency() against the amount of sent data ) to know when the output has finished prebuffering and started actual playback
|
|
// Outputs should be able to handle idling : idle(abort_callback&) => while(!update()) aborter.sleep(); or optimized for specific output
|
|
// =======================================================
|
|
|
|
//! Structure describing PCM audio data format, with basic helper functions.
|
|
struct t_pcmspec
|
|
{
|
|
unsigned m_sample_rate = 0;
|
|
unsigned m_bits_per_sample = 0;
|
|
unsigned m_channels = 0,m_channel_config = 0;
|
|
bool m_float = false;
|
|
|
|
inline unsigned align() const {return (m_bits_per_sample / 8) * m_channels;}
|
|
|
|
uint64_t time_to_bytes(double p_time) const {return audio_math::time_to_samples(p_time,m_sample_rate) * (m_bits_per_sample / 8) * m_channels;}
|
|
double bytes_to_time(uint64_t p_bytes) const {return (double) (p_bytes / ((m_bits_per_sample / 8) * m_channels)) / (double) m_sample_rate;}
|
|
|
|
inline bool operator==(/*const t_pcmspec & p_spec1,*/const t_pcmspec & p_spec2) const
|
|
{
|
|
return /*p_spec1.*/m_sample_rate == p_spec2.m_sample_rate
|
|
&& /*p_spec1.*/m_bits_per_sample == p_spec2.m_bits_per_sample
|
|
&& /*p_spec1.*/m_channels == p_spec2.m_channels
|
|
&& /*p_spec1.*/m_channel_config == p_spec2.m_channel_config
|
|
&& /*p_spec1.*/m_float == p_spec2.m_float;
|
|
}
|
|
|
|
inline bool operator!=(/*const t_pcmspec & p_spec1,*/const t_pcmspec & p_spec2) const
|
|
{
|
|
return !(*this == p_spec2);
|
|
}
|
|
|
|
inline void reset() { *this = t_pcmspec(); }
|
|
inline bool is_valid() const
|
|
{
|
|
return m_sample_rate >= 1000 && m_sample_rate <= 1000000 &&
|
|
m_channels > 0 && m_channels <= 256 && m_channel_config != 0 &&
|
|
(m_bits_per_sample == 8 || m_bits_per_sample == 16 || m_bits_per_sample == 24 || m_bits_per_sample == 32);
|
|
}
|
|
};
|
|
|
|
struct t_samplespec {
|
|
t_uint32 m_sample_rate;
|
|
t_uint32 m_channels,m_channel_config;
|
|
|
|
t_size time_to_samples(double p_time) const {PFC_ASSERT(is_valid());return (t_size)audio_math::time_to_samples(p_time,m_sample_rate);}
|
|
double samples_to_time(t_size p_samples) const {PFC_ASSERT(is_valid()); return audio_math::samples_to_time(p_samples,m_sample_rate);}
|
|
|
|
inline t_samplespec() {reset();}
|
|
inline t_samplespec(audio_chunk const & in) {fromchunk(in);}
|
|
|
|
inline void reset() {m_sample_rate = 0; m_channels = 0; m_channel_config = 0;}
|
|
|
|
inline bool operator==(const t_samplespec & p_spec2) const {
|
|
return m_sample_rate == p_spec2.m_sample_rate && m_channels == p_spec2.m_channels && m_channel_config == p_spec2.m_channel_config;
|
|
}
|
|
|
|
inline bool operator!=(const t_samplespec & p_spec2) const {
|
|
return !(*this == p_spec2);
|
|
}
|
|
|
|
inline bool is_valid() const {
|
|
return m_sample_rate > 0 && m_channels > 0 && audio_chunk::g_count_channels(m_channel_config) == m_channels;
|
|
}
|
|
|
|
static t_samplespec g_fromchunk(const audio_chunk & p_chunk) {
|
|
t_samplespec temp; temp.fromchunk(p_chunk); return temp;
|
|
}
|
|
|
|
void fromchunk(const audio_chunk & p_chunk) {
|
|
m_sample_rate = p_chunk.get_sample_rate();
|
|
m_channels = p_chunk.get_channels();
|
|
m_channel_config = p_chunk.get_channel_config();
|
|
}
|
|
};
|
|
|
|
class NOVTABLE output_device_enum_callback
|
|
{
|
|
public:
|
|
virtual void on_device(const GUID & p_guid,const char * p_name,unsigned p_name_length) = 0;
|
|
};
|
|
|
|
class NOVTABLE output : public service_base {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output,service_base);
|
|
public:
|
|
//! Retrieves amount of audio data queued for playback, in seconds.
|
|
virtual double get_latency() = 0;
|
|
//! Sends new samples to the device. Allowed to be called only when update() indicates that the device is ready.
|
|
virtual void process_samples(const audio_chunk & p_chunk) = 0;
|
|
//! Updates playback; queries whether the device is ready to receive new data.
|
|
//! @param p_ready On success, receives value indicating whether the device is ready for next process_samples() call.
|
|
virtual void update(bool & p_ready) = 0;
|
|
//! Pauses/unpauses playback.
|
|
virtual void pause(bool p_state) = 0;
|
|
//! Flushes queued audio data. Called after seeking.
|
|
virtual void flush() = 0;
|
|
//! Forces playback of queued data. Called when there's no more data to send, to prevent infinite waiting if output implementation starts actually playing after amount of data in internal buffer reaches some level.
|
|
virtual void force_play() = 0;
|
|
|
|
//! Sets playback volume.
|
|
//! @p_val Volume level in dB. Value of 0 indicates full ("100%") volume, negative values indciate different attenuation levels.
|
|
virtual void volume_set(double p_val) = 0;
|
|
|
|
//! Helper, see output_v4::is_progressing().
|
|
bool is_progressing_();
|
|
//! Helper, see output_v4::update_v2()
|
|
size_t update_v2_();
|
|
//! Helper, see output_v4::get_event_trigger()
|
|
pfc::eventHandle_t get_trigger_event_();
|
|
|
|
//! Helper for output_entry implementation.
|
|
static uint32_t g_extra_flags() { return 0; }
|
|
|
|
};
|
|
|
|
class NOVTABLE output_v2 : public output {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_v2, output);
|
|
public:
|
|
//! Obsolete, do not use.
|
|
virtual bool want_track_marks() {return false;}
|
|
//! Obsolete, do not use.
|
|
virtual void on_track_mark() {}
|
|
//! Obsolete, do not use.
|
|
virtual void enable_fading(bool state) {}
|
|
//! Called when flushing due to manual track change rather than seek-within-track
|
|
virtual void flush_changing_track() {flush();}
|
|
};
|
|
|
|
class dsp_chain_config;
|
|
|
|
//! \since 1.4
|
|
class NOVTABLE output_v3 : public output_v2 {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_v3, output_v2);
|
|
public:
|
|
//! Does this output require a specific sample rate? If yes, return the value, otherwise return zero. \n
|
|
//! Returning a nonzero will cause a resampler DSP to be injected.
|
|
virtual unsigned get_forced_sample_rate() { return 0; }
|
|
//! Allows the output to inject specific DSPs at the end of the used chain. \n
|
|
//! Default implementation queries get_forced_sample_rate() and injects a resampler.
|
|
virtual void get_injected_dsps( dsp_chain_config & );
|
|
};
|
|
|
|
//! \since 1.6
|
|
class NOVTABLE output_v4 : public output_v3 {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_v4, output_v3);
|
|
public:
|
|
//! Returns an event handle that becomes signaled once the output wants an update() call and possibly process_samples(). \n
|
|
//! Optional; may return pfc::eventInvalid if not available at this time or not supported. \n
|
|
//! If implemented, calling update() should clear the event each time.
|
|
virtual pfc::eventHandle_t get_trigger_event() {return pfc::eventInvalid;}
|
|
//! Returns whether the audio stream is currently being played or not. \n
|
|
//! Typically, for a short period of time, initially send data is not played until a sufficient amount is queued to initiate playback without glitches. \n
|
|
//! For old outputs that do not implement this, the value can be assumed to be true.
|
|
virtual bool is_progressing() {return true;}
|
|
|
|
//! Improved version of update(); returns 0 if the output isn't ready to receive any new data, otherwise an advisory number of samples - at the current stream format - that the output expects to take now. \n
|
|
//! If the caller changes the stream format, the value is irrelevant. \n
|
|
//! The output may return SIZE_MAX to indicate that it can take data but does not currently have any hints to tell how much.
|
|
virtual size_t update_v2();
|
|
};
|
|
|
|
//! \since 1.6
|
|
class output_v5 : public output_v4 {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_v5, output_v4);
|
|
public:
|
|
virtual unsigned get_forced_channel_mask() { return 0; }
|
|
};
|
|
|
|
class NOVTABLE output_entry : public service_base {
|
|
FB2K_MAKE_SERVICE_INTERFACE_ENTRYPOINT(output_entry);
|
|
public:
|
|
//! Instantiates output class.
|
|
virtual void instantiate(service_ptr_t<output> & p_out,const GUID & p_device,double p_buffer_length,bool p_dither,t_uint32 p_bitdepth) = 0;
|
|
//! Enumerates devices supported by this output_entry implementation.
|
|
virtual void enum_devices(output_device_enum_callback & p_callback) = 0;
|
|
//! For internal use by backend. Each output_entry implementation must have its own guid.
|
|
virtual GUID get_guid() = 0;
|
|
//! For internal use by backend. Retrieves human-readable name of this output_entry implementation.
|
|
virtual const char * get_name() = 0;
|
|
|
|
//! Obsolete, do not use.
|
|
virtual void advanced_settings_popup(HWND p_parent,POINT p_menupoint) = 0;
|
|
|
|
enum {
|
|
flag_needs_bitdepth_config = 1 << 0,
|
|
flag_needs_dither_config = 1 << 1,
|
|
//! Obsolete, do not use.
|
|
flag_needs_advanced_config = 1 << 2,
|
|
flag_needs_device_list_prefixes = 1 << 3,
|
|
|
|
//! Supports playing multiple simultaneous audio streams thru one device?
|
|
flag_supports_multiple_streams = 1 << 4,
|
|
|
|
//! High latency operation (such as remote network playback), mutually exclusive with flag_low_latency
|
|
flag_high_latency = 1 << 5,
|
|
//! Low latency operation (local playback), mutually exclusive with flag_high_latency
|
|
flag_low_latency = 1 << 6,
|
|
//! When set, the output will be used in special compatibility mode: guaranteed regular update() calls, injected padding (silence) at the end of stream.
|
|
flag_needs_shims = 1 << 7,
|
|
};
|
|
|
|
virtual t_uint32 get_config_flags() = 0;
|
|
|
|
uint32_t get_config_flags_compat();
|
|
|
|
bool is_high_latency();
|
|
bool is_low_latency();
|
|
|
|
pfc::string8 get_device_name( const GUID & deviceID);
|
|
bool get_device_name( const GUID & deviceID, pfc::string_base & out );
|
|
|
|
static bool g_find( const GUID & outputID, output_entry::ptr & outObj );
|
|
static output_entry::ptr g_find(const GUID & outputID );
|
|
};
|
|
|
|
//! Helper; implements output_entry for specific output class implementation. output_entry methods are forwarded to static methods of your output class. Use output_factory_t<myoutputclass> instead of using this class directly.
|
|
template<typename T, typename E = output_entry>
|
|
class output_entry_impl_t : public E
|
|
{
|
|
public:
|
|
void instantiate(service_ptr_t<output> & p_out,const GUID & p_device,double p_buffer_length,bool p_dither,t_uint32 p_bitdepth) {
|
|
p_out = new service_impl_t<T>(p_device,p_buffer_length,p_dither,p_bitdepth);
|
|
}
|
|
void enum_devices(output_device_enum_callback & p_callback) {T::g_enum_devices(p_callback);}
|
|
GUID get_guid() {return T::g_get_guid();}
|
|
const char * get_name() {return T::g_get_name();}
|
|
void advanced_settings_popup(HWND p_parent,POINT p_menupoint) {T::g_advanced_settings_popup(p_parent,p_menupoint);}
|
|
|
|
t_uint32 get_config_flags() {
|
|
t_uint32 flags = 0;
|
|
if (T::g_advanced_settings_query()) flags |= output_entry::flag_needs_advanced_config;
|
|
if (T::g_needs_bitdepth_config()) flags |= output_entry::flag_needs_bitdepth_config;
|
|
if (T::g_needs_dither_config()) flags |= output_entry::flag_needs_dither_config;
|
|
if (T::g_needs_device_list_prefixes()) flags |= output_entry::flag_needs_device_list_prefixes;
|
|
if (T::g_supports_multiple_streams()) flags |= output_entry::flag_supports_multiple_streams;
|
|
if (T::g_is_high_latency()) flags |= output_entry::flag_high_latency;
|
|
else flags |= output_entry::flag_low_latency;
|
|
flags |= T::g_extra_flags();
|
|
return flags;
|
|
}
|
|
};
|
|
|
|
|
|
//! Use this to register your output implementation.
|
|
template<class T>
|
|
class output_factory_t : public service_factory_single_t<output_entry_impl_t<T> > {};
|
|
|
|
class output_impl : public output_v5 {
|
|
protected:
|
|
output_impl() : m_incoming_ptr(0) {}
|
|
virtual void on_update() = 0;
|
|
//! Will never get more input than as returned by can_write_samples().
|
|
virtual void write(const audio_chunk & p_data) = 0;
|
|
virtual t_size can_write_samples() = 0;
|
|
virtual t_size get_latency_samples() = 0;
|
|
virtual void on_flush() = 0;
|
|
virtual void on_flush_changing_track() {on_flush();}
|
|
virtual void open(t_samplespec const & p_spec) = 0;
|
|
|
|
virtual void pause(bool p_state) = 0;
|
|
virtual void force_play() = 0;
|
|
virtual void volume_set(double p_val) = 0;
|
|
protected:
|
|
void on_need_reopen() {m_active_spec = t_samplespec();}
|
|
private:
|
|
void flush();
|
|
void flush_changing_track();
|
|
void update(bool & p_ready);
|
|
size_t update_v2();
|
|
double get_latency();
|
|
void process_samples(const audio_chunk & p_chunk);
|
|
|
|
pfc::array_t<audio_sample,pfc::alloc_fast_aggressive> m_incoming;
|
|
t_size m_incoming_ptr;
|
|
t_samplespec m_incoming_spec,m_active_spec;
|
|
};
|
|
|
|
|
|
class NOVTABLE volume_callback {
|
|
public:
|
|
virtual void on_volume_scale(float v) = 0;
|
|
virtual void on_volume_arbitrary(int v) = 0;
|
|
};
|
|
|
|
class NOVTABLE volume_control : public service_base {
|
|
FB2K_MAKE_SERVICE_INTERFACE(volume_control, service_base)
|
|
public:
|
|
virtual void add_callback(volume_callback * ptr) = 0;
|
|
virtual void remove_callback(volume_callback * ptr) = 0;
|
|
|
|
enum style_t {
|
|
styleScale,
|
|
styleArbitrary
|
|
};
|
|
|
|
virtual style_t getStyle() = 0;
|
|
|
|
virtual float scaleGet() = 0;
|
|
virtual void scaleSet(float v) = 0;
|
|
|
|
virtual void arbitrarySet(int val) = 0;
|
|
virtual int arbitraryGet() = 0;
|
|
virtual int arbitraryGetMin() = 0;
|
|
virtual int arbitraryGetMax() = 0;
|
|
virtual bool arbitraryGetMute() = 0;
|
|
virtual void arbitrarySetMute(bool val) = 0;
|
|
};
|
|
|
|
|
|
class NOVTABLE output_entry_v2 : public output_entry {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_entry_v2, output_entry)
|
|
public:
|
|
virtual bool get_volume_control(const GUID & id, volume_control::ptr & out) = 0;
|
|
virtual bool hasVisualisation() = 0;
|
|
};
|
|
|
|
//! \since 1.5
|
|
class NOVTABLE output_devices_notify {
|
|
public:
|
|
virtual void output_devices_changed() = 0;
|
|
protected:
|
|
output_devices_notify() {}
|
|
private:
|
|
output_devices_notify(const output_devices_notify &) = delete;
|
|
void operator=(const output_devices_notify &) = delete;
|
|
};
|
|
|
|
//! \since 1.5
|
|
class NOVTABLE output_entry_v3 : public output_entry_v2 {
|
|
FB2K_MAKE_SERVICE_INTERFACE(output_entry_v3, output_entry_v2)
|
|
public:
|
|
|
|
//! Main thread only!
|
|
virtual void add_notify(output_devices_notify *) = 0;
|
|
//! Main thread only!
|
|
virtual void remove_notify(output_devices_notify *) = 0;
|
|
|
|
//! Main thread only!
|
|
virtual void set_pinned_device(const GUID & guid) = 0;
|
|
};
|
|
|
|
#pragma pack(push, 1)
|
|
//! \since 1.3.5
|
|
struct outputCoreConfig_t {
|
|
|
|
static outputCoreConfig_t defaults();
|
|
|
|
GUID m_output;
|
|
GUID m_device;
|
|
double m_buffer_length;
|
|
uint32_t m_flags;
|
|
uint32_t m_bitDepth;
|
|
enum { flagUseDither = 1 << 0 };
|
|
};
|
|
#pragma pack(pop)
|
|
|
|
//! \since 1.3.5
|
|
//! Allows components to access foobar2000 core's output settings.
|
|
class NOVTABLE output_manager : public service_base {
|
|
FB2K_MAKE_SERVICE_COREAPI(output_manager);
|
|
public:
|
|
//! Instantiates an output instance with core settings.
|
|
//! @param overrideBufferLength Specify non zero to override user-configured buffer length in core settings.
|
|
//! @returns The new output instance. Throws exceptions on failure (invalid settings or other).
|
|
virtual output::ptr instantiateCoreDefault(double overrideBufferLength = 0) = 0;
|
|
virtual void getCoreConfig( void * out, size_t outSize ) = 0;
|
|
|
|
void getCoreConfig(outputCoreConfig_t & out ) { getCoreConfig(&out, sizeof(out) ); }
|
|
};
|
|
|
|
//! \since 1.3.16
|
|
class NOVTABLE output_device_list_callback {
|
|
public:
|
|
virtual void onDevice( const char * fullName, const GUID & output, const GUID & device ) = 0;
|
|
};
|
|
|
|
//! \since 1.3.16
|
|
class NOVTABLE output_config_change_callback {
|
|
public:
|
|
virtual void outputConfigChanged() = 0;
|
|
};
|
|
|
|
//! \since 1.4
|
|
class NOVTABLE output_manager_v2 : public output_manager {
|
|
FB2K_MAKE_SERVICE_COREAPI_EXTENSION(output_manager_v2, output_manager);
|
|
public:
|
|
virtual void setCoreConfig( const void * in, size_t inSize, bool bSuppressPlaybackRestart = false ) = 0;
|
|
void setCoreConfig( const outputCoreConfig_t & in ) { setCoreConfig(&in, sizeof(in) ); }
|
|
virtual void setCoreConfigDevice( const GUID & output, const GUID & device ) = 0;
|
|
virtual void listDevices( output_device_list_callback & callback ) = 0;
|
|
void listDevices( std::function< void ( const char*, const GUID&, const GUID&) > f );
|
|
virtual void addCallback( output_config_change_callback * ) = 0;
|
|
virtual void removeCallback( output_config_change_callback * ) = 0;
|
|
|
|
service_ptr addCallback( std::function<void()> f );
|
|
void addCallbackPermanent( std::function<void()> f );
|
|
};
|
|
|
|
extern const GUID output_id_null;
|
|
extern const GUID output_id_default;
|