DemuxerMP4 extracts both the audio and (H.264) video track from a general, interleaved MP4/ISO-BMFF stream - unlike M4AAudioDemuxer/ M4AAudioFileDemuxer, which only handle a single (audio) track. More...

#include <ContainerMP4.h>

Inheritance diagram for DemuxerMP4:

Classes
struct	StscEntry
	sample-to-chunk table entry (stsc) More...

struct	SttsEntry

struct	Track
	One track (audio or video) as found in 'moov'. More...

Public Types
using	Codec = M4ACommonDemuxer::Codec

Public Member Functions
	DemuxerMP4 ()

	DemuxerMP4 (Print &video_out, Print &audio_out)

	DemuxerMP4 (SeekableSource &seekSource)

	DemuxerMP4 (SpoolStorageFactory &spoolFactory)

	DemuxerMP4 (VideoOutput &video_out, Print &audio_out)

	~DemuxerMP4 ()

virtual void	addNotifyAudioChange (AudioInfoSupport &bi)
	Adds target to be notified about audio changes.

Vector< uint8_t > &	audioALACMagicCookie ()

AudioInfo	audioInfo () override
	provides the actual input AudioInfo

virtual AudioInfo	audioInfoOut ()

bool	begin () override

virtual bool	begin (AudioInfo info) override

virtual void	clearNotifyAudioChange ()
	Deletes all change notify subscriptions.

void	end () override

AudioInfoFormat	getAudioInfo () override

Print *	getOutput ()

VideoInfo	getVideoInfo () override

bool	isNotifyActive ()
	Checks if the automatic AudioInfo update is active.

bool	isValid (const uint8_t *data, size_t len) override

const char *	mime () override

const char *	mimeVideo () override
	Provides the container mime.

	operator bool () override

bool	quickStart ()

virtual bool	removeNotifyAudioChange (AudioInfoSupport &bi)
	Removes a target in order not to be notified about audio changes.

void	setAudioInfo (AudioInfo from) override
	for most decoders this is not needed

virtual bool	setCodecConfig (const uint8_t *data, size_t len)
	Some decoders need e.g. a magic cookie to provide the relevant info for decoding.

void	setNotifyActive (bool flag)
	Deactivate/Reactivate automatic AudioInfo updates: (default is active)

virtual void	setOutput (AudioOutput &out_stream)
	Defines where the decoded result is written to.

virtual void	setOutput (AudioStream &out_stream)
	Defines where the decoded result is written to.

void	setOutput (Print &out) override

void	setOutputAudio (Print &out) override

void	setOutputVideo (Print &out) override

void	setOutputVideo (VideoOutput &out) override

void	setSeekSource (SeekableSource &seekSource)

void	setSpoolStorageFactory (SpoolStorageFactory &spoolFactory)

VideoInfo	videoInfo () override

size_t	write (const uint8_t *data, size_t len) override

Public Attributes
int	id
	custom id to be used by application

Protected Member Functions
void	advanceChunkIfNeeded ()

bool	beginBoxAccum (const MP4Parser::Box &box)

uint64_t	boxPayloadFileOffset ()

void	configureSpoolStores (Track &t)

void	dispatchAudio (Track &t, const uint8_t *data, size_t size, bool isFirst)

void	dispatchSample (Track &t, const uint8_t *data, size_t size, bool isFirst)

void	dispatchVideo (Track &t, const uint8_t *data, size_t size, bool isFirst)

void	endBoxAccum ()
	Call once a box_accum-using handler has fully processed a complete box.

void	feed (const uint8_t *data, size_t len)

void	flushVideo ()

void	freeTracks ()

void	notifyAudioChange (AudioInfo info)

void	onAc3 (const MP4Parser::Box &box)

void	onAlac (const MP4Parser::Box &box)

void	onAvc1 (const MP4Parser::Box &box)

void	onAvcC (const MP4Parser::Box &box)

void	onEsds (const MP4Parser::Box &box)

void	onHdlr (MP4Parser::Box &box)

void	onHevc (const MP4Parser::Box &box)

void	onMdat (MP4Parser::Box &box)

void	onMdhd (MP4Parser::Box &box)

void	onMp4a (const MP4Parser::Box &box)

void	onStco (MP4Parser::Box &box, bool is64)

void	onStsc (MP4Parser::Box &box)

void	onStsd (MP4Parser::Box &box)

void	onStsz (MP4Parser::Box &box)

void	onStts (MP4Parser::Box &box)

void	onTrak ()

uint32_t	samplesInChunk (Track &t, uint32_t chunkIndex0)

void	sendWavHeader ()

void	setSendWavHeader (bool flag) override

void	setupAudioInfo (AudioFormat format, const uint8_t *entryData, size_t entrySize)

void	setupParser ()

void	writeBlocking (Print out, uint8_t data, size_t len)

void	writeVideo (uint8_t *data, size_t size)

Static Protected Member Functions
static StscEntry	readStscEntry (const uint8_t *p)

static SttsEntry	readSttsEntry (const uint8_t *p)

static uint16_t	readU16 (const uint8_t *p)

static uint32_t	readU32 (const uint8_t *p)

static uint64_t	readU32Widened (const uint8_t *p)

static uint64_t	readU64 (const uint8_t *p)

static void	writeAdtsHeader (uint8_t *adts, int aacProfile, int sampleRateIdx, int channelCfg, int frameLen)

Protected Attributes
AudioFormat	audio_format = AudioFormat::UNKNOWN

SingleBuffer< uint8_t >	box_accum {0}

bool	box_accum_active = false

uint64_t	box_start_file_offset = 0

Track *	current_chunk_track = nullptr

bool	current_sample_is_first_for_track = false

uint32_t	current_sample_size = 0

Track *	current_sample_track = nullptr

Track *	current_track
	track currently being parsed (scoped by 'trak')

uint32_t	dispatched_sample_count = 0

AudioInfo	info

bool	is_active = false

bool	is_notify_active = true

bool	is_wav_header_sent = false

bool	mdat_seen = false

Vector< uint8_t >	nal_tmp
	scratch buffer for Annex-B conversion

Vector< AudioInfoSupport * >	notify_vector

Track *	p_audio_track = nullptr

Print *	p_output_audio = nullptr

Print *	p_print = nullptr

SeekableSource *	p_seek_source = nullptr

SpoolStorageFactory *	p_spool_factory = nullptr

Print *	p_video_out = nullptr

VideoOutput *	p_video_out_video = nullptr

Track *	p_video_track = nullptr
	first video track found

MP4Parser	parser

int64_t	quickstart_bytes_remaining = -1

Vector< uint8_t >	sample_buffer

size_t	sample_buffer_filled = 0

uint32_t	samples_left_in_chunk = 0

bool	send_wav_header = false

bool	send_wav_header_explicit = false

bool	stsz_header_pending = true

uint32_t	total_bytes_received = 0

Vector< Track * >	tracks

Detailed Description

DemuxerMP4 extracts both the audio and (H.264) video track from a general, interleaved MP4/ISO-BMFF stream - unlike M4AAudioDemuxer/ M4AAudioFileDemuxer, which only handle a single (audio) track.

This is a streaming (forward-only) demuxer: it does not need a seekable source, so it works directly over e.g. an HTTP download (URLStream) or an SD file read sequentially with StreamCopy - the same way DemuxerAVI is used. It relies on the file being "faststart" muxed (moov before mdat, e.g. ffmpeg ... -movflags +faststart) - the same requirement MP4Parser and the M4A demuxers already document.

How the interleaving is resolved without seeking: once 'moov' has been fully parsed, every track's stsz (sample sizes) + stsc (sample-to-chunk)

stco/co64 (absolute chunk offsets) tables are known. Those chunk offsets already reveal the exact physical byte order the file was muxed in - i.e. the order 'mdat' bytes will arrive in over a plain sequential stream. So instead of seeking to the next-needed chunk, a "merge schedule" (which track's chunk comes next, and how many samples it holds) is computed once, and then simply followed by consuming that many bytes off the incoming stream for each successive sample.

Note: v1 scope: one video track (H.264/avc1 only - HEVC is detected and skipped with a warning) and one audio track (AAC or ALAC). Video output is the raw H.264 Annex-B elementary stream (SPS/PPS from avcC converted to Annex-B and prepended before the first frame) via VideoOutput - there is no pixel decoder anywhere in this library, same scope boundary as DemuxerAVI. No stts/timing support yet - write() dispatches samples as soon as they are complete; pacing is the caller's responsibility.; Relationship to ContainerM4A/M4AAudioDemuxer (AudioCodecs/ ContainerM4A.h, AudioCodecs/M4AAudioDemuxer.h): those handle a single audio-only track and implement ContainerDecoder, so they drop straight into AudioPlayer/MultiDecoder - use them if you only need audio out of an .m4a/.mp4. This class is multi-track (audio+video) and implements Demuxer instead, for the video pipeline (VideoPlayer, PacedVideoOutput, MultiVideoDemuxer). Both parse the same MP4 boxes for the audio sample entry (stsd/mp4a/esds for AAC, stsd/alac for ALAC's magic cookie), but this class keeps its own copy (onStsd/onMp4a/onEsds/onAlac below) rather than inheriting M4ACommonDemuxer like the M4A classes do - M4ACommonDemuxer's handlers assume one single-track audio_config, while this class needs per-track state (current_track) plus random-access sample tables (SampleTableStore) for the interleave merge schedule, so folding the two together isn't a straightforward inheritance change. Kept deliberately separate rather than unified; a fix to one track-audio-config parsing path may need to be mirrored in the other.

Author: Phil Schatzmann

Copyright: GPLv3

Member Typedef Documentation

◆ Codec

using Codec = M4ACommonDemuxer::Codec

Constructor & Destructor Documentation

◆ DemuxerMP4() [1/5]

DemuxerMP4 ( )

inline

This class only demuxes - it does not decode audio itself. Point setOutputAudio() at an EncodedAudioStream (wrapping whatever AudioDecoder matches the track's codec - AAC/ALAC) if you need the audio track decoded; the raw payload (ADTS-wrapped for AAC, raw for ALAC) is written through as-is otherwise. For ALAC, the decoder also needs the magic cookie exposed via audioALACMagicCookie() - configure your decoder with it (e.g. AudioDecoder::setCodecConfig()) once it's available (after the audio track's 'stsd' has been parsed, i.e. once getAudioInfo().format != AudioFormat::UNKNOWN). Point setOutputVideo() at a Print (e.g. a VideoOutput) to receive the demuxed H.264 Annex-B video track; leave unset to ignore video.

◆ DemuxerMP4() [2/5]

DemuxerMP4	(	Print &	video_out,
		Print &	audio_out
	)

inline

◆ DemuxerMP4() [3/5]

DemuxerMP4	(	VideoOutput &	video_out,
		Print &	audio_out
	)

inline

◆ DemuxerMP4() [4/5]

DemuxerMP4 ( SeekableSource & seekSource )

inline

'seekSource' wraps the same open file this instance will be fed from (e.g. via FileSeekableSource<File>(myFile)) - when set, newly discovered tracks' sample tables (stsz/stco/stsc/stts) are backed by SourceSeekSampleTableStore instead of the RAM-backed default, trading the ~4.5MB (for a feature-length file) RAM cost for small repeated seeks back into 'seekSource' during playback. Must stay valid and open for the lifetime of playback. See setSeekSource().

◆ DemuxerMP4() [5/5]

DemuxerMP4 ( SpoolStorageFactory & spoolFactory )

inline

'spoolFactory' provides a fresh SpoolStorage for each of a track's four sample tables as tracks are discovered - when set, backs new tracks' sample tables with SpoolFileSampleTableStore instead of the RAM-backed default. Unlike the SeekableSource overload above, this needs no seekability in the original MP4 source at all (it writes its own local copy as data streams past), at the cost of real disk I/O for both the writes during 'moov' parsing and the seek+reads during 'mdat' playback. See setSpoolStorageFactory().

◆ ~DemuxerMP4()

~DemuxerMP4 ( )

inline

Member Function Documentation

◆ addNotifyAudioChange()

virtual void addNotifyAudioChange ( AudioInfoSupport & bi )

inlinevirtualinherited

Adds target to be notified about audio changes.

Reimplemented in CodecNOP, EncodedAudioOutput, EncodedAudioStream, AACDecoderFDK, DecoderBasic, CodecChain, MP3DecoderHelix, MP3DecoderMAD, OggContainerDecoder, RTSPClient< TcpClient, UdpSocket >, Pipeline, and Pipeline::ModifyingStreamAdapter.

◆ advanceChunkIfNeeded()

void advanceChunkIfNeeded ( )

inlineprotected

Picks whichever track's next not-yet-consumed chunk comes first in the file (by chunk_offset) and makes it the active chunk - an incremental two-way merge instead of precomputing a whole-file schedule: each track's own chunk_offsets are already ascending (a muxer writes each track's chunks in file order), so the merged cross-track order falls out of comparing just the next pending chunk_offset per track, in O(1) per call instead of materializing and sorting every chunk from every track upfront.

◆ audioALACMagicCookie()

Vector< uint8_t > & audioALACMagicCookie ( )

inline

ALAC magic cookie (the 'alac' box payload, without its own size/type prefix) needed to configure an external ALAC decoder - empty unless getAudioInfo().format == AudioFormat::ALAC.

◆ audioInfo()

AudioInfo audioInfo ( )

inlineoverridevirtualinherited

provides the actual input AudioInfo

Implements AudioInfoSupport.

Reimplemented in CodecNOP, DecoderBasic, DSFDecoder, MP3DecoderHelix, MP3DecoderMAD, OpusAudioDecoder, OpusMultiStreamAudioDecoder, VorbisDecoder, WAVDecoder, WavIMADecoder, OggContainerDecoder, and VorbisDecoder.

◆ audioInfoOut()

virtual AudioInfo audioInfoOut ( )

inlinevirtualinherited

provides the actual output AudioInfo: this is usually the same as audioInfo() unless we use a transforming stream

Reimplemented in MP3EncoderShine, PureDataStream, PWMAudioOutput< PWMDriverT >, ChannelFormatConverterStreamT< T >, ChannelFormatConverterStream, NumberFormatConverterStreamT< TFrom, TTo >, NumberFormatConverterStream, FormatConverterStream, Pipeline, ResampleStream, ResampleStreamT< TInterpolator >, and SupportedRatesStream.

◆ begin() [1/2]

bool begin ( )

inlineoverridevirtual

Reimplemented from AudioDecoder.

◆ begin() [2/2]

virtual bool begin ( AudioInfo info )

inlineoverridevirtualinherited

Reimplemented from AudioWriter.

Reimplemented in DecoderBasic, and OggContainerDecoder.

◆ beginBoxAccum()

bool beginBoxAccum ( const MP4Parser::Box & box )

inlineprotected

Call at the top of every box_accum-using handler. Returns true the first time it's called for a given box (i.e. right when accumulation starts), false on continuations - callers that need to (re)initialize per-box state (e.g. onStsz's stsz_header_pending) should key off that.

◆ boxPayloadFileOffset()

uint64_t boxPayloadFileOffset ( )

inlineprotected

Absolute file offset of the first byte of this box's payload (box_accum's content), i.e. right after the box's own 8-byte size+type header - used, plus each table's own sub-header size on top, to tell a SourceSeekSampleTableStore where a table's first entry lives on disk.

Deliberately based on box_start_file_offset (captured once, at the box's first delivery - see beginBoxAccum()) rather than computed from total_bytes_received/box_accum.available() at entry-iteration time: MP4Parser advances its own internal fileOffset to the box's end as soon as an incremental box starts (see startIncrementalBox() in MP4Parser.h), before the box is actually fully delivered - a formula relying on "bytes received so far" at the time entries are iterated (which only happens on the box's last delivery) silently used the wrong file position for any box large enough to span multiple write() calls (which every sample table in a feature-length file is) - confirmed by hex-dumping the file at the computed offset and finding it landed inside 'mdat', nowhere near the actual table.

◆ clearNotifyAudioChange()

virtual void clearNotifyAudioChange ( )

inlinevirtualinherited

Deletes all change notify subscriptions.

Reimplemented in RTSPClient< TcpClient, UdpSocket >.

◆ configureSpoolStores()

void configureSpoolStores ( Track & t )

inlineprotected

Swaps a track's four sample tables for spool-backed stores, one SpoolStorage per table obtained from p_spool_factory. Guarded so it only ever runs once per track even if onHdlr() somehow saw its kind change or got called again.

◆ dispatchAudio()

void dispatchAudio	(	Track &	t,
		const uint8_t *	data,
		size_t	size,
		bool	isFirst
	)

inlineprotected

◆ dispatchSample()

void dispatchSample	(	Track &	t,
		const uint8_t *	data,
		size_t	size,
		bool	isFirst
	)

inlineprotected

◆ dispatchVideo()

void dispatchVideo	(	Track &	t,
		const uint8_t *	data,
		size_t	size,
		bool	isFirst
	)

inlineprotected

◆ end()

void end ( )

inlineoverridevirtual

Reimplemented from AudioDecoder.

◆ endBoxAccum()

void endBoxAccum ( )

inlineprotected

Call once a box_accum-using handler has fully processed a complete box.

◆ feed()

void feed	(	const uint8_t *	data,
		size_t	len
	)

inlineprotected

Feeds raw mdat bytes; accumulates until the current sample is complete, dispatches it, and moves on - possibly switching tracks at chunk boundaries per the incremental cross-track merge (see advanceChunkIfNeeded()).

◆ flushVideo()

void flushVideo ( )

inlineprotected

◆ freeTracks()

void freeTracks ( )

inlineprotected

◆ getAudioInfo()

AudioInfoFormat getAudioInfo ( )

inlineoverridevirtual

Common audio info (sample_rate/channels/bits_per_sample), extended with the parsed codec format tag - the same AudioInfoFormat type is also provided by DemuxerAVI. Named getAudioInfo() rather than audioInfo() because AudioDecoder already declares a virtual AudioInfo audioInfo() (returning the plain, unextended type by value) - by-value virtual returns can't be covariantly widened in C++, so reusing that name for a subclass-returning version isn't possible (it would be a hard "invalid covariant return type" error, not silent hiding).

Implements Demuxer.

◆ getOutput()

Print * getOutput ( )

inlineinherited

◆ getVideoInfo()

VideoInfo getVideoInfo ( )

inlineoverridevirtual

Common video info (width/height/format/frame_size/total_file_size), analogous to audioInfo() - the same VideoInfo type is also provided by DemuxerAVI. format is VideoFormat::H264 (the only video codec this class demuxes) once the video track's 'stsd' has been parsed, UNKNOWN before that or for an unsupported codec (e.g. HEVC). frame_size is always 0 (H264 frame size varies) - kept for symmetry with DemuxerAVI, whose raw pixel formats do have a fixed per-frame size. total_file_size has no declared-upfront equivalent in MP4 (no single header field states it, unlike AVI's RIFF header), so it reports the number of bytes received via write() so far instead - only equal to the true total once the whole stream has been fed in.

Implements Demuxer.

◆ isNotifyActive()

bool isNotifyActive ( )

inlineinherited

Checks if the automatic AudioInfo update is active.

◆ isValid()

bool isValid	(	const uint8_t *	data,
		size_t	len
	)

inlineoverridevirtual

True if data starts with an ISO base media 'ftyp' box (4-byte box size, then "ftyp" at offset 4) - the standard start of an MP4/M4A file - see Demuxer::isValid().

Reimplemented from Demuxer.

◆ mime()

const char * mime ( )

inlineoverridevirtual

Provides the audio mime (AudioFormat::UNKNOWN/nullptr until the audio track's 'stsd' has been parsed - see getAudioInfo())

Implements Demuxer.

◆ mimeVideo()

const char * mimeVideo ( )

inlineoverridevirtual

Provides the container mime.

Implements Demuxer.

◆ notifyAudioChange()

void notifyAudioChange ( AudioInfo info )

inlineprotectedinherited

◆ onAc3()

void onAc3 ( const MP4Parser::Box & box )

inlineprotected

AC-3 frames are self-synchronizing (each frame starts with its own 0x0B77 sync word and carries its own sample rate/channel info), so - unlike onMp4a/onAlac - there is no child config box to parse here.

◆ onAlac()

void onAlac ( const MP4Parser::Box & box )

inlineprotected

◆ onAvc1()

void onAvc1 ( const MP4Parser::Box & box )

inlineprotected

◆ onAvcC()

void onAvcC ( const MP4Parser::Box & box )

inlineprotected

◆ onEsds()

void onEsds ( const MP4Parser::Box & box )

inlineprotected

◆ onHdlr()

void onHdlr ( MP4Parser::Box & box )

inlineprotected

◆ onHevc()

void onHevc ( const MP4Parser::Box & box )

inlineprotected

◆ onMdat()

void onMdat ( MP4Parser::Box & box )

inlineprotected

◆ onMdhd()

void onMdhd ( MP4Parser::Box & box )

inlineprotected

'mdhd' (in 'mdia', scoped by the enclosing 'trak'): captures this track's own timescale (ticks/second), needed together with 'stts' to compute each sample's scheduled presentation time.

◆ onMp4a()

void onMp4a ( const MP4Parser::Box & box )

inlineprotected

◆ onStco()

void onStco	(	MP4Parser::Box &	box,
		bool	is64
	)

inlineprotected

◆ onStsc()

void onStsc ( MP4Parser::Box & box )

inlineprotected

◆ onStsd()

void onStsd ( MP4Parser::Box & box )

inlineprotected

◆ onStsz()

void onStsz ( MP4Parser::Box & box )

inlineprotected

◆ onStts()

void onStts ( MP4Parser::Box & box )

inlineprotected

'stts' (time-to-sample): run-length list of {sample_count, sample_delta} - see getVideoInfo(), which derives fps from the first entry's sample_delta.

◆ onTrak()

void onTrak ( )

inlineprotected

◆ operator bool()

operator bool ( )

inlineoverridevirtual

Implements AudioWriter.

◆ quickStart()

bool quickStart ( )

inline

Speeds up startup for a "non-faststart" MP4 (moov written after mdat) by locating and parsing 'moov' directly via seek()/readBytes() on the configured SeekableSource (setSeekSource()), instead of streaming through 'mdat' first to reach it. Only box headers (8 bytes each) are read - the payload is skipped via seek() - so this costs a handful of tiny reads regardless of 'mdat's size.

Call after begin(), before feeding data via write() - or just let FileSeekableSource call it automatically (its default behavior).

On success, leaves the SeekableSource positioned at 'mdat', ready for the normal write()-loop to resume there. Returns false (no side effects) if already faststart, no SeekableSource is configured, or a box form this scan doesn't understand (64-bit or "extends to EOF" size) is encountered.

◆ readStscEntry()

static StscEntry readStscEntry ( const uint8_t * p )

inlinestaticprotected

◆ readSttsEntry()

static SttsEntry readSttsEntry ( const uint8_t * p )

inlinestaticprotected

◆ readU16()

static uint16_t readU16 ( const uint8_t * p )

inlinestaticprotected

◆ readU32()

static uint32_t readU32 ( const uint8_t * p )

inlinestaticprotected

◆ readU32Widened()

static uint64_t readU32Widened ( const uint8_t * p )

inlinestaticprotected

◆ readU64()

static uint64_t readU64 ( const uint8_t * p )

inlinestaticprotected

◆ removeNotifyAudioChange()

virtual bool removeNotifyAudioChange ( AudioInfoSupport & bi )

inlinevirtualinherited

Removes a target in order not to be notified about audio changes.

Reimplemented in RTSPClient< TcpClient, UdpSocket >.

◆ samplesInChunk()

uint32_t samplesInChunk	(	Track &	t,
		uint32_t	chunkIndex0
	)

inlineprotected

Number of samples in the given (0-based) chunk index, per 'stsc'. Keeps a per-track cursor rather than rescanning 'stsc' from the start every call - correct as long as chunkIndex0 is non-decreasing across calls for a given track, which holds for the sequential per-chunk consumption in advanceChunkIfNeeded()/feed().

◆ sendWavHeader()

void sendWavHeader ( )

inlineprotected

Synthesizes a valid WAV header (via the shared WAVHeader writer) from the parsed AudioInfo and sends it to p_output_audio, once, before any real audio payload - so a WAVDecoder-based output can bootstrap itself from it. Length is written as streamed/unknown, since the track's length is not known upfront while demuxing. Only ever relevant for a raw-PCM ('lpcm') audio track - not currently produced by this class (AAC/ALAC are the only codecs parsed), but kept symmetric with DemuxerAVI for when PCM tracks are added.

◆ setAudioInfo()

void setAudioInfo ( AudioInfo from )

inlineoverridevirtualinherited

for most decoders this is not needed

Implements AudioWriter.

Reimplemented in DecoderALAC, CodecChain, DSFDecoder, DecoderL8, OpusAudioDecoder, OpusMultiStreamAudioDecoder, CodecNOP, G729Decoder, and G7xxDecoder.

◆ setCodecConfig()

virtual bool setCodecConfig	(	const uint8_t *	data,
		size_t	len
	)

inlinevirtualinherited

Some decoders need e.g. a magic cookie to provide the relevant info for decoding.

Reimplemented in DecoderALAC, and MultiDecoder.

◆ setNotifyActive()

void setNotifyActive ( bool flag )

inlineinherited

Deactivate/Reactivate automatic AudioInfo updates: (default is active)

◆ setOutput() [1/3]

virtual void setOutput ( AudioOutput & out_stream )

inlinevirtualinherited

Defines where the decoded result is written to.

Reimplemented in ADTSDecoder, CodecChain, MTSDecoder, MTSDecoderTSDemux, and MetaDataFilterDecoder.

◆ setOutput() [2/3]

virtual void setOutput ( AudioStream & out_stream )

inlinevirtualinherited

Defines where the decoded result is written to.

Reimplemented in ADTSDecoder, CodecChain, MTSDecoder, MTSDecoderTSDemux, and MetaDataFilterDecoder.

◆ setOutput() [3/3]

void setOutput ( Print & out )

inlineoverridevirtual

Satisfies the AudioWriter/AudioDecoder interface - needed so EncodedAudioOutput/EncodedAudioStream's polymorphic AudioDecoder* wiring (which calls setOutput() through that base class pointer) still reaches the audio output correctly; delegates to setOutputAudio().

Reimplemented from AudioDecoder.

◆ setOutputAudio()

void setOutputAudio ( Print & out )

inlineoverridevirtual

Defines the audio output stream - e.g. an EncodedAudioStream wrapping an AudioDecoder that matches the audio track's codec, or any other Print if you want the raw payload as-is.

Implements Demuxer.

◆ setOutputVideo() [1/2]

void setOutputVideo ( Print & out )

inlineoverridevirtual

Defines the video output - e.g. a VideoOutput implementation, or any other Print if you want the raw payload as-is.

Implements Demuxer.

◆ setOutputVideo() [2/2]

void setOutputVideo ( VideoOutput & out )

inlineoverridevirtual

Defines the video output - the VideoOutput-specific counterpart of setOutputVideo(Print&), for a target like PacedVideoOutput/ OutputTinyGPU/OutputOpenCV that implements VideoOutput but not Print (the two are unrelated types - see VideoFrameMeter's own comment). Part of the base interface (unlike other concrete-class-specific extras) so code holding only a Demuxer& (e.g. VideoPlayer) can wire it to an arbitrary VideoOutput& without knowing the concrete container format - every current demuxer already implemented this identically before it was pulled up into the interface.

Implements Demuxer.

◆ setSeekSource()

void setSeekSource ( SeekableSource & seekSource )

inline

See the SeekableSource constructor overload above - can also be called any time before begin(), instead of via the constructor.

◆ setSendWavHeader()

void setSendWavHeader ( bool flag )

inlineoverrideprotectedvirtual

Overrides the automatic decision of whether a synthetic WAV header is sent to the audio output before any audio payload. By default this is decided automatically from the parsed codec: on whenever isWavFormat() of the parsed format is true (raw PCM), off otherwise (AAC/ALAC, which have their own decoder and don't expect a WAV header). Call this to force it either way instead.

Implements Demuxer.

◆ setSpoolStorageFactory()

void setSpoolStorageFactory ( SpoolStorageFactory & spoolFactory )

inline

See the SpoolStorageFactory constructor overload above - can also be called any time before begin(), instead of via the constructor.

◆ setupAudioInfo()

void setupAudioInfo	(	AudioFormat	format,
		const uint8_t *	entryData,
		size_t	entrySize
	)

inlineprotected

Populates the shared AudioInfo (channels/sample_rate/format/mime) from an AudioSampleEntry's fixed 28-byte header (ISO/IEC 14496-12 12.2.3): channelcount at rel-offset 16-17, samplerate (16.16 fixed point, upper 16 bits = Hz) at rel-offset 24-27 - the same layout for any audio sample entry (mp4a, alac, ...), so this is codec-agnostic.

◆ setupParser()

void setupParser ( )

inlineprotected

◆ videoInfo()

VideoInfo videoInfo ( )

inlineoverridevirtualinherited

Fulfills VideoInfoSource::videoInfo(), so any Demuxer can be passed directly to e.g. OutputOpenCV::setVideoInfoSource() - note the reported 'format' is the container's codec (e.g. H264), not necessarily what ends up written to a raw-picture VideoOutput/Print (that depends on whatever decoder sits between the demuxer and the output); consumers that care about the decoded format should still set it explicitly and rely on this only for width/height.

Implements VideoInfoSource.

◆ write()

size_t write	(	const uint8_t *	data,
		size_t	len
	)

inlineoverridevirtual

Partial-write contract (like AVIDecoder::write()): MP4Parser's own internal buffer has finite capacity, so a single call may accept fewer bytes than requested - the caller is expected to call write() again with the remainder, exactly like the underlying Print contract.

Implements AudioWriter.

◆ writeAdtsHeader()

static void writeAdtsHeader	(	uint8_t *	adts,
		int	aacProfile,
		int	sampleRateIdx,
		int	channelCfg,
		int	frameLen
	)

inlinestaticprotected

◆ writeBlocking()

void writeBlocking	(	Print *	out,
		uint8_t *	data,
		size_t	len
	)

inlineprotectedinherited

◆ writeVideo()

void writeVideo	(	uint8_t *	data,
		size_t	size
	)

inlineprotected

Member Data Documentation

◆ audio_format

AudioFormat audio_format = AudioFormat::UNKNOWN

protected

Codec format tag of the (first) audio track - AudioFormat::UNKNOWN until its 'stsd' has been parsed. Kept separately from AudioInfo (see AudioInfoFormat) since AudioInfo itself has no format field.

◆ box_accum

SingleBuffer<uint8_t> box_accum {0}

protected

◆ box_accum_active

bool box_accum_active = false

protected

◆ box_start_file_offset

uint64_t box_start_file_offset = 0

protected

Absolute file offset of the current box_accum-accumulating box's own header - see beginBoxAccum().

◆ current_chunk_track

Track* current_chunk_track = nullptr

protected

◆ current_sample_is_first_for_track

bool current_sample_is_first_for_track = false

protected

◆ current_sample_size

uint32_t current_sample_size = 0

protected

◆ current_sample_track

Track* current_sample_track = nullptr

protected

◆ current_track

Track* current_track

protected

Initial value:

=

nullptr

track currently being parsed (scoped by 'trak')

◆ dispatched_sample_count

uint32_t dispatched_sample_count = 0

protected

◆ id

int id

inherited

custom id to be used by application

◆ info

AudioInfo info

protectedinherited

◆ is_active

bool is_active = false

protected

◆ is_notify_active

bool is_notify_active = true

protectedinherited

◆ is_wav_header_sent

bool is_wav_header_sent = false

protected

◆ mdat_seen

bool mdat_seen = false

protected

◆ nal_tmp

Vector<uint8_t> nal_tmp

protected

scratch buffer for Annex-B conversion

◆ notify_vector

Vector<AudioInfoSupport*> notify_vector

protectedinherited

◆ p_audio_track

Track* p_audio_track = nullptr

protected

first audio track found

◆ p_output_audio

Print* p_output_audio = nullptr

protected

◆ p_print

Print* p_print = nullptr

protectedinherited

◆ p_seek_source

SeekableSource* p_seek_source = nullptr

protected

Set via setSeekSource()/the SeekableSource constructor - when set, onTrak() backs new tracks' sample tables with SourceSeekSampleTableStore instead of the RAM-backed default.

◆ p_spool_factory

SpoolStorageFactory* p_spool_factory = nullptr

protected

Set via setSpoolStorageFactory()/the SpoolStorageFactory constructor

when set, onHdlr() (once a track's kind is known) backs its sample tables with SpoolFileSampleTableStore instead of the RAM-backed default.

◆ p_video_out

Print* p_video_out = nullptr

protected

◆ p_video_out_video

VideoOutput* p_video_out_video = nullptr

protected

◆ p_video_track

Track* p_video_track = nullptr

protected

first video track found

◆ parser

MP4Parser parser

protected

◆ quickstart_bytes_remaining

int64_t quickstart_bytes_remaining = -1

protected

Set by quickStart() on success, to 'mdat's declared size. -1 (the begin() default) means no boundary active - see write().

◆ sample_buffer

Vector<uint8_t> sample_buffer

protected

◆ sample_buffer_filled

size_t sample_buffer_filled = 0

protected

◆ samples_left_in_chunk

uint32_t samples_left_in_chunk = 0

protected

◆ send_wav_header

bool send_wav_header = false

protected

◆ send_wav_header_explicit

bool send_wav_header_explicit = false

protected

◆ stsz_header_pending

bool stsz_header_pending = true

protected

◆ total_bytes_received

uint32_t total_bytes_received = 0

protected

Bytes received via write() so far this stream - MP4 has no declared total-file-size field to report instead (see getVideoInfo()).

◆ tracks

Vector<Track*> tracks

protected

The documentation for this class was generated from the following file:

src/AudioTools/AudioCodecs/ContainerMP4.h

Classes

Public Types

Public Member Functions

Public Attributes

Protected Member Functions

Static Protected Member Functions

Protected Attributes

Detailed Description

Member Typedef Documentation

◆ Codec

Constructor & Destructor Documentation

◆ DemuxerMP4() [1/5]

◆ DemuxerMP4() [2/5]

◆ DemuxerMP4() [3/5]

◆ DemuxerMP4() [4/5]

◆ DemuxerMP4() [5/5]

◆ ~DemuxerMP4()

Member Function Documentation

◆ addNotifyAudioChange()

◆ advanceChunkIfNeeded()

◆ audioALACMagicCookie()

◆ audioInfo()

◆ audioInfoOut()

◆ begin() [1/2]

◆ begin() [2/2]

◆ beginBoxAccum()

◆ boxPayloadFileOffset()

◆ clearNotifyAudioChange()

◆ configureSpoolStores()

◆ dispatchAudio()

◆ dispatchSample()

◆ dispatchVideo()

◆ end()

◆ endBoxAccum()

◆ feed()

◆ flushVideo()

◆ freeTracks()

◆ getAudioInfo()

◆ getOutput()

◆ getVideoInfo()

◆ isNotifyActive()

◆ isValid()

◆ mime()

◆ mimeVideo()

◆ notifyAudioChange()

◆ onAc3()

◆ onAlac()

◆ onAvc1()

◆ onAvcC()

◆ onEsds()

◆ onHdlr()

◆ onHevc()

◆ onMdat()

◆ onMdhd()

◆ onMp4a()

◆ onStco()

◆ onStsc()

◆ onStsd()

◆ onStsz()

◆ onStts()

◆ onTrak()

◆ operator bool()

◆ quickStart()

◆ readStscEntry()

◆ readSttsEntry()

◆ readU16()

◆ readU32()

◆ readU32Widened()

◆ readU64()

◆ removeNotifyAudioChange()

◆ samplesInChunk()

◆ sendWavHeader()

◆ setAudioInfo()

◆ setCodecConfig()

◆ setNotifyActive()

◆ setOutput() [1/3]

◆ setOutput() [2/3]

◆ setOutput() [3/3]

◆ setOutputAudio()

◆ setOutputVideo() [1/2]