gstreamer/subprojects/gst-plugins-good/gst/isomp4/qtdemux.h

568 lines
17 KiB
C

/* GStreamer
* Copyright (C) <1999> Erik Walthinsen <omega@cse.ogi.edu>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
#ifndef __GST_QTDEMUX_H__
#define __GST_QTDEMUX_H__
#include <gst/gst.h>
#include <gst/base/gstadapter.h>
#include <gst/base/gstflowcombiner.h>
#include <gst/base/gstbytereader.h>
#include <gst/video/video.h>
#include "gstisoff.h"
G_BEGIN_DECLS
#define GST_TYPE_QTDEMUX \
(gst_qtdemux_get_type())
#define GST_QTDEMUX(obj) \
(G_TYPE_CHECK_INSTANCE_CAST((obj),GST_TYPE_QTDEMUX,GstQTDemux))
#define GST_QTDEMUX_CLASS(klass) \
(G_TYPE_CHECK_CLASS_CAST((klass),GST_TYPE_QTDEMUX,GstQTDemuxClass))
#define GST_IS_QTDEMUX(obj) \
(G_TYPE_CHECK_INSTANCE_TYPE((obj),GST_TYPE_QTDEMUX))
#define GST_IS_QTDEMUX_CLASS(klass) \
(G_TYPE_CHECK_CLASS_TYPE((klass),GST_TYPE_QTDEMUX))
#define GST_QTDEMUX_CAST(obj) ((GstQTDemux *)(obj))
/* qtdemux produces these for atoms it cannot parse */
#define GST_QT_DEMUX_PRIVATE_TAG "private-qt-tag"
#define GST_QT_DEMUX_CLASSIFICATION_TAG "classification"
typedef struct _GstQTDemux GstQTDemux;
typedef struct _GstQTDemuxClass GstQTDemuxClass;
typedef struct _QtDemuxStream QtDemuxStream;
typedef struct _QtDemuxSample QtDemuxSample;
typedef struct _QtDemuxSegment QtDemuxSegment;
typedef struct _QtDemuxRandomAccessEntry QtDemuxRandomAccessEntry;
typedef struct _QtDemuxStreamStsdEntry QtDemuxStreamStsdEntry;
typedef struct _QtDemuxGaplessAudioInfo QtDemuxGaplessAudioInfo;
typedef GstBuffer * (*QtDemuxProcessFunc)(GstQTDemux * qtdemux, QtDemuxStream * stream, GstBuffer * buf);
enum QtDemuxState
{
QTDEMUX_STATE_INITIAL, /* Initial state (haven't got the header yet) */
QTDEMUX_STATE_HEADER, /* Parsing the header */
QTDEMUX_STATE_MOVIE, /* Parsing/Playing the media data */
QTDEMUX_STATE_BUFFER_MDAT /* Buffering the mdat atom */
};
typedef enum {
/* Regular behaviour */
VARIANT_NONE,
/* We're working with a MediaSource Extensions ISO BMFF Bytestream. */
VARIANT_MSE_BYTESTREAM,
/* We're working with a smoothstreaming fragment.
* Mss doesn't have 'moov' or any information about the streams format,
* requiring qtdemux to expose and create the streams */
VARIANT_MSS_FRAGMENTED,
} Variant;
typedef enum {
/* No valid gapless audio info present. Types other than this one
* are used only if all of these apply:
*
* 1. There is embedded gapless audio information available
* 2. Only one stream exists
* 3. Said stream has only one segment
* 4. Said stream is an audio stream
*/
GAPLESS_AUDIO_INFO_TYPE_NONE,
/* Using information from the iTunes iTunSMPB revdns tag. */
GAPLESS_AUDIO_INFO_TYPE_ITUNES,
/* Using known Nero encoder delay information. */
GAPLESS_AUDIO_INFO_TYPE_NERO
} QtDemuxGaplessAudioInfoType;
/* Gapless audio information, only used for single-stream audio-only media. */
struct _QtDemuxGaplessAudioInfo {
QtDemuxGaplessAudioInfoType type;
guint64 num_start_padding_pcm_frames;
guint64 num_end_padding_pcm_frames;
guint64 num_valid_pcm_frames;
/* PCM frame amounts converted to nanoseconds. */
GstClockTime start_padding_duration;
GstClockTime end_padding_duration;
GstClockTime valid_duration;
};
struct _GstQTDemux {
GstElement element;
/* Global state */
enum QtDemuxState state;
/* static sink pad */
GstPad *sinkpad;
/* TRUE if pull-based */
gboolean pullbased;
gchar *redirect_location;
/* Protect pad exposing from flush event */
GMutex expose_lock;
/* list of QtDemuxStream */
GPtrArray *active_streams;
GPtrArray *old_streams;
gint n_video_streams;
gint n_audio_streams;
gint n_sub_streams;
gint n_meta_streams;
GstFlowCombiner *flowcombiner;
/* Incoming stream group-id to set on downstream STREAM_START events.
* If upstream doesn't contain one, a global one will be generated */
gboolean have_group_id;
guint group_id;
guint major_brand;
GstBuffer *comp_brands;
/* [moov] header.
* FIXME : This is discarded just after it's created. Just move it
* to a temporary variable ? */
GNode *moov_node;
/* FIXME : This is never freed. It is only assigned once. memleak ? */
GNode *moov_node_compressed;
/* Set to TRUE when the [moov] header has been fully parsed */
gboolean got_moov;
/* Global timescale for the incoming stream. Use the QTTIME macros
* to convert values to/from GstClockTime */
guint32 timescale;
/* Global duration (in global timescale). Use QTTIME macros to get GstClockTime */
guint64 duration;
/* Start UTC time as extracted from the AFIdentification box, reset on every
* moov */
GstClockTime start_utc_time;
/* Total size of header atoms. Used to calculate fallback overall bitrate */
guint header_size;
GstTagList *tag_list;
/* configured playback region */
GstSegment segment;
/* State for key_units trickmode */
GstClockTime trickmode_interval;
/* PUSH-BASED only: If the initial segment event, or a segment consequence of
* a seek or incoming TIME segment from upstream needs to be pushed. This
* variable is used instead of pushing the event directly because at that
* point we may not have yet emitted the srcpads. */
gboolean need_segment;
guint32 segment_seqnum;
Variant variant;
/* Set to TRUE if the incoming stream is either a MSS stream or
* a Fragmented MP4 (containing the [mvex] atom in the header) */
gboolean fragmented;
/* PULL-BASED only : If TRUE there is a pending seek */
gboolean fragmented_seek_pending;
/* PULL-BASED : offset of first [moof] or of fragment to seek to
* PUSH-BASED : offset of latest [moof] */
guint64 moof_offset;
/* MSS streams have a single media that is unspecified at the atoms, so
* upstream provides it at the caps */
GstCaps *media_caps;
/* Set to TRUE when all streams have been exposed */
gboolean exposed;
gint64 chapters_track_id;
QtDemuxGaplessAudioInfo gapless_audio_info;
/* protection support */
GPtrArray *protection_system_ids; /* Holds identifiers of all content protection systems for all tracks */
GQueue protection_event_queue; /* holds copy of upstream protection events */
guint64 cenc_aux_info_offset;
guint8 *cenc_aux_info_sizes;
guint32 cenc_aux_sample_count;
gchar *preferred_protection_system_id;
/* Whether the parent bin is streams-aware, meaning we can
* add/remove streams at any point in time */
gboolean streams_aware;
/*
* ALL VARIABLES BELOW ARE ONLY USED IN PUSH-BASED MODE
*/
GstAdapter *adapter;
guint neededbytes;
guint todrop;
/* Used to store data if [mdat] is before the headers */
GstBuffer *mdatbuffer;
/* Amount of bytes left to read in the current [mdat] */
guint64 mdatleft, mdatsize;
/* When restoring the mdat to the adapter, this buffer stores any
* trailing data that was after the last atom parsed as it has to be
* restored later along with the correct offset. Used in fragmented
* scenario where mdat/moof are one after the other in any order.
*
* Check https://bugzilla.gnome.org/show_bug.cgi?id=710623 */
GstBuffer *restoredata_buffer;
guint64 restoredata_offset;
/* The current offset in bytes from upstream.
* Note: While it makes complete sense when we are PULL-BASED (pulling
* in BYTES from upstream) and PUSH-BASED with a BYTE SEGMENT (receiving
* buffers with actual offsets), it is undefined in PUSH-BASED with a
* TIME SEGMENT */
guint64 offset;
/* offset of the mdat atom */
guint64 mdatoffset;
/* Offset of the first mdat */
guint64 first_mdat;
/* offset of last [moov] seen */
guint64 last_moov_offset;
/* If TRUE, qtdemux received upstream newsegment in TIME format
* which likely means that upstream is driving the pipeline (such as
* adaptive demuxers or dlna sources) */
gboolean upstream_format_is_time;
/* Seqnum of the seek event sent upstream. Will be used to
* detect incoming FLUSH events corresponding to that */
guint32 offset_seek_seqnum;
/* UPSTREAM BYTE: Requested upstream byte seek offset.
* Currently it is only used to check if an incoming BYTE SEGMENT
* corresponds to a seek event that was sent upstream */
gint64 seek_offset;
/* UPSTREAM BYTE: Requested start/stop TIME values from
* downstream.
* Used to set on the downstream segment once the corresponding upstream
* BYTE SEEK has succeeded */
gint64 push_seek_start;
gint64 push_seek_stop;
#if 0
/* gst index support */
GstIndex *element_index;
gint index_id;
#endif
/* Whether upstream is seekable in BYTES */
gboolean upstream_seekable;
/* UPSTREAM BYTE: Size of upstream content.
* Note : This is only computed once ! If upstream grows in the meantime
* it will not be updated */
gint64 upstream_size;
/* UPSTREAM TIME : Contains the PTS (if any) of the
* buffer that contains a [moof] header. Will be used to establish
* the actual PTS of the samples contained within that fragment. */
guint64 fragment_start;
/* UPSTREAM TIME : The offset in bytes of the [moof]
* header start.
* Note : This is not computed from the GST_BUFFER_OFFSET field */
guint64 fragment_start_offset;
/* These two fields are used to perform an implicit seek when a fragmented
* file whose first tfdt is not zero. This way if the first fragment starts
* at 1 hour, the user does not have to wait 1 hour or perform a manual seek
* for the image to move and the sound to play.
*
* This implicit seek is only done if the first parsed fragment has a non-zero
* decode base time and a seek has not been received previously, hence these
* fields. */
gboolean received_seek;
gboolean first_moof_already_parsed;
};
struct _GstQTDemuxClass {
GstElementClass parent_class;
};
GType gst_qtdemux_get_type (void);
struct _QtDemuxStreamStsdEntry
{
GstCaps *caps;
guint32 fourcc;
gboolean sparse;
/* video info */
gint width;
gint height;
gint par_w;
gint par_h;
/* Numerator/denominator framerate */
gint fps_n;
gint fps_d;
GstVideoColorimetry colorimetry;
guint16 bits_per_sample;
guint16 color_table_id;
GstMemory *rgb8_palette;
guint interlace_mode;
guint field_order;
/* audio info */
gdouble rate;
gint n_channels;
guint samples_per_packet;
guint samples_per_frame;
guint bytes_per_packet;
guint bytes_per_sample;
guint bytes_per_frame;
guint compression;
/* if we use chunks or samples */
gboolean sampled;
guint padding;
};
struct _QtDemuxSample
{
guint32 size;
gint32 pts_offset; /* Add this value to timestamp to get the pts */
guint64 offset;
guint64 timestamp; /* DTS In mov time */
guint32 duration; /* In mov time */
gboolean keyframe; /* TRUE when this packet is a keyframe */
};
struct _QtDemuxStream
{
GstPad *pad;
GstQTDemux *demux;
gchar *stream_id;
QtDemuxStreamStsdEntry *stsd_entries;
guint stsd_entries_length;
guint cur_stsd_entry_index;
/* stream type */
guint32 subtype;
gboolean new_caps; /* If TRUE, caps need to be generated (by
* calling _configure_stream()) This happens
* for MSS and fragmented streams */
gboolean new_stream; /* signals that a stream_start is required */
gboolean on_keyframe; /* if this stream last pushed buffer was a
* keyframe. This is important to identify
* where to stop pushing buffers after a
* segment stop time */
/* if the stream has a redirect URI in its headers, we store it here */
gchar *redirect_uri;
/* track id */
guint track_id;
/* duration/scale */
guint64 duration; /* in timescale units */
guint32 timescale;
/* language */
gchar lang_id[4]; /* ISO 639-2T language code */
/* our samples */
guint32 n_samples;
QtDemuxSample *samples;
gboolean all_keyframe; /* TRUE when all samples are keyframes (no stss) */
guint32 n_samples_moof; /* sample count in a moof */
guint64 duration_moof; /* duration in timescale of a moof, used for figure out
* the framerate of fragmented format stream */
guint64 duration_last_moof;
guint32 offset_in_sample; /* Offset in the current sample, used for
* streams which have got exceedingly big
* sample size (such as 24s of raw audio).
* Only used when max_buffer_size is non-NULL */
guint32 min_buffer_size; /* Minimum allowed size for output buffers.
* Currently only set for raw audio streams*/
guint32 max_buffer_size; /* Maximum allowed size for output buffers.
* Currently only set for raw audio streams*/
/* video info */
/* aspect ratio */
gint display_width;
gint display_height;
/* allocation */
gboolean use_allocator;
GstAllocator *allocator;
GstAllocationParams params;
gsize alignment;
/* when a discontinuity is pending */
gboolean discont;
/* list of buffers to push first */
GSList *buffers;
/* if we need to clip this buffer. This is only needed for uncompressed
* data */
gboolean need_clip;
/* If the buffer needs some custom processing, e.g. subtitles, pass them
* through this function */
QtDemuxProcessFunc process_func;
/* buffer needs potentially be split, e.g. CEA608 subtitles */
gboolean need_split;
/* current position */
guint32 segment_index;
guint32 sample_index;
GstClockTime time_position; /* in gst time */
guint64 accumulated_base;
/* the Gst segment we are processing out, used for clipping */
GstSegment segment;
/* quicktime segments */
guint32 n_segments;
QtDemuxSegment *segments;
gboolean dummy_segment;
guint32 from_sample;
guint32 to_sample;
gboolean sent_eos;
GstTagList *stream_tags;
gboolean send_global_tags;
GstEvent *pending_event;
GstByteReader stco;
GstByteReader stsz;
GstByteReader stsc;
GstByteReader stts;
GstByteReader stss;
GstByteReader stps;
GstByteReader ctts;
gboolean chunks_are_samples; /* TRUE means treat chunks as samples */
gint64 stbl_index;
/* stco */
guint co_size;
GstByteReader co_chunk;
guint32 first_chunk;
guint32 current_chunk;
guint32 last_chunk;
guint32 samples_per_chunk;
guint32 stsd_sample_description_id;
guint32 stco_sample_index;
/* stsz */
guint32 sample_size; /* 0 means variable sizes are stored in stsz */
/* stsc */
guint32 stsc_index;
guint32 n_samples_per_chunk;
guint32 stsc_chunk_index;
guint32 stsc_sample_index;
guint64 chunk_offset;
/* stts */
guint32 stts_index;
guint32 stts_samples;
guint32 n_sample_times;
guint32 stts_sample_index;
guint64 stts_time;
guint32 stts_duration;
/* stss */
gboolean stss_present;
guint32 n_sample_syncs;
guint32 stss_index;
/* stps */
gboolean stps_present;
guint32 n_sample_partial_syncs;
guint32 stps_index;
QtDemuxRandomAccessEntry *ra_entries;
guint n_ra_entries;
const QtDemuxRandomAccessEntry *pending_seek;
/* ctts */
gboolean ctts_present;
guint32 n_composition_times;
guint32 ctts_index;
guint32 ctts_sample_index;
guint32 ctts_count;
gint32 ctts_soffset;
/* cslg composition_to_dts_shift or based on the smallest negative
* composition time offset.
*
* This is unsigned because only negative composition time offsets /
* positive composition_to_dts_shift matter here. In all other cases,
* DTS/PTS can be inferred directly without ending up with PTS>DTS.
*
* See 14496-12 6.4
*/
guint64 cslg_shift;
/* fragmented */
gboolean parsed_trex;
guint32 def_sample_description_index; /* index is 1-based */
guint32 def_sample_duration;
guint32 def_sample_size;
guint32 def_sample_flags;
gboolean disabled;
/* stereoscopic video streams */
GstVideoMultiviewMode multiview_mode;
GstVideoMultiviewFlags multiview_flags;
/* protected streams */
gboolean protected;
guint32 protection_scheme_type;
guint32 protection_scheme_version;
gpointer protection_scheme_info; /* specific to the protection scheme */
GQueue protection_scheme_event_queue;
/* KEY_UNITS trickmode with an interval */
GstClockTime last_keyframe_dts;
gint ref_count; /* atomic */
};
G_END_DECLS
#endif /* __GST_QTDEMUX_H__ */