gstreamer/subprojects/gst-plugins-bad/gst/videoparsers/gsth264parse.c

3796 lines
124 KiB
C
Raw Normal View History

/* GStreamer H.264 Parser
* Copyright (C) <2010> Collabora ltd
* Copyright (C) <2010> Nokia Corporation
* Copyright (C) <2011> Intel Corporation
*
* Copyright (C) <2010> Mark Nauwelaerts <mark.nauwelaerts@collabora.co.uk>
* Copyright (C) <2011> Thibault Saunier <thibault.saunier@collabora.com>
*
* 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.
*/
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
2013-12-16 09:18:23 +00:00
#include <gst/base/base.h>
#include <gst/pbutils/pbutils.h>
#include <gst/video/video.h>
#include "gstvideoparserselements.h"
#include "gsth264parse.h"
#include <string.h>
GST_DEBUG_CATEGORY (h264_parse_debug);
#define GST_CAT_DEFAULT h264_parse_debug
#define DEFAULT_CONFIG_INTERVAL (0)
#define DEFAULT_UPDATE_TIMECODE FALSE
enum
{
PROP_0,
PROP_CONFIG_INTERVAL,
PROP_UPDATE_TIMECODE,
};
enum
{
GST_H264_PARSE_FORMAT_NONE,
GST_H264_PARSE_FORMAT_AVC,
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
GST_H264_PARSE_FORMAT_BYTE,
GST_H264_PARSE_FORMAT_AVC3
};
enum
{
GST_H264_PARSE_ALIGN_NONE = 0,
GST_H264_PARSE_ALIGN_NAL,
GST_H264_PARSE_ALIGN_AU
};
enum
{
GST_H264_PARSE_STATE_GOT_SPS = 1 << 0,
GST_H264_PARSE_STATE_GOT_PPS = 1 << 1,
GST_H264_PARSE_STATE_GOT_SLICE = 1 << 2,
GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS = (GST_H264_PARSE_STATE_GOT_SPS |
GST_H264_PARSE_STATE_GOT_PPS),
GST_H264_PARSE_STATE_VALID_PICTURE =
(GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS |
GST_H264_PARSE_STATE_GOT_SLICE)
};
enum
{
GST_H264_PARSE_SEI_EXPIRED = 0,
GST_H264_PARSE_SEI_ACTIVE = 1,
GST_H264_PARSE_SEI_PARSED = 2,
};
#define GST_H264_PARSE_STATE_VALID(parse, expected_state) \
(((parse)->state & (expected_state)) == (expected_state))
static GstStaticPadTemplate sinktemplate = GST_STATIC_PAD_TEMPLATE ("sink",
GST_PAD_SINK,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("video/x-h264"));
static GstStaticPadTemplate srctemplate = GST_STATIC_PAD_TEMPLATE ("src",
GST_PAD_SRC,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("video/x-h264, parsed = (boolean) true, "
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
"stream-format=(string) { avc, avc3, byte-stream }, "
"alignment=(string) { au, nal }"));
#define parent_class gst_h264_parse_parent_class
G_DEFINE_TYPE (GstH264Parse, gst_h264_parse, GST_TYPE_BASE_PARSE);
GST_ELEMENT_REGISTER_DEFINE_WITH_CODE (h264parse, "h264parse",
GST_RANK_PRIMARY + 1, GST_TYPE_H264_PARSE,
videoparsers_element_init (plugin));
static void gst_h264_parse_finalize (GObject * object);
static gboolean gst_h264_parse_start (GstBaseParse * parse);
static gboolean gst_h264_parse_stop (GstBaseParse * parse);
static GstFlowReturn gst_h264_parse_handle_frame (GstBaseParse * parse,
GstBaseParseFrame * frame, gint * skipsize);
static GstFlowReturn gst_h264_parse_parse_frame (GstBaseParse * parse,
GstBaseParseFrame * frame);
static GstFlowReturn gst_h264_parse_pre_push_frame (GstBaseParse * parse,
GstBaseParseFrame * frame);
static void gst_h264_parse_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec);
static void gst_h264_parse_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec);
static gboolean gst_h264_parse_set_caps (GstBaseParse * parse, GstCaps * caps);
static GstCaps *gst_h264_parse_get_caps (GstBaseParse * parse,
GstCaps * filter);
static gboolean gst_h264_parse_event (GstBaseParse * parse, GstEvent * event);
static gboolean gst_h264_parse_src_event (GstBaseParse * parse,
GstEvent * event);
static void gst_h264_parse_update_src_caps (GstH264Parse * h264parse,
GstCaps * caps);
static void
gst_h264_parse_class_init (GstH264ParseClass * klass)
{
GObjectClass *gobject_class = (GObjectClass *) klass;
GstBaseParseClass *parse_class = GST_BASE_PARSE_CLASS (klass);
GstElementClass *gstelement_class = GST_ELEMENT_CLASS (klass);
GST_DEBUG_CATEGORY_INIT (h264_parse_debug, "h264parse", 0, "h264 parser");
gobject_class->finalize = gst_h264_parse_finalize;
gobject_class->set_property = gst_h264_parse_set_property;
gobject_class->get_property = gst_h264_parse_get_property;
g_object_class_install_property (gobject_class, PROP_CONFIG_INTERVAL,
g_param_spec_int ("config-interval",
"SPS PPS Send Interval",
"Send SPS and PPS Insertion Interval in seconds (sprop parameter sets "
"will be multiplexed in the data stream when detected.) "
"(0 = disabled, -1 = send with every IDR frame)",
-1, 3600, DEFAULT_CONFIG_INTERVAL,
G_PARAM_READWRITE | G_PARAM_CONSTRUCT | G_PARAM_STATIC_STRINGS));
/**
* GstH264Parse:update-timecode:
*
* If the stream contains Picture Timing SEI, update their timecode values
* using upstream GstVideoTimeCodeMeta. However, if there are no Picture
* Timing SEI in bitstream, this property will not insert the SEI into the
* bitstream - it only modifies existing ones.
* Moreover, even if both GstVideoTimeCodeMeta and Picture Timing SEI
* are present, if pic_struct_present_flag of VUI is equal to zero,
* timecode values will not updated as there is not enough information
* in the stream to do so.
*
* Since: 1.18
*/
g_object_class_install_property (gobject_class, PROP_UPDATE_TIMECODE,
g_param_spec_boolean ("update-timecode",
"Update Timecode",
"Update time code values in Picture Timing SEI if GstVideoTimeCodeMeta "
"is attached to incoming buffer and also Picture Timing SEI exists "
"in the bitstream. To make this property work, SPS must contain "
"VUI and pic_struct_present_flag of VUI must be non-zero",
DEFAULT_UPDATE_TIMECODE, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
/* Override BaseParse vfuncs */
parse_class->start = GST_DEBUG_FUNCPTR (gst_h264_parse_start);
parse_class->stop = GST_DEBUG_FUNCPTR (gst_h264_parse_stop);
parse_class->handle_frame = GST_DEBUG_FUNCPTR (gst_h264_parse_handle_frame);
parse_class->pre_push_frame =
GST_DEBUG_FUNCPTR (gst_h264_parse_pre_push_frame);
parse_class->set_sink_caps = GST_DEBUG_FUNCPTR (gst_h264_parse_set_caps);
parse_class->get_sink_caps = GST_DEBUG_FUNCPTR (gst_h264_parse_get_caps);
parse_class->sink_event = GST_DEBUG_FUNCPTR (gst_h264_parse_event);
parse_class->src_event = GST_DEBUG_FUNCPTR (gst_h264_parse_src_event);
gst_element_class_add_static_pad_template (gstelement_class, &srctemplate);
gst_element_class_add_static_pad_template (gstelement_class, &sinktemplate);
gst_element_class_set_static_metadata (gstelement_class, "H.264 parser",
"Codec/Parser/Converter/Video",
"Parses H.264 streams",
"Mark Nauwelaerts <mark.nauwelaerts@collabora.co.uk>");
}
static void
gst_h264_parse_init (GstH264Parse * h264parse)
{
h264parse->frame_out = gst_adapter_new ();
gst_base_parse_set_pts_interpolation (GST_BASE_PARSE (h264parse), FALSE);
gst_base_parse_set_infer_ts (GST_BASE_PARSE (h264parse), FALSE);
GST_PAD_SET_ACCEPT_INTERSECT (GST_BASE_PARSE_SINK_PAD (h264parse));
GST_PAD_SET_ACCEPT_TEMPLATE (GST_BASE_PARSE_SINK_PAD (h264parse));
h264parse->aud_needed = TRUE;
h264parse->aud_insert = TRUE;
h264parse->update_timecode = DEFAULT_UPDATE_TIMECODE;
}
static void
gst_h264_parse_finalize (GObject * object)
{
GstH264Parse *h264parse = GST_H264_PARSE (object);
g_object_unref (h264parse->frame_out);
2011-03-22 12:18:03 +00:00
G_OBJECT_CLASS (parent_class)->finalize (object);
}
static void
gst_h264_parse_reset_frame (GstH264Parse * h264parse)
{
GST_DEBUG_OBJECT (h264parse, "reset frame");
/* done parsing; reset state */
h264parse->current_off = -1;
h264parse->update_caps = FALSE;
h264parse->idr_pos = -1;
h264parse->sei_pos = -1;
h264parse->pic_timing_sei_pos = -1;
h264parse->pic_timing_sei_size = -1;
h264parse->keyframe = FALSE;
h264parse->predicted = FALSE;
h264parse->bidirectional = FALSE;
h264parse->header = FALSE;
h264parse->frame_start = FALSE;
h264parse->have_sps_in_frame = FALSE;
h264parse->have_pps_in_frame = FALSE;
gst_adapter_clear (h264parse->frame_out);
}
static void
gst_h264_parse_reset_stream_info (GstH264Parse * h264parse)
{
gint i;
h264parse->width = 0;
h264parse->height = 0;
h264parse->fps_num = 0;
h264parse->fps_den = 0;
h264parse->upstream_par_n = -1;
h264parse->upstream_par_d = -1;
h264parse->parsed_par_n = 0;
h264parse->parsed_par_d = 0;
h264parse->parsed_colorimetry.range = GST_VIDEO_COLOR_RANGE_UNKNOWN;
h264parse->parsed_colorimetry.matrix = GST_VIDEO_COLOR_MATRIX_UNKNOWN;
h264parse->parsed_colorimetry.transfer = GST_VIDEO_TRANSFER_UNKNOWN;
h264parse->parsed_colorimetry.primaries = GST_VIDEO_COLOR_PRIMARIES_UNKNOWN;
h264parse->have_pps = FALSE;
h264parse->have_sps = FALSE;
h264parse->multiview_mode = GST_VIDEO_MULTIVIEW_MODE_NONE;
h264parse->multiview_flags = GST_VIDEO_MULTIVIEW_FLAGS_NONE;
h264parse->first_in_bundle = TRUE;
h264parse->align = GST_H264_PARSE_ALIGN_NONE;
h264parse->format = GST_H264_PARSE_FORMAT_NONE;
h264parse->transform = FALSE;
h264parse->nal_length_size = 4;
h264parse->packetized = FALSE;
h264parse->push_codec = FALSE;
h264parse->first_frame = TRUE;
gst_buffer_replace (&h264parse->codec_data, NULL);
gst_buffer_replace (&h264parse->codec_data_in, NULL);
gst_h264_parse_reset_frame (h264parse);
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++)
gst_buffer_replace (&h264parse->sps_nals[i], NULL);
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++)
gst_buffer_replace (&h264parse->pps_nals[i], NULL);
gst_video_mastering_display_info_init (&h264parse->mastering_display_info);
h264parse->mastering_display_info_state = GST_H264_PARSE_SEI_EXPIRED;
gst_video_content_light_level_init (&h264parse->content_light_level);
h264parse->content_light_level_state = GST_H264_PARSE_SEI_EXPIRED;
}
static void
gst_h264_parse_reset (GstH264Parse * h264parse)
{
h264parse->last_report = GST_CLOCK_TIME_NONE;
h264parse->dts = GST_CLOCK_TIME_NONE;
h264parse->ts_trn_nb = GST_CLOCK_TIME_NONE;
h264parse->do_ts = TRUE;
h264parse->pending_key_unit_ts = GST_CLOCK_TIME_NONE;
gst_event_replace (&h264parse->force_key_unit_event, NULL);
h264parse->discont = FALSE;
h264parse->discard_bidirectional = FALSE;
2018-09-28 20:49:52 +00:00
h264parse->marker = FALSE;
gst_h264_parse_reset_stream_info (h264parse);
}
static gboolean
gst_h264_parse_start (GstBaseParse * parse)
{
GstH264Parse *h264parse = GST_H264_PARSE (parse);
GST_DEBUG_OBJECT (parse, "start");
gst_h264_parse_reset (h264parse);
h264parse->nalparser = gst_h264_nal_parser_new ();
h264parse->state = 0;
h264parse->dts = GST_CLOCK_TIME_NONE;
h264parse->ts_trn_nb = GST_CLOCK_TIME_NONE;
h264parse->sei_pic_struct_pres_flag = FALSE;
h264parse->sei_pic_struct = 0;
h264parse->field_pic_flag = 0;
h264parse->aud_needed = TRUE;
h264parse->aud_insert = FALSE;
gst_base_parse_set_min_frame_size (parse, 4);
return TRUE;
}
static gboolean
gst_h264_parse_stop (GstBaseParse * parse)
{
GstH264Parse *h264parse = GST_H264_PARSE (parse);
GST_DEBUG_OBJECT (parse, "stop");
gst_h264_parse_reset (h264parse);
gst_h264_nal_parser_free (h264parse->nalparser);
return TRUE;
}
static const gchar *
gst_h264_parse_get_string (GstH264Parse * parse, gboolean format, gint code)
{
if (format) {
switch (code) {
case GST_H264_PARSE_FORMAT_AVC:
return "avc";
case GST_H264_PARSE_FORMAT_BYTE:
return "byte-stream";
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
case GST_H264_PARSE_FORMAT_AVC3:
return "avc3";
default:
return "none";
}
} else {
switch (code) {
case GST_H264_PARSE_ALIGN_NAL:
return "nal";
case GST_H264_PARSE_ALIGN_AU:
return "au";
default:
return "none";
}
}
}
static void
gst_h264_parse_format_from_caps (GstCaps * caps, guint * format, guint * align)
{
if (format)
*format = GST_H264_PARSE_FORMAT_NONE;
if (align)
*align = GST_H264_PARSE_ALIGN_NONE;
g_return_if_fail (gst_caps_is_fixed (caps));
GST_DEBUG ("parsing caps: %" GST_PTR_FORMAT, caps);
if (caps && gst_caps_get_size (caps) > 0) {
GstStructure *s = gst_caps_get_structure (caps, 0);
const gchar *str = NULL;
if (format) {
if ((str = gst_structure_get_string (s, "stream-format"))) {
if (strcmp (str, "avc") == 0)
*format = GST_H264_PARSE_FORMAT_AVC;
else if (strcmp (str, "byte-stream") == 0)
*format = GST_H264_PARSE_FORMAT_BYTE;
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
else if (strcmp (str, "avc3") == 0)
*format = GST_H264_PARSE_FORMAT_AVC3;
}
}
if (align) {
if ((str = gst_structure_get_string (s, "alignment"))) {
if (strcmp (str, "au") == 0)
*align = GST_H264_PARSE_ALIGN_AU;
else if (strcmp (str, "nal") == 0)
*align = GST_H264_PARSE_ALIGN_NAL;
}
}
}
}
/* check downstream caps to configure format and alignment */
static void
gst_h264_parse_negotiate (GstH264Parse * h264parse, gint in_format,
GstCaps * in_caps)
{
GstCaps *caps;
guint format = h264parse->format;
guint align = h264parse->align;
g_return_if_fail ((in_caps == NULL) || gst_caps_is_fixed (in_caps));
caps = gst_pad_get_allowed_caps (GST_BASE_PARSE_SRC_PAD (h264parse));
GST_DEBUG_OBJECT (h264parse, "allowed caps: %" GST_PTR_FORMAT, caps);
/* concentrate on leading structure, since decodebin parser
* capsfilter always includes parser template caps */
if (caps) {
2012-03-11 18:06:59 +00:00
caps = gst_caps_truncate (caps);
GST_DEBUG_OBJECT (h264parse, "negotiating with caps: %" GST_PTR_FORMAT,
caps);
}
h264parse->can_passthrough = FALSE;
if (in_caps && caps) {
if (gst_caps_can_intersect (in_caps, caps)) {
GST_DEBUG_OBJECT (h264parse, "downstream accepts upstream caps");
gst_h264_parse_format_from_caps (in_caps, &format, &align);
gst_caps_unref (caps);
caps = NULL;
h264parse->can_passthrough = TRUE;
}
}
2019-09-02 19:08:44 +00:00
/* FIXME We could fail the negotiation immediately if caps are empty */
if (caps && !gst_caps_is_empty (caps)) {
/* fixate to avoid ambiguity with lists when parsing */
2012-03-11 18:06:59 +00:00
caps = gst_caps_fixate (caps);
gst_h264_parse_format_from_caps (caps, &format, &align);
}
/* default */
if (!format)
format = GST_H264_PARSE_FORMAT_BYTE;
if (!align)
align = GST_H264_PARSE_ALIGN_AU;
GST_DEBUG_OBJECT (h264parse, "selected format %s, alignment %s",
gst_h264_parse_get_string (h264parse, TRUE, format),
gst_h264_parse_get_string (h264parse, FALSE, align));
h264parse->format = format;
h264parse->align = align;
h264parse->transform = in_format != h264parse->format ||
align == GST_H264_PARSE_ALIGN_AU;
if (caps)
gst_caps_unref (caps);
}
static GstBuffer *
gst_h264_parse_wrap_nal (GstH264Parse * h264parse, guint format, guint8 * data,
guint size)
{
GstBuffer *buf;
guint nl = h264parse->nal_length_size;
guint32 tmp;
GST_DEBUG_OBJECT (h264parse, "nal length %d", size);
buf = gst_buffer_new_allocate (NULL, 4 + size, NULL);
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
if (format == GST_H264_PARSE_FORMAT_AVC
|| format == GST_H264_PARSE_FORMAT_AVC3) {
tmp = GUINT32_TO_BE (size << (32 - 8 * nl));
} else {
/* HACK: nl should always be 4 here, otherwise this won't work.
* There are legit cases where nl in avc stream is 2, but byte-stream
* SC is still always 4 bytes. */
nl = 4;
tmp = GUINT32_TO_BE (1);
}
gst_buffer_fill (buf, 0, &tmp, sizeof (guint32));
gst_buffer_fill (buf, nl, data, size);
gst_buffer_set_size (buf, size + nl);
return buf;
}
static void
gst_h264_parser_store_nal (GstH264Parse * h264parse, guint id,
GstH264NalUnitType naltype, GstH264NalUnit * nalu)
{
GstBuffer *buf, **store;
guint size = nalu->size, store_size;
if (naltype == GST_H264_NAL_SPS || naltype == GST_H264_NAL_SUBSET_SPS) {
store_size = GST_H264_MAX_SPS_COUNT;
store = h264parse->sps_nals;
GST_DEBUG_OBJECT (h264parse, "storing sps %u", id);
} else if (naltype == GST_H264_NAL_PPS) {
store_size = GST_H264_MAX_PPS_COUNT;
store = h264parse->pps_nals;
GST_DEBUG_OBJECT (h264parse, "storing pps %u", id);
} else
return;
if (id >= store_size) {
GST_DEBUG_OBJECT (h264parse, "unable to store nal, id out-of-range %d", id);
return;
}
2012-03-15 12:37:36 +00:00
buf = gst_buffer_new_allocate (NULL, size, NULL);
gst_buffer_fill (buf, 0, nalu->data + nalu->offset, size);
/* Indicate that buffer contain a header needed for decoding */
if (naltype == GST_H264_NAL_SPS || naltype == GST_H264_NAL_PPS)
GST_BUFFER_FLAG_SET (buf, GST_BUFFER_FLAG_HEADER);
if (store[id])
gst_buffer_unref (store[id]);
store[id] = buf;
}
#ifndef GST_DISABLE_GST_DEBUG
static const gchar *nal_names[] = {
"Unknown",
"Slice",
"Slice DPA",
"Slice DPB",
"Slice DPC",
"Slice IDR",
"SEI",
"SPS",
"PPS",
"AU delimiter",
"Sequence End",
"Stream End",
"Filler Data",
"SPS extension",
"Prefix",
"SPS Subset",
"Depth Parameter Set",
"Reserved", "Reserved",
"Slice Aux Unpartitioned",
"Slice Extension",
"Slice Depth/3D-AVC Extension"
};
static const gchar *
_nal_name (GstH264NalUnitType nal_type)
{
if (nal_type <= GST_H264_NAL_SLICE_DEPTH)
return nal_names[nal_type];
return "Invalid";
}
#endif
static void
gst_h264_parse_process_sei_user_data (GstH264Parse * h264parse,
GstH264RegisteredUserData * rud)
{
guint16 provider_code;
GstByteReader br;
GstVideoParseUtilsField field = GST_VIDEO_PARSE_UTILS_FIELD_1;
/* only US country code is currently supported */
switch (rud->country_code) {
case ITU_T_T35_COUNTRY_CODE_US:
break;
default:
GST_LOG_OBJECT (h264parse, "Unsupported country code %d",
rud->country_code);
return;
}
if (rud->data == NULL || rud->size < 2)
return;
gst_byte_reader_init (&br, rud->data, rud->size);
provider_code = gst_byte_reader_get_uint16_be_unchecked (&br);
if (h264parse->sei_pic_struct ==
(guint8) GST_H264_SEI_PIC_STRUCT_BOTTOM_FIELD)
field = GST_VIDEO_PARSE_UTILS_FIELD_2;
gst_video_parse_user_data ((GstElement *) h264parse, &h264parse->user_data,
&br, field, provider_code);
}
static void
gst_h264_parse_process_sei (GstH264Parse * h264parse, GstH264NalUnit * nalu)
{
GstH264SEIMessage sei;
GstH264NalParser *nalparser = h264parse->nalparser;
GstH264ParserResult pres;
GArray *messages;
guint i;
pres = gst_h264_parser_parse_sei (nalparser, nalu, &messages);
if (pres != GST_H264_PARSER_OK)
2015-06-20 15:07:57 +00:00
GST_WARNING_OBJECT (h264parse, "failed to parse one or more SEI message");
/* Even if pres != GST_H264_PARSER_OK, some message could have been parsed and
* stored in messages.
*/
for (i = 0; i < messages->len; i++) {
sei = g_array_index (messages, GstH264SEIMessage, i);
switch (sei.payloadType) {
case GST_H264_SEI_PIC_TIMING:
{
guint j;
h264parse->sei_pic_struct_pres_flag =
sei.payload.pic_timing.pic_struct_present_flag;
h264parse->sei_cpb_removal_delay =
sei.payload.pic_timing.cpb_removal_delay;
if (h264parse->sei_pic_struct_pres_flag) {
h264parse->sei_pic_struct = sei.payload.pic_timing.pic_struct;
}
h264parse->num_clock_timestamp = 0;
memcpy (&h264parse->pic_timing_sei, &sei.payload.pic_timing,
sizeof (GstH264PicTiming));
for (j = 0; j < 3; j++) {
if (sei.payload.pic_timing.clock_timestamp_flag[j]) {
h264parse->num_clock_timestamp++;
}
}
if (h264parse->sei_pic_struct_pres_flag && h264parse->update_timecode) {
/* FIXME: add support multiple messages in a SEI nalu.
* Updating only this SEI message and preserving the others
* is a bit complicated */
if (messages->len == 1) {
h264parse->pic_timing_sei_pos = nalu->sc_offset;
h264parse->pic_timing_sei_size =
nalu->size + (nalu->offset - nalu->sc_offset);
}
}
GST_LOG_OBJECT (h264parse, "pic timing updated");
break;
}
case GST_H264_SEI_REGISTERED_USER_DATA:
gst_h264_parse_process_sei_user_data (h264parse,
&sei.payload.registered_user_data);
break;
case GST_H264_SEI_BUF_PERIOD:
if (h264parse->ts_trn_nb == GST_CLOCK_TIME_NONE ||
h264parse->dts == GST_CLOCK_TIME_NONE)
h264parse->ts_trn_nb = 0;
else
h264parse->ts_trn_nb = h264parse->dts;
GST_LOG_OBJECT (h264parse,
"new buffering period; ts_trn_nb updated: %" GST_TIME_FORMAT,
GST_TIME_ARGS (h264parse->ts_trn_nb));
break;
/* Additional messages that are not innerly useful to the
* element but for debugging purposes */
case GST_H264_SEI_RECOVERY_POINT:
GST_LOG_OBJECT (h264parse, "recovery point found: %u %u %u %u",
sei.payload.recovery_point.recovery_frame_cnt,
sei.payload.recovery_point.exact_match_flag,
sei.payload.recovery_point.broken_link_flag,
sei.payload.recovery_point.changing_slice_group_idc);
h264parse->keyframe = TRUE;
break;
/* Additional messages that are not innerly useful to the
* element but for debugging purposes */
case GST_H264_SEI_STEREO_VIDEO_INFO:{
GstVideoMultiviewMode mview_mode = GST_VIDEO_MULTIVIEW_MODE_NONE;
GstVideoMultiviewFlags mview_flags = GST_VIDEO_MULTIVIEW_FLAGS_NONE;
GST_LOG_OBJECT (h264parse, "Stereo video information %u %u %u %u %u %u",
sei.payload.stereo_video_info.field_views_flag,
sei.payload.stereo_video_info.top_field_is_left_view_flag,
sei.payload.stereo_video_info.current_frame_is_left_view_flag,
sei.payload.stereo_video_info.next_frame_is_second_view_flag,
sei.payload.stereo_video_info.left_view_self_contained_flag,
sei.payload.stereo_video_info.right_view_self_contained_flag);
if (sei.payload.stereo_video_info.field_views_flag) {
mview_mode = GST_VIDEO_MULTIVIEW_MODE_ROW_INTERLEAVED;
if (!sei.payload.stereo_video_info.top_field_is_left_view_flag)
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST;
} else {
mview_mode = GST_VIDEO_MULTIVIEW_MODE_FRAME_BY_FRAME;
if (sei.payload.stereo_video_info.next_frame_is_second_view_flag) {
/* Mark current frame as first in bundle */
h264parse->first_in_bundle = TRUE;
if (!sei.payload.stereo_video_info.current_frame_is_left_view_flag)
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST;
}
}
if (mview_mode != h264parse->multiview_mode ||
mview_flags != h264parse->multiview_flags) {
h264parse->multiview_mode = mview_mode;
h264parse->multiview_flags = mview_flags;
/* output caps need to be changed */
gst_h264_parse_update_src_caps (h264parse, NULL);
}
break;
}
case GST_H264_SEI_FRAME_PACKING:{
GstVideoMultiviewMode mview_mode = GST_VIDEO_MULTIVIEW_MODE_NONE;
GstVideoMultiviewFlags mview_flags = GST_VIDEO_MULTIVIEW_FLAGS_NONE;
GST_LOG_OBJECT (h264parse,
"frame packing arrangement message: id %u cancelled %u "
"type %u quincunx %u content_interpretation %d flip %u "
"right_first %u field_views %u is_frame0 %u "
"frame0_self_contained %u frame1_self_contained %u "
"frame0_grid (%u, %u) frame1_grid (%u, %u) "
"repetition_period %" G_GUINT16_FORMAT,
sei.payload.frame_packing.frame_packing_id,
sei.payload.frame_packing.frame_packing_cancel_flag,
sei.payload.frame_packing.frame_packing_type,
sei.payload.frame_packing.quincunx_sampling_flag,
sei.payload.frame_packing.content_interpretation_type,
sei.payload.frame_packing.spatial_flipping_flag,
sei.payload.frame_packing.frame0_flipped_flag,
sei.payload.frame_packing.field_views_flag,
sei.payload.frame_packing.current_frame_is_frame0_flag,
sei.payload.frame_packing.frame0_self_contained_flag,
sei.payload.frame_packing.frame1_self_contained_flag,
sei.payload.frame_packing.frame0_grid_position_x,
sei.payload.frame_packing.frame0_grid_position_y,
sei.payload.frame_packing.frame1_grid_position_x,
sei.payload.frame_packing.frame1_grid_position_y,
sei.payload.frame_packing.frame_packing_repetition_period);
/* Only IDs from 0->255 and 512->2^31-1 are valid. Ignore others */
if ((sei.payload.frame_packing.frame_packing_id >= 256 &&
sei.payload.frame_packing.frame_packing_id < 512) ||
(sei.payload.frame_packing.frame_packing_id >= (1U << 31)))
break; /* ignore */
if (!sei.payload.frame_packing.frame_packing_cancel_flag) {
/* Cancel flag sets things back to no-info */
if (sei.payload.frame_packing.content_interpretation_type == 2)
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST;
switch (sei.payload.frame_packing.frame_packing_type) {
case 0:
mview_mode = GST_VIDEO_MULTIVIEW_MODE_CHECKERBOARD;
break;
case 1:
mview_mode = GST_VIDEO_MULTIVIEW_MODE_COLUMN_INTERLEAVED;
break;
case 2:
mview_mode = GST_VIDEO_MULTIVIEW_MODE_ROW_INTERLEAVED;
break;
case 3:
if (sei.payload.frame_packing.quincunx_sampling_flag)
mview_mode = GST_VIDEO_MULTIVIEW_MODE_SIDE_BY_SIDE_QUINCUNX;
else
mview_mode = GST_VIDEO_MULTIVIEW_MODE_SIDE_BY_SIDE;
if (sei.payload.frame_packing.spatial_flipping_flag) {
/* One of the views is flopped. */
if (sei.payload.frame_packing.frame0_flipped_flag !=
! !(mview_flags &
GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST))
/* the left view is flopped */
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_LEFT_FLOPPED;
else
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_FLOPPED;
}
break;
case 4:
mview_mode = GST_VIDEO_MULTIVIEW_MODE_TOP_BOTTOM;
if (sei.payload.frame_packing.spatial_flipping_flag) {
/* One of the views is flipped, */
if (sei.payload.frame_packing.frame0_flipped_flag !=
! !(mview_flags &
GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST))
/* the left view is flipped */
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_LEFT_FLIPPED;
else
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_FLIPPED;
}
break;
case 5:
if (sei.payload.frame_packing.content_interpretation_type == 0)
mview_mode = GST_VIDEO_MULTIVIEW_MODE_MULTIVIEW_FRAME_BY_FRAME;
else
mview_mode = GST_VIDEO_MULTIVIEW_MODE_FRAME_BY_FRAME;
break;
default:
GST_DEBUG_OBJECT (h264parse, "Invalid frame packing type %u",
sei.payload.frame_packing.frame_packing_type);
break;
}
}
if (mview_mode != h264parse->multiview_mode ||
mview_flags != h264parse->multiview_flags) {
h264parse->multiview_mode = mview_mode;
h264parse->multiview_flags = mview_flags;
/* output caps need to be changed */
gst_h264_parse_update_src_caps (h264parse, NULL);
}
break;
}
case GST_H264_SEI_MASTERING_DISPLAY_COLOUR_VOLUME:
{
/* Precision defined by spec.
* See D.2.29 Mastering display colour volume SEI message semantics */
GstVideoMasteringDisplayInfo minfo;
gint j, k;
/* GstVideoMasteringDisplayInfo::display_primaries is rgb order but
* AVC uses gbr order.
* See spec D.2.29 display_primaries_x and display_primaries_y
*/
for (j = 0, k = 2; j < G_N_ELEMENTS (minfo.display_primaries); j++, k++) {
minfo.display_primaries[j].x =
sei.payload.
mastering_display_colour_volume.display_primaries_x[k % 3];
minfo.display_primaries[j].y =
sei.payload.
mastering_display_colour_volume.display_primaries_y[k % 3];
}
minfo.white_point.x =
sei.payload.mastering_display_colour_volume.white_point_x;
minfo.white_point.y =
sei.payload.mastering_display_colour_volume.white_point_y;
minfo.max_display_mastering_luminance =
sei.payload.mastering_display_colour_volume.
max_display_mastering_luminance;
minfo.min_display_mastering_luminance =
sei.payload.mastering_display_colour_volume.
min_display_mastering_luminance;
GST_LOG_OBJECT (h264parse, "mastering display info found: "
"Red(%u, %u) "
"Green(%u, %u) "
"Blue(%u, %u) "
"White(%u, %u) "
"max_luminance(%u) "
"min_luminance(%u) ",
minfo.display_primaries[0].x, minfo.display_primaries[0].y,
minfo.display_primaries[1].x, minfo.display_primaries[1].y,
minfo.display_primaries[2].x, minfo.display_primaries[2].y,
minfo.white_point.x, minfo.white_point.y,
minfo.max_display_mastering_luminance,
minfo.min_display_mastering_luminance);
if (h264parse->mastering_display_info_state ==
GST_H264_PARSE_SEI_EXPIRED) {
h264parse->update_caps = TRUE;
} else if (!gst_video_mastering_display_info_is_equal
(&h264parse->mastering_display_info, &minfo)) {
h264parse->update_caps = TRUE;
}
h264parse->mastering_display_info_state = GST_H264_PARSE_SEI_PARSED;
h264parse->mastering_display_info = minfo;
break;
}
case GST_H264_SEI_CONTENT_LIGHT_LEVEL:
{
GstVideoContentLightLevel cll;
cll.max_content_light_level =
sei.payload.content_light_level.max_content_light_level;
cll.max_frame_average_light_level =
sei.payload.content_light_level.max_pic_average_light_level;
GST_LOG_OBJECT (h264parse, "content light level found: "
"maxCLL:(%u), maxFALL:(%u)", cll.max_content_light_level,
cll.max_frame_average_light_level);
if (h264parse->content_light_level_state == GST_H264_PARSE_SEI_EXPIRED) {
h264parse->update_caps = TRUE;
} else if (cll.max_content_light_level !=
h264parse->content_light_level.max_content_light_level ||
cll.max_frame_average_light_level !=
h264parse->content_light_level.max_frame_average_light_level) {
h264parse->update_caps = TRUE;
}
h264parse->content_light_level_state = GST_H264_PARSE_SEI_PARSED;
h264parse->content_light_level = cll;
break;
}
default:{
gint payload_type = sei.payloadType;
if (payload_type == GST_H264_SEI_UNHANDLED_PAYLOAD) {
GstH264SEIUnhandledPayload *unhandled =
&sei.payload.unhandled_payload;
payload_type = unhandled->payloadType;
}
GST_LOG_OBJECT (h264parse, "Unsupported payload type %d", payload_type);
break;
}
}
}
g_array_free (messages, TRUE);
}
/* caller guarantees 2 bytes of nal payload */
static gboolean
gst_h264_parse_process_nal (GstH264Parse * h264parse, GstH264NalUnit * nalu)
{
guint nal_type;
GstH264PPS pps = { 0, };
GstH264SPS sps = { 0, };
GstH264NalParser *nalparser = h264parse->nalparser;
GstH264ParserResult pres;
GstH264SliceHdr slice;
/* nothing to do for broken input */
if (G_UNLIKELY (nalu->size < 2)) {
GST_DEBUG_OBJECT (h264parse, "not processing nal size %u", nalu->size);
return TRUE;
}
/* we have a peek as well */
nal_type = nalu->type;
GST_DEBUG_OBJECT (h264parse, "processing nal of type %u %s, size %u",
nal_type, _nal_name (nal_type), nalu->size);
switch (nal_type) {
case GST_H264_NAL_SUBSET_SPS:
if (!GST_H264_PARSE_STATE_VALID (h264parse, GST_H264_PARSE_STATE_GOT_SPS))
return FALSE;
pres = gst_h264_parser_parse_subset_sps (nalparser, nalu, &sps);
goto process_sps;
case GST_H264_NAL_SPS:
/* reset state, everything else is obsolete */
h264parse->state &= GST_H264_PARSE_STATE_GOT_PPS;
pres = gst_h264_parser_parse_sps (nalparser, nalu, &sps);
process_sps:
/* arranged for a fallback sps.id, so use that one and only warn */
if (pres != GST_H264_PARSER_OK) {
GST_WARNING_OBJECT (h264parse, "failed to parse SPS:");
h264parse->state |= GST_H264_PARSE_STATE_GOT_SPS;
h264parse->header = TRUE;
return FALSE;
}
GST_DEBUG_OBJECT (h264parse, "triggering src caps check");
h264parse->update_caps = TRUE;
h264parse->have_sps = TRUE;
h264parse->have_sps_in_frame = TRUE;
if (h264parse->push_codec && h264parse->have_pps) {
/* SPS and PPS found in stream before the first pre_push_frame, no need
* to forcibly push at start */
GST_INFO_OBJECT (h264parse, "have SPS/PPS in stream");
h264parse->push_codec = FALSE;
h264parse->have_sps = FALSE;
h264parse->have_pps = FALSE;
}
gst_h264_parser_store_nal (h264parse, sps.id, nal_type, nalu);
gst_h264_sps_clear (&sps);
h264parse->state |= GST_H264_PARSE_STATE_GOT_SPS;
h264parse->header = TRUE;
break;
case GST_H264_NAL_PPS:
/* expected state: got-sps */
h264parse->state &= GST_H264_PARSE_STATE_GOT_SPS;
if (!GST_H264_PARSE_STATE_VALID (h264parse, GST_H264_PARSE_STATE_GOT_SPS))
return FALSE;
pres = gst_h264_parser_parse_pps (nalparser, nalu, &pps);
/* arranged for a fallback pps.id, so use that one and only warn */
if (pres != GST_H264_PARSER_OK) {
GST_WARNING_OBJECT (h264parse, "failed to parse PPS:");
if (pres != GST_H264_PARSER_BROKEN_LINK)
return FALSE;
}
/* parameters might have changed, force caps check */
if (!h264parse->have_pps) {
GST_DEBUG_OBJECT (h264parse, "triggering src caps check");
h264parse->update_caps = TRUE;
}
h264parse->have_pps = TRUE;
h264parse->have_pps_in_frame = TRUE;
if (h264parse->push_codec && h264parse->have_sps) {
/* SPS and PPS found in stream before the first pre_push_frame, no need
* to forcibly push at start */
GST_INFO_OBJECT (h264parse, "have SPS/PPS in stream");
h264parse->push_codec = FALSE;
h264parse->have_sps = FALSE;
h264parse->have_pps = FALSE;
}
gst_h264_parser_store_nal (h264parse, pps.id, nal_type, nalu);
gst_h264_pps_clear (&pps);
h264parse->state |= GST_H264_PARSE_STATE_GOT_PPS;
h264parse->header = TRUE;
break;
case GST_H264_NAL_SEI:
/* expected state: got-sps */
if (!GST_H264_PARSE_STATE_VALID (h264parse, GST_H264_PARSE_STATE_GOT_SPS))
return FALSE;
h264parse->header = TRUE;
gst_h264_parse_process_sei (h264parse, nalu);
/* mark SEI pos */
if (h264parse->sei_pos == -1) {
if (h264parse->transform)
h264parse->sei_pos = gst_adapter_available (h264parse->frame_out);
else
h264parse->sei_pos = nalu->sc_offset;
GST_DEBUG_OBJECT (h264parse, "marking SEI in frame at offset %d",
h264parse->sei_pos);
}
break;
case GST_H264_NAL_SLICE:
case GST_H264_NAL_SLICE_DPA:
case GST_H264_NAL_SLICE_DPB:
case GST_H264_NAL_SLICE_DPC:
2011-12-05 12:09:17 +00:00
case GST_H264_NAL_SLICE_IDR:
case GST_H264_NAL_SLICE_EXT:
/* expected state: got-sps|got-pps (valid picture headers) */
h264parse->state &= GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS;
if (!GST_H264_PARSE_STATE_VALID (h264parse,
GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS))
return FALSE;
/* This is similar to the GOT_SLICE state, but is only reset when the
* AU is complete. This is used to keep track of AU */
h264parse->picture_start = TRUE;
/* don't need to parse the whole slice (header) here */
if (*(nalu->data + nalu->offset + nalu->header_bytes) & 0x80) {
2011-12-05 12:09:17 +00:00
/* means first_mb_in_slice == 0 */
/* real frame data */
GST_DEBUG_OBJECT (h264parse, "first_mb_in_slice = 0");
h264parse->frame_start = TRUE;
}
2011-12-05 12:09:17 +00:00
GST_DEBUG_OBJECT (h264parse, "frame start: %i", h264parse->frame_start);
if (nal_type == GST_H264_NAL_SLICE_EXT && !GST_H264_IS_MVC_NALU (nalu))
break;
pres = gst_h264_parser_parse_slice_hdr (nalparser, nalu, &slice,
FALSE, FALSE);
GST_DEBUG_OBJECT (h264parse,
"parse result %d, first MB: %u, slice type: %u",
pres, slice.first_mb_in_slice, slice.type);
if (pres == GST_H264_PARSER_OK) {
if (GST_H264_IS_I_SLICE (&slice) || GST_H264_IS_SI_SLICE (&slice))
h264parse->keyframe = TRUE;
else if (GST_H264_IS_P_SLICE (&slice)
|| GST_H264_IS_SP_SLICE (&slice))
h264parse->predicted = TRUE;
else if (GST_H264_IS_B_SLICE (&slice))
h264parse->bidirectional = TRUE;
h264parse->state |= GST_H264_PARSE_STATE_GOT_SLICE;
h264parse->field_pic_flag = slice.field_pic_flag;
2011-12-05 12:09:17 +00:00
}
2011-12-05 12:09:17 +00:00
if (G_LIKELY (nal_type != GST_H264_NAL_SLICE_IDR &&
!h264parse->push_codec))
break;
/* if we need to sneak codec NALs into the stream,
* this is a good place, so fake it as IDR
* (which should be at start anyway) */
/* mark where config needs to go if interval expired */
/* mind replacement buffer if applicable */
if (h264parse->idr_pos == -1) {
if (h264parse->transform)
h264parse->idr_pos = gst_adapter_available (h264parse->frame_out);
else
h264parse->idr_pos = nalu->sc_offset;
GST_DEBUG_OBJECT (h264parse, "marking IDR in frame at offset %d",
h264parse->idr_pos);
}
2019-09-02 19:08:44 +00:00
/* if SEI precedes (faked) IDR, then we have to insert config there */
if (h264parse->sei_pos >= 0 && h264parse->idr_pos > h264parse->sei_pos) {
h264parse->idr_pos = h264parse->sei_pos;
GST_DEBUG_OBJECT (h264parse, "moved IDR mark to SEI position %d",
h264parse->idr_pos);
}
/* Reset state only on first IDR slice of CVS D.2.29 */
if (slice.first_mb_in_slice == 0) {
if (h264parse->mastering_display_info_state ==
GST_H264_PARSE_SEI_PARSED)
h264parse->mastering_display_info_state = GST_H264_PARSE_SEI_ACTIVE;
else if (h264parse->mastering_display_info_state ==
GST_H264_PARSE_SEI_ACTIVE)
h264parse->mastering_display_info_state = GST_H264_PARSE_SEI_EXPIRED;
if (h264parse->content_light_level_state == GST_H264_PARSE_SEI_PARSED)
h264parse->content_light_level_state = GST_H264_PARSE_SEI_ACTIVE;
else if (h264parse->content_light_level_state ==
GST_H264_PARSE_SEI_ACTIVE)
h264parse->content_light_level_state = GST_H264_PARSE_SEI_EXPIRED;
}
break;
case GST_H264_NAL_AU_DELIMITER:
/* Just accumulate AU Delimiter, whether it's before SPS or not */
pres = gst_h264_parser_parse_nal (nalparser, nalu);
if (pres != GST_H264_PARSER_OK)
return FALSE;
h264parse->aud_needed = FALSE;
break;
default:
/* drop anything before the initial SPS */
if (!GST_H264_PARSE_STATE_VALID (h264parse, GST_H264_PARSE_STATE_GOT_SPS))
return FALSE;
pres = gst_h264_parser_parse_nal (nalparser, nalu);
if (pres != GST_H264_PARSER_OK)
return FALSE;
break;
}
/* if AVC output needed, collect properly prefixed nal in adapter,
* and use that to replace outgoing buffer data later on */
if (h264parse->transform) {
GstBuffer *buf;
GST_LOG_OBJECT (h264parse, "collecting NAL in AVC frame");
buf = gst_h264_parse_wrap_nal (h264parse, h264parse->format,
nalu->data + nalu->offset, nalu->size);
gst_adapter_push (h264parse->frame_out, buf);
}
return TRUE;
}
/* caller guarantees at least 2 bytes of nal payload for each nal
* returns TRUE if next_nal indicates that nal terminates an AU */
static inline gboolean
2019-09-20 19:21:17 +00:00
gst_h264_parse_collect_nal (GstH264Parse * h264parse, GstH264NalUnit * nalu)
{
GstH264NalUnitType nal_type = nalu->type;
gboolean complete;
/* determine if AU complete */
GST_LOG_OBJECT (h264parse, "next nal type: %d %s (picture started %i)",
nal_type, _nal_name (nal_type), h264parse->picture_start);
/* consider a coded slices (IDR or not) to start a picture,
* (so ending the previous one) if first_mb_in_slice == 0
* (non-0 is part of previous one) */
/* NOTE this is not entirely according to Access Unit specs in 7.4.1.2.4,
* but in practice it works in sane cases, needs not much parsing,
* and also works with broken frame_num in NAL
* (where spec-wise would fail) */
complete = h264parse->picture_start && ((nal_type >= GST_H264_NAL_SEI &&
nal_type <= GST_H264_NAL_AU_DELIMITER) ||
(nal_type >= 14 && nal_type <= 18));
/* first_mb_in_slice == 0 considered start of frame */
if (nalu->size > nalu->header_bytes)
complete |= h264parse->picture_start && (nal_type == GST_H264_NAL_SLICE
|| nal_type == GST_H264_NAL_SLICE_DPA
|| nal_type == GST_H264_NAL_SLICE_IDR) &&
(nalu->data[nalu->offset + nalu->header_bytes] & 0x80);
GST_LOG_OBJECT (h264parse, "au complete: %d", complete);
if (complete)
h264parse->picture_start = FALSE;
return complete;
}
static guint8 au_delim[6] = {
0x00, 0x00, 0x00, 0x01, /* nal prefix */
0x09, /* nal unit type = access unit delimiter */
0xf0 /* allow any slice type */
};
static GstFlowReturn
gst_h264_parse_handle_frame_packetized (GstBaseParse * parse,
GstBaseParseFrame * frame)
{
GstH264Parse *h264parse = GST_H264_PARSE (parse);
GstBuffer *buffer = frame->buffer;
GstFlowReturn ret = GST_FLOW_OK;
GstH264ParserResult parse_res;
GstH264NalUnit nalu;
const guint nl = h264parse->nal_length_size;
GstMapInfo map;
gint left;
if (nl < 1 || nl > 4) {
GST_DEBUG_OBJECT (h264parse, "insufficient data to split input");
return GST_FLOW_NOT_NEGOTIATED;
}
/* need to save buffer from invalidation upon _finish_frame */
if (h264parse->split_packetized)
buffer = gst_buffer_copy (frame->buffer);
gst_buffer_map (buffer, &map, GST_MAP_READ);
left = map.size;
GST_LOG_OBJECT (h264parse,
"processing packet buffer of size %" G_GSIZE_FORMAT, map.size);
parse_res = gst_h264_parser_identify_nalu_avc (h264parse->nalparser,
map.data, 0, map.size, nl, &nalu);
/* there is no AUD in AVC, always enable insertion, the pre_push function
* will only add it once, and will only add it for byte-stream output. */
h264parse->aud_insert = TRUE;
while (parse_res == GST_H264_PARSER_OK) {
GST_DEBUG_OBJECT (h264parse, "AVC nal offset %d", nalu.offset + nalu.size);
/* either way, have a look at it */
gst_h264_parse_process_nal (h264parse, &nalu);
/* dispatch per NALU if needed */
if (h264parse->split_packetized) {
GstBaseParseFrame tmp_frame;
gst_base_parse_frame_init (&tmp_frame);
tmp_frame.flags |= frame->flags;
tmp_frame.offset = frame->offset;
tmp_frame.overhead = frame->overhead;
tmp_frame.buffer = gst_buffer_copy_region (buffer, GST_BUFFER_COPY_ALL,
nalu.offset, nalu.size);
2018-09-28 20:49:52 +00:00
/* Set marker on last packet */
if (nl + nalu.size == left) {
if (GST_BUFFER_FLAG_IS_SET (frame->buffer, GST_BUFFER_FLAG_MARKER))
h264parse->marker = TRUE;
}
/* note we don't need to come up with a sub-buffer, since
* subsequent code only considers input buffer's metadata.
* Real data is either taken from input by baseclass or
* a replacement output buffer is provided anyway. */
gst_h264_parse_parse_frame (parse, &tmp_frame);
ret = gst_base_parse_finish_frame (parse, &tmp_frame, nl + nalu.size);
left -= nl + nalu.size;
}
parse_res = gst_h264_parser_identify_nalu_avc (h264parse->nalparser,
map.data, nalu.offset + nalu.size, map.size, nl, &nalu);
}
gst_buffer_unmap (buffer, &map);
if (!h264parse->split_packetized) {
2018-09-28 20:49:52 +00:00
h264parse->marker = TRUE;
gst_h264_parse_parse_frame (parse, frame);
ret = gst_base_parse_finish_frame (parse, frame, map.size);
} else {
gst_buffer_unref (buffer);
if (G_UNLIKELY (left)) {
/* should not be happening for nice AVC */
GST_WARNING_OBJECT (parse, "skipping leftover AVC data %d", left);
frame->flags |= GST_BASE_PARSE_FRAME_FLAG_DROP;
ret = gst_base_parse_finish_frame (parse, frame, map.size);
}
}
if (parse_res == GST_H264_PARSER_NO_NAL_END ||
parse_res == GST_H264_PARSER_BROKEN_DATA) {
if (h264parse->split_packetized) {
GST_ELEMENT_ERROR (h264parse, STREAM, FAILED, (NULL),
("invalid AVC input data"));
return GST_FLOW_ERROR;
} else {
/* do not meddle to much in this case */
GST_DEBUG_OBJECT (h264parse, "parsing packet failed");
}
}
return ret;
}
static GstFlowReturn
gst_h264_parse_handle_frame (GstBaseParse * parse,
GstBaseParseFrame * frame, gint * skipsize)
{
GstH264Parse *h264parse = GST_H264_PARSE (parse);
GstBuffer *buffer = frame->buffer;
2012-01-25 15:20:41 +00:00
GstMapInfo map;
guint8 *data;
gsize size;
2012-02-13 11:26:11 +00:00
gint current_off = 0;
gboolean drain, nonext;
GstH264NalParser *nalparser = h264parse->nalparser;
GstH264NalUnit nalu;
GstH264ParserResult pres;
gint framesize;
if (G_UNLIKELY (GST_BUFFER_FLAG_IS_SET (frame->buffer,
GST_BUFFER_FLAG_DISCONT))) {
h264parse->discont = TRUE;
}
/* delegate in packetized case, no skipping should be needed */
if (h264parse->packetized)
return gst_h264_parse_handle_frame_packetized (parse, frame);
2012-01-25 15:20:41 +00:00
gst_buffer_map (buffer, &map, GST_MAP_READ);
data = map.data;
size = map.size;
/* expect at least 3 bytes start_code, and 1 bytes NALU header.
* the length of the NALU payload can be zero.
* (e.g. EOS/EOB placed at the end of an AU.) */
if (G_UNLIKELY (size < 4)) {
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (buffer, &map);
*skipsize = 1;
return GST_FLOW_OK;
}
/* need to configure aggregation */
if (G_UNLIKELY (h264parse->format == GST_H264_PARSE_FORMAT_NONE))
gst_h264_parse_negotiate (h264parse, GST_H264_PARSE_FORMAT_BYTE, NULL);
2011-05-18 07:47:43 +00:00
/* avoid stale cached parsing state */
if (frame->flags & GST_BASE_PARSE_FRAME_FLAG_NEW_FRAME) {
2011-05-18 07:47:43 +00:00
GST_LOG_OBJECT (h264parse, "parsing new frame");
gst_h264_parse_reset_frame (h264parse);
} else {
GST_LOG_OBJECT (h264parse, "resuming frame parsing");
}
/* Always consume the entire input buffer when in_align == ALIGN_AU */
drain = GST_BASE_PARSE_DRAINING (parse)
|| h264parse->in_align == GST_H264_PARSE_ALIGN_AU;
nonext = FALSE;
current_off = h264parse->current_off;
if (current_off < 0)
current_off = 0;
/* The parser is being drain, but no new data was added, just prentend this
* AU is complete */
if (drain && current_off == size) {
GST_DEBUG_OBJECT (h264parse, "draining with no new data");
nalu.size = 0;
nalu.offset = current_off;
goto end;
}
g_assert (current_off < size);
GST_DEBUG_OBJECT (h264parse, "last parse position %d", current_off);
/* check for initial skip */
if (h264parse->current_off == -1) {
pres =
gst_h264_parser_identify_nalu_unchecked (nalparser, data, current_off,
size, &nalu);
switch (pres) {
case GST_H264_PARSER_OK:
if (nalu.sc_offset > 0) {
*skipsize = nalu.sc_offset;
goto skip;
}
break;
case GST_H264_PARSER_NO_NAL:
/* Start code may have up to 4 bytes */
*skipsize = size - 4;
goto skip;
break;
default:
/* should not really occur either */
GST_ELEMENT_ERROR (h264parse, STREAM, FORMAT,
("Error parsing H.264 stream"), ("Invalid H.264 stream"));
goto invalid_stream;
}
/* Ensure we use the TS of the first NAL. This avoids broken timestamp in
* the case of a miss-placed filler byte. */
gst_base_parse_set_ts_at_offset (parse, nalu.offset);
}
while (TRUE) {
pres =
gst_h264_parser_identify_nalu (nalparser, data, current_off, size,
&nalu);
switch (pres) {
case GST_H264_PARSER_OK:
GST_DEBUG_OBJECT (h264parse, "complete nal (offset, size): (%u, %u) ",
nalu.offset, nalu.size);
break;
case GST_H264_PARSER_NO_NAL_END:
/* In NAL alignment, assume the NAL is complete */
if (h264parse->in_align == GST_H264_PARSE_ALIGN_NAL ||
h264parse->in_align == GST_H264_PARSE_ALIGN_AU) {
nonext = TRUE;
nalu.size = size - nalu.offset;
break;
}
GST_DEBUG_OBJECT (h264parse, "not a complete nal found at offset %u",
nalu.offset);
/* if draining, accept it as complete nal */
if (drain) {
nonext = TRUE;
nalu.size = size - nalu.offset;
GST_DEBUG_OBJECT (h264parse, "draining, accepting with size %u",
nalu.size);
/* if it's not too short at least */
if (nalu.size < 2)
goto broken;
break;
}
/* otherwise need more */
goto more;
case GST_H264_PARSER_BROKEN_LINK:
GST_ELEMENT_ERROR (h264parse, STREAM, FORMAT,
("Error parsing H.264 stream"),
("The link to structure needed for the parsing couldn't be found"));
goto invalid_stream;
case GST_H264_PARSER_ERROR:
/* should not really occur either */
GST_ELEMENT_ERROR (h264parse, STREAM, FORMAT,
("Error parsing H.264 stream"), ("Invalid H.264 stream"));
goto invalid_stream;
case GST_H264_PARSER_NO_NAL:
GST_ELEMENT_ERROR (h264parse, STREAM, FORMAT,
("Error parsing H.264 stream"), ("No H.264 NAL unit found"));
goto invalid_stream;
case GST_H264_PARSER_BROKEN_DATA:
GST_WARNING_OBJECT (h264parse, "input stream is corrupt; "
"it contains a NAL unit of length %u", nalu.size);
broken:
/* broken nal at start -> arrange to skip it,
* otherwise have it terminate current au
* (and so it will be skipped on next frame round) */
GST_ELEMENT_WARNING (h264parse, STREAM, DECODE,
(NULL), ("Broken bit stream"));
if (current_off == 0) {
GST_DEBUG_OBJECT (h264parse, "skipping broken nal");
*skipsize = nalu.offset;
goto skip;
} else {
GST_DEBUG_OBJECT (h264parse, "terminating au");
nalu.size = 0;
nalu.offset = nalu.sc_offset;
goto end;
}
break;
default:
g_assert_not_reached ();
break;
}
GST_DEBUG_OBJECT (h264parse, "%p complete nal found. Off: %u, Size: %u",
data, nalu.offset, nalu.size);
2019-09-20 19:21:17 +00:00
if (gst_h264_parse_collect_nal (h264parse, &nalu)) {
h264parse->aud_needed = TRUE;
/* complete current frame, if it exist */
if (current_off > 0) {
nalu.size = 0;
nalu.offset = nalu.sc_offset;
2018-09-28 20:49:52 +00:00
h264parse->marker = TRUE;
break;
}
}
if (!gst_h264_parse_process_nal (h264parse, &nalu)) {
GST_WARNING_OBJECT (h264parse,
"broken/invalid nal Type: %d %s, Size: %u will be dropped",
nalu.type, _nal_name (nalu.type), nalu.size);
*skipsize = nalu.size;
goto skip;
}
/* Make sure the next buffer will contain an AUD */
if (h264parse->aud_needed) {
h264parse->aud_insert = TRUE;
h264parse->aud_needed = FALSE;
}
/* Do not push immediately if we don't have all headers. This ensure that
* our caps are complete, avoiding a renegotiation */
if (h264parse->align == GST_H264_PARSE_ALIGN_NAL &&
!GST_H264_PARSE_STATE_VALID (h264parse,
GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS))
frame->flags |= GST_BASE_PARSE_FRAME_FLAG_QUEUE;
/* if no next nal, we reached the end of this buffer */
if (nonext) {
/* If there is a marker flag, or input is AU, we know this is complete */
if (GST_BUFFER_FLAG_IS_SET (frame->buffer, GST_BUFFER_FLAG_MARKER) ||
h264parse->in_align == GST_H264_PARSE_ALIGN_AU) {
2018-09-28 20:49:52 +00:00
h264parse->marker = TRUE;
break;
}
/* or if we are draining */
if (drain || h264parse->align == GST_H264_PARSE_ALIGN_NAL)
break;
current_off = nalu.offset + nalu.size;
goto more;
}
/* If the output is NAL, we are done */
if (h264parse->align == GST_H264_PARSE_ALIGN_NAL)
break;
GST_DEBUG_OBJECT (h264parse, "Looking for more");
current_off = nalu.offset + nalu.size;
/* expect at least 3 bytes start_code, and 1 bytes NALU header.
* the length of the NALU payload can be zero.
* (e.g. EOS/EOB placed at the end of an AU.) */
if (size - current_off < 4) {
/* Finish the frame if there is no more data in the stream */
if (drain)
break;
goto more;
}
}
end:
framesize = nalu.offset + nalu.size;
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (buffer, &map);
gst_h264_parse_parse_frame (parse, frame);
return gst_base_parse_finish_frame (parse, frame, framesize);
more:
*skipsize = 0;
/* Restart parsing from here next time */
if (current_off > 0)
h264parse->current_off = current_off;
/* Fall-through. */
out:
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (buffer, &map);
return GST_FLOW_OK;
skip:
GST_DEBUG_OBJECT (h264parse, "skipping %d", *skipsize);
/* If we are collecting access units, we need to preserve the initial
* config headers (SPS, PPS et al.) and only reset the frame if another
* slice NAL was received. This means that broken pictures are discarded */
if (h264parse->align != GST_H264_PARSE_ALIGN_AU ||
!(h264parse->state & GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS) ||
(h264parse->state & GST_H264_PARSE_STATE_GOT_SLICE))
gst_h264_parse_reset_frame (h264parse);
goto out;
invalid_stream:
gst_buffer_unmap (buffer, &map);
return GST_FLOW_ERROR;
}
/* byte together avc codec data based on collected pps and sps so far */
static GstBuffer *
gst_h264_parse_make_codec_data (GstH264Parse * h264parse)
{
GstBuffer *buf, *nal;
gint i, sps_size = 0, pps_size = 0, num_sps = 0, num_pps = 0;
guint8 profile_idc = 0, profile_comp = 0, level_idc = 0;
gboolean found = FALSE;
2012-01-25 15:20:41 +00:00
GstMapInfo map;
guint8 *data;
gint nl;
/* only nal payload in stored nals */
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++) {
if ((nal = h264parse->sps_nals[i])) {
gsize size = gst_buffer_get_size (nal);
num_sps++;
/* size bytes also count */
sps_size += size + 2;
if (size >= 4) {
guint8 tmp[3];
found = TRUE;
gst_buffer_extract (nal, 1, tmp, 3);
profile_idc = tmp[0];
profile_comp = tmp[1];
level_idc = tmp[2];
}
}
}
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++) {
if ((nal = h264parse->pps_nals[i])) {
num_pps++;
/* size bytes also count */
pps_size += gst_buffer_get_size (nal) + 2;
}
}
/* AVC3 has SPS/PPS inside the stream, not in the codec_data */
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
if (h264parse->format == GST_H264_PARSE_FORMAT_AVC3) {
num_sps = sps_size = 0;
num_pps = pps_size = 0;
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
}
GST_DEBUG_OBJECT (h264parse,
"constructing codec_data: num_sps=%d, num_pps=%d", num_sps, num_pps);
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
if (!found || (0 == num_pps
&& GST_H264_PARSE_FORMAT_AVC3 != h264parse->format))
return NULL;
2012-03-15 12:37:36 +00:00
buf = gst_buffer_new_allocate (NULL, 5 + 1 + sps_size + 1 + pps_size, NULL);
2012-01-25 15:20:41 +00:00
gst_buffer_map (buf, &map, GST_MAP_WRITE);
data = map.data;
nl = h264parse->nal_length_size;
data[0] = 1; /* AVC Decoder Configuration Record ver. 1 */
data[1] = profile_idc; /* profile_idc */
data[2] = profile_comp; /* profile_compability */
data[3] = level_idc; /* level_idc */
data[4] = 0xfc | (nl - 1); /* nal_length_size_minus1 */
data[5] = 0xe0 | num_sps; /* number of SPSs */
data += 6;
if (h264parse->format != GST_H264_PARSE_FORMAT_AVC3) {
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++) {
if ((nal = h264parse->sps_nals[i])) {
gsize nal_size = gst_buffer_get_size (nal);
GST_WRITE_UINT16_BE (data, nal_size);
gst_buffer_extract (nal, 0, data + 2, nal_size);
data += 2 + nal_size;
}
}
}
data[0] = num_pps;
data++;
if (h264parse->format != GST_H264_PARSE_FORMAT_AVC3) {
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++) {
if ((nal = h264parse->pps_nals[i])) {
gsize nal_size = gst_buffer_get_size (nal);
GST_WRITE_UINT16_BE (data, nal_size);
gst_buffer_extract (nal, 0, data + 2, nal_size);
data += 2 + nal_size;
}
}
}
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (buf, &map);
return buf;
}
static void
gst_h264_parse_get_par (GstH264Parse * h264parse, gint * num, gint * den)
{
if (h264parse->upstream_par_n != -1 && h264parse->upstream_par_d != -1) {
*num = h264parse->upstream_par_n;
*den = h264parse->upstream_par_d;
} else {
*num = h264parse->parsed_par_n;
*den = h264parse->parsed_par_d;
}
}
static GstCaps *
get_compatible_profile_caps (GstH264SPS * sps)
{
GstCaps *caps = NULL;
const gchar **profiles = NULL;
gint i;
GValue compat_profiles = G_VALUE_INIT;
g_value_init (&compat_profiles, GST_TYPE_LIST);
switch (sps->profile_idc) {
case GST_H264_PROFILE_EXTENDED:
if (sps->constraint_set0_flag) { /* A.2.1 */
if (sps->constraint_set1_flag) {
static const gchar *profile_array[] =
{ "constrained-baseline", "baseline", "main", "high",
"high-10", "high-4:2:2", "high-4:4:4", NULL
};
profiles = profile_array;
} else {
static const gchar *profile_array[] = { "baseline", NULL };
profiles = profile_array;
}
} else if (sps->constraint_set1_flag) { /* A.2.2 */
static const gchar *profile_array[] =
{ "main", "high", "high-10", "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
}
break;
case GST_H264_PROFILE_BASELINE:
if (sps->constraint_set1_flag) { /* A.2.1 */
static const gchar *profile_array[] =
{ "baseline", "main", "high", "high-10", "high-4:2:2",
"high-4:4:4", NULL
};
profiles = profile_array;
} else {
static const gchar *profile_array[] = { "extended", NULL };
profiles = profile_array;
}
break;
case GST_H264_PROFILE_MAIN:
{
static const gchar *profile_array[] =
{ "high", "high-10", "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
}
break;
case GST_H264_PROFILE_HIGH:
if (sps->constraint_set1_flag) {
static const gchar *profile_array[] =
{ "main", "high-10", "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
} else {
static const gchar *profile_array[] =
{ "high-10", "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
}
break;
case GST_H264_PROFILE_HIGH10:
if (sps->constraint_set1_flag) {
static const gchar *profile_array[] =
{ "main", "high", "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
} else {
if (sps->constraint_set3_flag) { /* A.2.8 */
static const gchar *profile_array[] =
{ "high-10", "high-4:2:2", "high-4:4:4", "high-4:2:2-intra",
"high-4:4:4-intra", NULL
};
profiles = profile_array;
} else {
static const gchar *profile_array[] =
{ "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
}
}
break;
case GST_H264_PROFILE_HIGH_422:
if (sps->constraint_set1_flag) {
static const gchar *profile_array[] =
{ "main", "high", "high-10", "high-4:4:4", NULL };
profiles = profile_array;
} else {
if (sps->constraint_set3_flag) { /* A.2.9 */
static const gchar *profile_array[] =
{ "high-4:2:2", "high-4:4:4", "high-4:2:2-intra",
"high-4:4:4-intra", NULL
};
profiles = profile_array;
} else {
static const gchar *profile_array[] =
{ "high-4:2:2", "high-4:4:4", NULL };
profiles = profile_array;
}
}
break;
case GST_H264_PROFILE_HIGH_444:
if (sps->constraint_set1_flag) {
static const gchar *profile_array[] =
{ "main", "high", "high-10", "high-4:2:2", NULL };
profiles = profile_array;
} else if (sps->constraint_set3_flag) { /* A.2.10 */
static const gchar *profile_array[] = { "high-4:4:4", NULL };
profiles = profile_array;
}
break;
case GST_H264_PROFILE_MULTIVIEW_HIGH:
if (sps->extension_type == GST_H264_NAL_EXTENSION_MVC
&& sps->extension.mvc.num_views_minus1 == 1) {
static const gchar *profile_array[] =
{ "stereo-high", "multiview-high", NULL };
profiles = profile_array;
} else {
static const gchar *profile_array[] = { "multiview-high", NULL };
profiles = profile_array;
}
break;
default:
break;
}
if (profiles) {
GValue value = G_VALUE_INIT;
caps = gst_caps_new_empty_simple ("video/x-h264");
for (i = 0; profiles[i]; i++) {
g_value_init (&value, G_TYPE_STRING);
g_value_set_string (&value, profiles[i]);
gst_value_list_append_value (&compat_profiles, &value);
g_value_unset (&value);
}
gst_caps_set_value (caps, "profile", &compat_profiles);
g_value_unset (&compat_profiles);
}
return caps;
}
/* if downstream didn't support the exact profile indicated in sps header,
* check for the compatible profiles also */
static void
ensure_caps_profile (GstH264Parse * h264parse, GstCaps * caps, GstH264SPS * sps)
{
GstCaps *peer_caps, *compat_caps;
peer_caps = gst_pad_get_current_caps (GST_BASE_PARSE_SRC_PAD (h264parse));
if (!peer_caps || !gst_caps_can_intersect (caps, peer_caps)) {
GstCaps *filter_caps = gst_caps_new_empty_simple ("video/x-h264");
if (peer_caps)
gst_caps_unref (peer_caps);
peer_caps =
gst_pad_peer_query_caps (GST_BASE_PARSE_SRC_PAD (h264parse),
filter_caps);
gst_caps_unref (filter_caps);
}
if (peer_caps && !gst_caps_can_intersect (caps, peer_caps)) {
GstStructure *structure;
compat_caps = get_compatible_profile_caps (sps);
if (compat_caps != NULL) {
GstCaps *res_caps = NULL;
res_caps = gst_caps_intersect (peer_caps, compat_caps);
if (res_caps && !gst_caps_is_empty (res_caps)) {
const gchar *profile_str = NULL;
res_caps = gst_caps_fixate (res_caps);
structure = gst_caps_get_structure (res_caps, 0);
profile_str = gst_structure_get_string (structure, "profile");
if (profile_str) {
gst_caps_set_simple (caps, "profile", G_TYPE_STRING, profile_str,
NULL);
GST_DEBUG_OBJECT (h264parse,
"Setting compatible profile %s to the caps", profile_str);
}
}
if (res_caps)
gst_caps_unref (res_caps);
gst_caps_unref (compat_caps);
}
}
if (peer_caps)
gst_caps_unref (peer_caps);
}
static const gchar *
digit_to_string (guint digit)
{
static const char itoa[][2] = {
"0", "1", "2", "3", "4", "5", "6", "7", "8", "9"
};
if (G_LIKELY (digit < 10))
return itoa[digit];
else
return NULL;
}
static const gchar *
get_profile_string (GstH264SPS * sps)
{
const gchar *profile = NULL;
switch (sps->profile_idc) {
case 66:
if (sps->constraint_set1_flag)
profile = "constrained-baseline";
else
profile = "baseline";
break;
case 77:
profile = "main";
break;
case 88:
profile = "extended";
break;
case 100:
if (sps->constraint_set4_flag) {
if (sps->constraint_set5_flag)
profile = "constrained-high";
else
profile = "progressive-high";
} else
profile = "high";
break;
case 110:
if (sps->constraint_set3_flag)
profile = "high-10-intra";
else if (sps->constraint_set4_flag)
profile = "progressive-high-10";
else
profile = "high-10";
break;
case 122:
if (sps->constraint_set3_flag)
profile = "high-4:2:2-intra";
else
profile = "high-4:2:2";
break;
case 244:
if (sps->constraint_set3_flag)
profile = "high-4:4:4-intra";
else
profile = "high-4:4:4";
break;
case 44:
profile = "cavlc-4:4:4-intra";
break;
case 118:
profile = "multiview-high";
break;
case 128:
profile = "stereo-high";
break;
case 83:
if (sps->constraint_set5_flag)
profile = "scalable-constrained-baseline";
else
profile = "scalable-baseline";
break;
case 86:
if (sps->constraint_set3_flag)
profile = "scalable-high-intra";
else if (sps->constraint_set5_flag)
profile = "scalable-constrained-high";
else
profile = "scalable-high";
break;
default:
return NULL;
}
return profile;
}
static const gchar *
get_level_string (GstH264SPS * sps)
{
if (sps->level_idc == 0)
return NULL;
else if ((sps->level_idc == 11 && sps->constraint_set3_flag)
|| sps->level_idc == 9)
return "1b";
else if (sps->level_idc % 10 == 0)
return digit_to_string (sps->level_idc / 10);
else {
switch (sps->level_idc) {
case 11:
return "1.1";
case 12:
return "1.2";
case 13:
return "1.3";
case 21:
return "2.1";
case 22:
return "2.2";
case 31:
return "3.1";
case 32:
return "3.2";
case 41:
return "4.1";
case 42:
return "4.2";
case 51:
return "5.1";
case 52:
return "5.2";
case 61:
return "6.1";
case 62:
return "6.2";
default:
return NULL;
}
}
}
static void
gst_h264_parse_update_src_caps (GstH264Parse * h264parse, GstCaps * caps)
{
GstH264SPS *sps;
GstCaps *sink_caps, *src_caps;
gboolean modified = FALSE;
GstBuffer *buf = NULL;
GstStructure *s = NULL;
if (G_UNLIKELY (!gst_pad_has_current_caps (GST_BASE_PARSE_SRC_PAD
(h264parse))))
modified = TRUE;
else if (G_UNLIKELY (!h264parse->update_caps))
return;
/* if this is being called from the first _setcaps call, caps on the sinkpad
* aren't set yet and so they need to be passed as an argument */
if (caps)
2012-02-09 13:42:23 +00:00
sink_caps = gst_caps_ref (caps);
else
sink_caps = gst_pad_get_current_caps (GST_BASE_PARSE_SINK_PAD (h264parse));
/* carry over input caps as much as possible; override with our own stuff */
if (!sink_caps)
sink_caps = gst_caps_new_empty_simple ("video/x-h264");
else
s = gst_caps_get_structure (sink_caps, 0);
sps = h264parse->nalparser->last_sps;
GST_DEBUG_OBJECT (h264parse, "sps: %p", sps);
/* only codec-data for nice-and-clean au aligned packetized avc format */
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
if ((h264parse->format == GST_H264_PARSE_FORMAT_AVC
|| h264parse->format == GST_H264_PARSE_FORMAT_AVC3)
&& h264parse->align == GST_H264_PARSE_ALIGN_AU) {
buf = gst_h264_parse_make_codec_data (h264parse);
if (buf && h264parse->codec_data) {
2012-01-25 15:20:41 +00:00
GstMapInfo map;
2012-01-25 15:20:41 +00:00
gst_buffer_map (buf, &map, GST_MAP_READ);
if (map.size != gst_buffer_get_size (h264parse->codec_data) ||
gst_buffer_memcmp (h264parse->codec_data, 0, map.data, map.size))
modified = TRUE;
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (buf, &map);
} else {
if (!buf && h264parse->codec_data_in)
buf = gst_buffer_ref (h264parse->codec_data_in);
modified = TRUE;
}
}
caps = NULL;
if (G_UNLIKELY (!sps)) {
caps = gst_caps_copy (sink_caps);
} else {
gint crop_width, crop_height;
gint fps_num, fps_den;
gint par_n, par_d;
GstH264VUIParams *vui = &sps->vui_parameters;
gchar *colorimetry = NULL;
if (sps->frame_cropping_flag) {
crop_width = sps->crop_rect_width;
crop_height = sps->crop_rect_height;
} else {
crop_width = sps->width;
crop_height = sps->height;
}
if (G_UNLIKELY (h264parse->width != crop_width ||
h264parse->height != crop_height)) {
GST_INFO_OBJECT (h264parse, "resolution changed %dx%d",
crop_width, crop_height);
h264parse->width = crop_width;
h264parse->height = crop_height;
modified = TRUE;
}
/* 0/1 is set as the default in the codec parser, we will set
* it in case we have no info */
gst_h264_video_calculate_framerate (sps, h264parse->field_pic_flag,
h264parse->sei_pic_struct, &fps_num, &fps_den);
if (G_UNLIKELY (h264parse->fps_num != fps_num
|| h264parse->fps_den != fps_den)) {
GST_DEBUG_OBJECT (h264parse, "framerate changed %d/%d", fps_num, fps_den);
h264parse->fps_num = fps_num;
h264parse->fps_den = fps_den;
modified = TRUE;
}
if (vui->aspect_ratio_info_present_flag) {
if (G_UNLIKELY ((h264parse->parsed_par_n != vui->par_n)
|| (h264parse->parsed_par_d != vui->par_d))) {
h264parse->parsed_par_n = vui->par_n;
h264parse->parsed_par_d = vui->par_d;
GST_INFO_OBJECT (h264parse, "pixel aspect ratio has been changed %d/%d",
h264parse->parsed_par_n, h264parse->parsed_par_d);
modified = TRUE;
}
}
if (vui->video_signal_type_present_flag &&
vui->colour_description_present_flag) {
GstVideoColorimetry ci = { 0, };
gchar *old_colorimetry = NULL;
if (vui->video_full_range_flag)
ci.range = GST_VIDEO_COLOR_RANGE_0_255;
else
ci.range = GST_VIDEO_COLOR_RANGE_16_235;
ci.matrix = gst_video_color_matrix_from_iso (vui->matrix_coefficients);
ci.transfer =
gst_video_transfer_function_from_iso (vui->transfer_characteristics);
ci.primaries = gst_video_color_primaries_from_iso (vui->colour_primaries);
old_colorimetry =
gst_video_colorimetry_to_string (&h264parse->parsed_colorimetry);
colorimetry = gst_video_colorimetry_to_string (&ci);
if (colorimetry && g_strcmp0 (old_colorimetry, colorimetry)) {
GST_INFO_OBJECT (h264parse,
"colorimetry has been changed from %s to %s",
GST_STR_NULL (old_colorimetry), colorimetry);
h264parse->parsed_colorimetry = ci;
modified = TRUE;
}
g_free (old_colorimetry);
}
if (G_UNLIKELY (modified || h264parse->update_caps)) {
gint width, height;
GstClockTime latency = 0;
const gchar *caps_mview_mode = NULL;
GstVideoMultiviewMode mview_mode = h264parse->multiview_mode;
GstVideoMultiviewFlags mview_flags = h264parse->multiview_flags;
const gchar *chroma_format = NULL;
guint bit_depth_chroma;
fps_num = h264parse->fps_num;
fps_den = h264parse->fps_den;
caps = gst_caps_copy (sink_caps);
/* sps should give this but upstream overrides */
if (s && gst_structure_has_field (s, "width"))
gst_structure_get_int (s, "width", &width);
else
width = h264parse->width;
if (s && gst_structure_has_field (s, "height"))
gst_structure_get_int (s, "height", &height);
else
height = h264parse->height;
if (s == NULL ||
!gst_structure_get_fraction (s, "pixel-aspect-ratio", &par_n,
&par_d)) {
gst_h264_parse_get_par (h264parse, &par_n, &par_d);
if (par_n != 0 && par_d != 0) {
GST_INFO_OBJECT (h264parse, "PAR %d/%d", par_n, par_d);
gst_caps_set_simple (caps, "pixel-aspect-ratio", GST_TYPE_FRACTION,
par_n, par_d, NULL);
} else {
/* Assume par_n/par_d of 1/1 for calcs below, but don't set into caps */
par_n = par_d = 1;
}
}
/* Pass through or set output stereo/multiview config */
if (s && gst_structure_has_field (s, "multiview-mode")) {
caps_mview_mode = gst_structure_get_string (s, "multiview-mode");
2016-11-26 10:16:39 +00:00
gst_structure_get_flagset (s, "multiview-flags",
(guint *) & mview_flags, NULL);
} else if (mview_mode != GST_VIDEO_MULTIVIEW_MODE_NONE) {
if (gst_video_multiview_guess_half_aspect (mview_mode,
width, height, par_n, par_d)) {
mview_flags |= GST_VIDEO_MULTIVIEW_FLAGS_HALF_ASPECT;
}
caps_mview_mode = gst_video_multiview_mode_to_caps_string (mview_mode);
gst_caps_set_simple (caps, "multiview-mode", G_TYPE_STRING,
caps_mview_mode, "multiview-flags",
GST_TYPE_VIDEO_MULTIVIEW_FLAGSET, mview_flags,
GST_FLAG_SET_MASK_EXACT, NULL);
}
gst_caps_set_simple (caps, "width", G_TYPE_INT, width,
"height", G_TYPE_INT, height, NULL);
/* upstream overrides */
if (s && gst_structure_has_field (s, "framerate")) {
gst_structure_get_fraction (s, "framerate", &fps_num, &fps_den);
}
/* but not necessarily or reliably this */
if (fps_den > 0) {
GstStructure *s2;
gst_caps_set_simple (caps, "framerate",
GST_TYPE_FRACTION, fps_num, fps_den, NULL);
s2 = gst_caps_get_structure (caps, 0);
gst_structure_get_fraction (s2, "framerate", &h264parse->parsed_fps_n,
&h264parse->parsed_fps_d);
/* If we know the frame duration, and if we are not in one of the zero
* latency pattern, add one frame of latency */
if (fps_num > 0 && h264parse->in_align != GST_H264_PARSE_ALIGN_AU &&
!(h264parse->in_align == GST_H264_PARSE_ALIGN_NAL &&
h264parse->align == GST_H264_PARSE_ALIGN_NAL)) {
latency = gst_util_uint64_scale (GST_SECOND, fps_den, fps_num);
}
gst_base_parse_set_latency (GST_BASE_PARSE (h264parse), latency,
latency);
}
bit_depth_chroma = sps->bit_depth_chroma_minus8 + 8;
switch (sps->chroma_format_idc) {
case 0:
chroma_format = "4:0:0";
bit_depth_chroma = 0;
break;
case 1:
chroma_format = "4:2:0";
break;
case 2:
chroma_format = "4:2:2";
break;
case 3:
chroma_format = "4:4:4";
break;
default:
break;
}
if (chroma_format)
gst_caps_set_simple (caps,
"chroma-format", G_TYPE_STRING, chroma_format,
"bit-depth-luma", G_TYPE_UINT, sps->bit_depth_luma_minus8 + 8,
"bit-depth-chroma", G_TYPE_UINT, bit_depth_chroma, NULL);
if (colorimetry && (!s || !gst_structure_has_field (s, "colorimetry"))) {
gst_caps_set_simple (caps, "colorimetry", G_TYPE_STRING, colorimetry,
NULL);
}
}
g_free (colorimetry);
}
if (caps) {
const gchar *mdi_str = NULL;
const gchar *cll_str = NULL;
gboolean codec_data_modified = FALSE;
gst_caps_set_simple (caps, "parsed", G_TYPE_BOOLEAN, TRUE,
"stream-format", G_TYPE_STRING,
gst_h264_parse_get_string (h264parse, TRUE, h264parse->format),
"alignment", G_TYPE_STRING,
gst_h264_parse_get_string (h264parse, FALSE, h264parse->align), NULL);
/* set profile and level in caps */
if (sps) {
const gchar *profile, *level;
profile = get_profile_string (sps);
if (profile != NULL)
gst_caps_set_simple (caps, "profile", G_TYPE_STRING, profile, NULL);
level = get_level_string (sps);
if (level != NULL)
gst_caps_set_simple (caps, "level", G_TYPE_STRING, level, NULL);
/* relax the profile constraint to find a suitable decoder */
ensure_caps_profile (h264parse, caps, sps);
}
if (s)
mdi_str = gst_structure_get_string (s, "mastering-display-info");
if (mdi_str) {
gst_caps_set_simple (caps, "mastering-display-info", G_TYPE_STRING,
mdi_str, NULL);
} else if (h264parse->mastering_display_info_state !=
GST_H264_PARSE_SEI_EXPIRED &&
!gst_video_mastering_display_info_add_to_caps
(&h264parse->mastering_display_info, caps)) {
GST_WARNING_OBJECT (h264parse,
"Couldn't set mastering display info to caps");
}
if (s)
cll_str = gst_structure_get_string (s, "content-light-level");
if (cll_str) {
gst_caps_set_simple (caps, "content-light-level", G_TYPE_STRING, cll_str,
NULL);
} else if (h264parse->content_light_level_state !=
GST_H264_PARSE_SEI_EXPIRED &&
!gst_video_content_light_level_add_to_caps
(&h264parse->content_light_level, caps)) {
GST_WARNING_OBJECT (h264parse,
"Couldn't set content light level to caps");
}
src_caps = gst_pad_get_current_caps (GST_BASE_PARSE_SRC_PAD (h264parse));
if (src_caps) {
GstStructure *src_caps_str = gst_caps_get_structure (src_caps, 0);
/* use codec data from old caps for comparison if we have pushed frame for now.
* we don't want to resend caps if everything is same except codec data.
* However, if the updated sps/pps is not in bitstream, we should put
* it on bitstream */
if (gst_structure_has_field (src_caps_str, "codec_data")) {
const GValue *codec_data_value =
gst_structure_get_value (src_caps_str, "codec_data");
if (!GST_VALUE_HOLDS_BUFFER (codec_data_value)) {
GST_WARNING_OBJECT (h264parse, "codec_data does not hold buffer");
} else if (!h264parse->first_frame) {
/* If there is no pushed frame before, we can update caps without worry.
* But updating codec_data in the middle of frames
* (especially on non-keyframe) might make downstream be confused.
* Therefore we are setting old codec data
* (i.e., was pushed to downstream previously) to new caps candidate
* here for gst_caps_is_strictly_equal() to be returned TRUE if only
* the codec_data is different, and to avoid re-sending caps it
* that case.
*/
gst_caps_set_value (caps, "codec_data", codec_data_value);
/* check for codec_data update to re-send sps/pps inband data if
* current frame has no sps/pps but upstream codec_data was updated */
if ((!h264parse->have_sps_in_frame || !h264parse->have_pps_in_frame)
&& buf) {
GstBuffer *codec_data_buf = gst_value_get_buffer (codec_data_value);
GstMapInfo map;
gst_buffer_map (buf, &map, GST_MAP_READ);
if (map.size != gst_buffer_get_size (codec_data_buf) ||
gst_buffer_memcmp (codec_data_buf, 0, map.data, map.size)) {
codec_data_modified = TRUE;
}
gst_buffer_unmap (buf, &map);
}
}
} else if (!buf) {
GstStructure *s;
/* remove any left-over codec-data hanging around */
s = gst_caps_get_structure (caps, 0);
gst_structure_remove_field (s, "codec_data");
}
}
if (!(src_caps && gst_caps_is_strictly_equal (src_caps, caps))) {
/* update codec data to new value */
if (buf) {
gst_caps_set_simple (caps, "codec_data", GST_TYPE_BUFFER, buf, NULL);
gst_buffer_replace (&h264parse->codec_data, buf);
gst_buffer_unref (buf);
buf = NULL;
} else {
GstStructure *s;
/* remove any left-over codec-data hanging around */
s = gst_caps_get_structure (caps, 0);
gst_structure_remove_field (s, "codec_data");
gst_buffer_replace (&h264parse->codec_data, NULL);
}
gst_pad_set_caps (GST_BASE_PARSE_SRC_PAD (h264parse), caps);
} else if (codec_data_modified) {
GST_DEBUG_OBJECT (h264parse,
"Only codec_data is different, need inband sps/pps update");
/* this will insert updated codec_data with next idr */
h264parse->push_codec = TRUE;
}
if (src_caps)
gst_caps_unref (src_caps);
gst_caps_unref (caps);
}
gst_caps_unref (sink_caps);
if (buf)
gst_buffer_unref (buf);
}
static GstClockTime
gst_h264_parse_get_duration (GstH264Parse * h264parse, gboolean frame)
{
GstClockTime ret = GST_CLOCK_TIME_NONE;
GstH264SPS *sps = h264parse->nalparser->last_sps;
gint duration = 1;
if (!frame) {
GST_LOG_OBJECT (h264parse, "no frame data -> 0 duration");
ret = 0;
goto done;
}
if (!sps) {
GST_DEBUG_OBJECT (h264parse, "referred SPS invalid");
goto fps_duration;
} else if (!sps->vui_parameters_present_flag) {
GST_DEBUG_OBJECT (h264parse, "unable to compute duration: VUI not present");
goto fps_duration;
} else if (!sps->vui_parameters.timing_info_present_flag) {
GST_DEBUG_OBJECT (h264parse,
"unable to compute duration: timing info not present");
goto fps_duration;
} else if (sps->vui_parameters.time_scale == 0) {
GST_DEBUG_OBJECT (h264parse,
"unable to compute duration: time_scale = 0 "
"(this is forbidden in spec; bitstream probably contains error)");
goto fps_duration;
}
if (h264parse->sei_pic_struct_pres_flag &&
h264parse->sei_pic_struct != (guint8) - 1) {
/* Note that when h264parse->sei_pic_struct == -1 (unspecified), there
* are ways to infer its value. This is related to computing the
* TopFieldOrderCnt and BottomFieldOrderCnt, which looks
* complicated and thus not implemented for the time being. Yet
* the value we have here is correct for many applications
*/
switch (h264parse->sei_pic_struct) {
case GST_H264_SEI_PIC_STRUCT_TOP_FIELD:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_FIELD:
duration = 1;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME:
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP:
duration = 2;
break;
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
duration = 3;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME_DOUBLING:
duration = 4;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME_TRIPLING:
duration = 6;
break;
default:
GST_DEBUG_OBJECT (h264parse,
"h264parse->sei_pic_struct of unknown value %d. Not parsed",
h264parse->sei_pic_struct);
break;
}
} else {
duration = h264parse->field_pic_flag ? 1 : 2;
}
GST_LOG_OBJECT (h264parse, "frame tick duration %d", duration);
ret = gst_util_uint64_scale (duration * GST_SECOND,
sps->vui_parameters.num_units_in_tick, sps->vui_parameters.time_scale);
/* sanity check */
if (ret < GST_MSECOND) {
GST_DEBUG_OBJECT (h264parse, "discarding dur %" GST_TIME_FORMAT,
GST_TIME_ARGS (ret));
goto fps_duration;
}
done:
return ret;
fps_duration:
if (h264parse->parsed_fps_d > 0 && h264parse->parsed_fps_n > 0)
ret =
gst_util_uint64_scale (GST_SECOND, h264parse->parsed_fps_d,
h264parse->parsed_fps_n);
goto done;
}
static void
gst_h264_parse_get_timestamp (GstH264Parse * h264parse,
GstClockTime * out_ts, GstClockTime * out_dur, gboolean frame)
{
GstH264SPS *sps = h264parse->nalparser->last_sps;
GstClockTime upstream;
gint duration = 1;
g_return_if_fail (out_dur != NULL);
g_return_if_fail (out_ts != NULL);
upstream = *out_ts;
GST_LOG_OBJECT (h264parse, "Upstream ts %" GST_TIME_FORMAT,
GST_TIME_ARGS (upstream));
if (!frame) {
GST_LOG_OBJECT (h264parse, "no frame data -> 0 duration");
*out_dur = 0;
goto exit;
}
if (!sps) {
GST_DEBUG_OBJECT (h264parse, "referred SPS invalid");
goto exit;
} else if (!sps->vui_parameters_present_flag) {
GST_DEBUG_OBJECT (h264parse,
"unable to compute timestamp: VUI not present");
goto exit;
} else if (!sps->vui_parameters.timing_info_present_flag) {
GST_DEBUG_OBJECT (h264parse,
"unable to compute timestamp: timing info not present");
goto exit;
} else if (sps->vui_parameters.time_scale == 0) {
GST_DEBUG_OBJECT (h264parse,
"unable to compute timestamp: time_scale = 0 "
"(this is forbidden in spec; bitstream probably contains error)");
goto exit;
}
if (h264parse->sei_pic_struct_pres_flag &&
h264parse->sei_pic_struct != (guint8) - 1) {
/* Note that when h264parse->sei_pic_struct == -1 (unspecified), there
* are ways to infer its value. This is related to computing the
* TopFieldOrderCnt and BottomFieldOrderCnt, which looks
* complicated and thus not implemented for the time being. Yet
* the value we have here is correct for many applications
*/
switch (h264parse->sei_pic_struct) {
case GST_H264_SEI_PIC_STRUCT_TOP_FIELD:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_FIELD:
duration = 1;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME:
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP:
duration = 2;
break;
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
duration = 3;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME_DOUBLING:
duration = 4;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME_TRIPLING:
duration = 6;
break;
default:
GST_DEBUG_OBJECT (h264parse,
"h264parse->sei_pic_struct of unknown value %d. Not parsed",
h264parse->sei_pic_struct);
break;
}
} else {
duration = h264parse->field_pic_flag ? 1 : 2;
}
GST_LOG_OBJECT (h264parse, "frame tick duration %d", duration);
/*
* h264parse.264 C.1.2 Timing of coded picture removal (equivalent to DTS):
* Tr,n(0) = initial_cpb_removal_delay[ SchedSelIdx ] / 90000
* Tr,n(n) = Tr,n(nb) + Tc * cpb_removal_delay(n)
* where
* Tc = num_units_in_tick / time_scale
*/
if (h264parse->ts_trn_nb != GST_CLOCK_TIME_NONE) {
GST_LOG_OBJECT (h264parse, "buffering based ts");
/* buffering period is present */
if (upstream != GST_CLOCK_TIME_NONE) {
/* If upstream timestamp is valid, we respect it and adjust current
* reference point */
h264parse->ts_trn_nb = upstream -
(GstClockTime) gst_util_uint64_scale
(h264parse->sei_cpb_removal_delay * GST_SECOND,
sps->vui_parameters.num_units_in_tick,
sps->vui_parameters.time_scale);
} else {
/* If no upstream timestamp is given, we write in new timestamp */
upstream = h264parse->dts = h264parse->ts_trn_nb +
(GstClockTime) gst_util_uint64_scale
(h264parse->sei_cpb_removal_delay * GST_SECOND,
sps->vui_parameters.num_units_in_tick,
sps->vui_parameters.time_scale);
}
} else {
GstClockTime dur;
GST_LOG_OBJECT (h264parse, "duration based ts");
/* naive method: no removal delay specified
* track upstream timestamp and provide best guess frame duration */
dur = gst_util_uint64_scale (duration * GST_SECOND,
sps->vui_parameters.num_units_in_tick, sps->vui_parameters.time_scale);
/* sanity check */
if (dur < GST_MSECOND) {
GST_DEBUG_OBJECT (h264parse, "discarding dur %" GST_TIME_FORMAT,
GST_TIME_ARGS (dur));
} else {
*out_dur = dur;
}
}
exit:
if (GST_CLOCK_TIME_IS_VALID (upstream))
*out_ts = h264parse->dts = upstream;
if (GST_CLOCK_TIME_IS_VALID (*out_dur) &&
GST_CLOCK_TIME_IS_VALID (h264parse->dts))
h264parse->dts += *out_dur;
}
static GstFlowReturn
gst_h264_parse_parse_frame (GstBaseParse * parse, GstBaseParseFrame * frame)
{
GstH264Parse *h264parse;
GstBuffer *buffer;
guint av;
h264parse = GST_H264_PARSE (parse);
buffer = frame->buffer;
gst_h264_parse_update_src_caps (h264parse, NULL);
/* don't mess with timestamps if provided by upstream,
* particularly since our ts not that good they handle seeking etc */
if (h264parse->do_ts) {
gst_h264_parse_get_timestamp (h264parse,
&GST_BUFFER_DTS (buffer), &GST_BUFFER_DURATION (buffer),
h264parse->frame_start);
}
/* We don't want to let baseparse select a duration itself based
* solely on the framerate, as we have more per-frame information
* available */
if (!GST_CLOCK_TIME_IS_VALID (GST_BUFFER_DURATION (buffer))) {
GST_BUFFER_DURATION (buffer) =
gst_h264_parse_get_duration (h264parse, h264parse->frame_start);
}
if (h264parse->keyframe)
GST_BUFFER_FLAG_UNSET (buffer, GST_BUFFER_FLAG_DELTA_UNIT);
else
GST_BUFFER_FLAG_SET (buffer, GST_BUFFER_FLAG_DELTA_UNIT);
if (h264parse->discard_bidirectional && h264parse->bidirectional)
goto discard;
if (h264parse->header)
GST_BUFFER_FLAG_SET (buffer, GST_BUFFER_FLAG_HEADER);
else
GST_BUFFER_FLAG_UNSET (buffer, GST_BUFFER_FLAG_HEADER);
if (h264parse->discont) {
GST_BUFFER_FLAG_SET (buffer, GST_BUFFER_FLAG_DISCONT);
h264parse->discont = FALSE;
}
2018-09-28 20:49:52 +00:00
if (h264parse->marker) {
GST_BUFFER_FLAG_SET (buffer, GST_BUFFER_FLAG_MARKER);
h264parse->marker = FALSE;
} else {
GST_BUFFER_FLAG_UNSET (buffer, GST_BUFFER_FLAG_MARKER);
}
/* replace with transformed AVC output if applicable */
av = gst_adapter_available (h264parse->frame_out);
if (av) {
GstBuffer *buf;
buf = gst_adapter_take_buffer (h264parse->frame_out, av);
gst_buffer_copy_into (buf, buffer, GST_BUFFER_COPY_METADATA, 0, -1);
gst_buffer_replace (&frame->out_buffer, buf);
gst_buffer_unref (buf);
}
done:
return GST_FLOW_OK;
discard:
GST_DEBUG_OBJECT (h264parse, "Discarding bidirectional frame");
frame->flags |= GST_BASE_PARSE_FRAME_FLAG_DROP;
gst_h264_parse_reset_frame (h264parse);
goto done;
}
/* sends a codec NAL downstream, decorating and transforming as needed.
* No ownership is taken of @nal */
static GstFlowReturn
gst_h264_parse_push_codec_buffer (GstH264Parse * h264parse,
GstBuffer * nal, GstBuffer * buffer)
{
2012-01-25 15:20:41 +00:00
GstMapInfo map;
GstBuffer *wrapped_nal;
2012-01-25 15:20:41 +00:00
gst_buffer_map (nal, &map, GST_MAP_READ);
wrapped_nal = gst_h264_parse_wrap_nal (h264parse, h264parse->format,
2012-01-25 15:20:41 +00:00
map.data, map.size);
gst_buffer_unmap (nal, &map);
GST_BUFFER_PTS (wrapped_nal) = GST_BUFFER_PTS (buffer);
GST_BUFFER_DTS (wrapped_nal) = GST_BUFFER_DTS (buffer);
GST_BUFFER_DURATION (wrapped_nal) = 0;
return gst_pad_push (GST_BASE_PARSE_SRC_PAD (h264parse), wrapped_nal);
}
static GstEvent *
check_pending_key_unit_event (GstEvent * pending_event,
GstSegment * segment, GstClockTime timestamp, guint flags,
GstClockTime pending_key_unit_ts)
{
GstClockTime running_time, stream_time;
gboolean all_headers;
guint count;
GstEvent *event = NULL;
g_return_val_if_fail (segment != NULL, NULL);
if (pending_event == NULL)
goto out;
if (GST_CLOCK_TIME_IS_VALID (pending_key_unit_ts) &&
timestamp == GST_CLOCK_TIME_NONE)
goto out;
running_time = gst_segment_to_running_time (segment,
GST_FORMAT_TIME, timestamp);
GST_INFO ("now %" GST_TIME_FORMAT " wanted %" GST_TIME_FORMAT,
GST_TIME_ARGS (running_time), GST_TIME_ARGS (pending_key_unit_ts));
if (GST_CLOCK_TIME_IS_VALID (pending_key_unit_ts) &&
running_time < pending_key_unit_ts)
goto out;
if (flags & GST_BUFFER_FLAG_DELTA_UNIT) {
GST_DEBUG ("pending force key unit, waiting for keyframe");
goto out;
}
stream_time = gst_segment_to_stream_time (segment,
GST_FORMAT_TIME, timestamp);
if (!gst_video_event_parse_upstream_force_key_unit (pending_event,
NULL, &all_headers, &count)) {
gst_video_event_parse_downstream_force_key_unit (pending_event, NULL,
NULL, NULL, &all_headers, &count);
}
event =
gst_video_event_new_downstream_force_key_unit (timestamp, stream_time,
running_time, all_headers, count);
gst_event_set_seqnum (event, gst_event_get_seqnum (pending_event));
out:
return event;
}
static void
gst_h264_parse_prepare_key_unit (GstH264Parse * parse, GstEvent * event)
{
GstClockTime running_time;
guint count;
#ifndef GST_DISABLE_GST_DEBUG
gboolean have_sps, have_pps;
gint i;
#endif
parse->pending_key_unit_ts = GST_CLOCK_TIME_NONE;
gst_event_replace (&parse->force_key_unit_event, NULL);
gst_video_event_parse_downstream_force_key_unit (event,
NULL, NULL, &running_time, NULL, &count);
GST_INFO_OBJECT (parse, "pushing downstream force-key-unit event %d "
"%" GST_TIME_FORMAT " count %d", gst_event_get_seqnum (event),
GST_TIME_ARGS (running_time), count);
gst_pad_push_event (GST_BASE_PARSE_SRC_PAD (parse), event);
#ifndef GST_DISABLE_GST_DEBUG
have_sps = have_pps = FALSE;
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++) {
if (parse->sps_nals[i] != NULL) {
have_sps = TRUE;
break;
}
}
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++) {
if (parse->pps_nals[i] != NULL) {
have_pps = TRUE;
break;
}
}
GST_INFO_OBJECT (parse, "preparing key unit, have sps %d have pps %d",
have_sps, have_pps);
#endif
/* set push_codec to TRUE so that pre_push_frame sends SPS/PPS again */
parse->push_codec = TRUE;
}
static gboolean
gst_h264_parse_handle_sps_pps_nals (GstH264Parse * h264parse,
GstBuffer * buffer, GstBaseParseFrame * frame)
{
GstBuffer *codec_nal;
gint i;
gboolean send_done = FALSE;
if (h264parse->have_sps_in_frame && h264parse->have_pps_in_frame) {
GST_DEBUG_OBJECT (h264parse, "SPS/PPS exist in frame, will not insert");
return TRUE;
}
if (h264parse->align == GST_H264_PARSE_ALIGN_NAL) {
/* send separate config NAL buffers */
GST_DEBUG_OBJECT (h264parse, "- sending SPS/PPS");
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++) {
if ((codec_nal = h264parse->sps_nals[i])) {
GST_DEBUG_OBJECT (h264parse, "sending SPS nal");
gst_h264_parse_push_codec_buffer (h264parse, codec_nal, buffer);
send_done = TRUE;
}
}
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++) {
if ((codec_nal = h264parse->pps_nals[i])) {
GST_DEBUG_OBJECT (h264parse, "sending PPS nal");
gst_h264_parse_push_codec_buffer (h264parse, codec_nal, buffer);
send_done = TRUE;
}
}
} else {
/* insert config NALs into AU */
GstByteWriter bw;
GstBuffer *new_buf;
const gboolean bs = h264parse->format == GST_H264_PARSE_FORMAT_BYTE;
const gint nls = 4 - h264parse->nal_length_size;
gboolean ok;
gst_byte_writer_init_with_size (&bw, gst_buffer_get_size (buffer), FALSE);
ok = gst_byte_writer_put_buffer (&bw, buffer, 0, h264parse->idr_pos);
GST_DEBUG_OBJECT (h264parse, "- inserting SPS/PPS");
for (i = 0; i < GST_H264_MAX_SPS_COUNT; i++) {
if ((codec_nal = h264parse->sps_nals[i])) {
gsize nal_size = gst_buffer_get_size (codec_nal);
GST_DEBUG_OBJECT (h264parse, "inserting SPS nal");
if (bs) {
ok &= gst_byte_writer_put_uint32_be (&bw, 1);
} else {
ok &= gst_byte_writer_put_uint32_be (&bw, (nal_size << (nls * 8)));
ok &= gst_byte_writer_set_pos (&bw,
gst_byte_writer_get_pos (&bw) - nls);
}
ok &= gst_byte_writer_put_buffer (&bw, codec_nal, 0, nal_size);
send_done = TRUE;
}
}
for (i = 0; i < GST_H264_MAX_PPS_COUNT; i++) {
if ((codec_nal = h264parse->pps_nals[i])) {
gsize nal_size = gst_buffer_get_size (codec_nal);
GST_DEBUG_OBJECT (h264parse, "inserting PPS nal");
if (bs) {
ok &= gst_byte_writer_put_uint32_be (&bw, 1);
} else {
ok &= gst_byte_writer_put_uint32_be (&bw, (nal_size << (nls * 8)));
ok &= gst_byte_writer_set_pos (&bw,
gst_byte_writer_get_pos (&bw) - nls);
}
ok &= gst_byte_writer_put_buffer (&bw, codec_nal, 0, nal_size);
send_done = TRUE;
}
}
ok &= gst_byte_writer_put_buffer (&bw, buffer, h264parse->idr_pos, -1);
/* collect result and push */
new_buf = gst_byte_writer_reset_and_get_buffer (&bw);
gst_buffer_copy_into (new_buf, buffer, GST_BUFFER_COPY_METADATA, 0, -1);
/* should already be keyframe/IDR, but it may not have been,
* so mark it as such to avoid being discarded by picky decoder */
GST_BUFFER_FLAG_UNSET (new_buf, GST_BUFFER_FLAG_DELTA_UNIT);
gst_buffer_replace (&frame->out_buffer, new_buf);
gst_buffer_unref (new_buf);
/* some result checking seems to make some compilers happy */
if (G_UNLIKELY (!ok)) {
GST_ERROR_OBJECT (h264parse, "failed to insert SPS/PPS");
}
}
return send_done;
}
static GstBuffer *
gst_h264_parse_create_pic_timing_sei (GstH264Parse * h264parse,
GstBuffer * buffer)
{
guint num_meta;
const guint8 num_clock_ts_table[9] = {
1, 1, 1, 2, 2, 3, 3, 2, 3
};
guint num_clock_ts;
GstBuffer *out_buf = NULL;
GstMemory *sei_mem;
GArray *msg_array;
gint i, j;
GstH264SEIMessage sei;
GstH264PicTiming *pic_timing;
GstVideoTimeCodeMeta *tc_meta;
gpointer iter = NULL;
guint8 ct_type = GST_H264_CT_TYPE_PROGRESSIVE;
if (!h264parse->update_timecode)
return NULL;
num_meta = gst_buffer_get_n_meta (buffer, GST_VIDEO_TIME_CODE_META_API_TYPE);
if (num_meta == 0)
return NULL;
if (!h264parse->sei_pic_struct_pres_flag || h264parse->pic_timing_sei_pos < 0) {
GST_ELEMENT_WARNING (h264parse, STREAM, NOT_IMPLEMENTED, (NULL),
("timecode update was requested but VUI doesn't support timecode"));
return NULL;
}
g_assert (h264parse->sei_pic_struct <=
GST_H264_SEI_PIC_STRUCT_FRAME_TRIPLING);
num_clock_ts = num_clock_ts_table[h264parse->sei_pic_struct];
if (num_meta > num_clock_ts) {
GST_LOG_OBJECT (h264parse,
"The number of timecode meta %d is superior to required %d",
num_meta, num_clock_ts);
return NULL;
}
GST_LOG_OBJECT (h264parse,
"The number of timecode meta %d is compatible", num_meta);
memset (&sei, 0, sizeof (GstH264SEIMessage));
sei.payloadType = GST_H264_SEI_PIC_TIMING;
memcpy (&sei.payload.pic_timing,
&h264parse->pic_timing_sei, sizeof (GstH264PicTiming));
pic_timing = &sei.payload.pic_timing;
switch (h264parse->sei_pic_struct) {
case GST_H264_SEI_PIC_STRUCT_FRAME:
case GST_H264_SEI_PIC_STRUCT_FRAME_DOUBLING:
case GST_H264_SEI_PIC_STRUCT_FRAME_TRIPLING:
ct_type = GST_H264_CT_TYPE_PROGRESSIVE;
break;
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP:
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
ct_type = GST_H264_CT_TYPE_INTERLACED;
break;
default:
ct_type = GST_H264_CT_TYPE_UNKNOWN;
break;
}
i = 0;
while ((tc_meta =
(GstVideoTimeCodeMeta *) gst_buffer_iterate_meta_filtered (buffer,
&iter, GST_VIDEO_TIME_CODE_META_API_TYPE))) {
GstH264ClockTimestamp *tim = &pic_timing->clock_timestamp[i];
GstVideoTimeCode *tc = &tc_meta->tc;
pic_timing->clock_timestamp_flag[i] = 1;
tim->ct_type = ct_type;
tim->nuit_field_based_flag = 1;
tim->counting_type = 0;
if ((tc->config.flags & GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME)
== GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME)
tim->counting_type = 4;
tim->discontinuity_flag = 0;
tim->cnt_dropped_flag = 0;
tim->n_frames = tc->frames;
tim->hours_value = tc->hours;
tim->minutes_value = tc->minutes;
tim->seconds_value = tc->seconds;
tim->full_timestamp_flag =
tim->seconds_flag = tim->minutes_flag = tim->hours_flag = 0;
if (tc->hours > 0)
tim->full_timestamp_flag = 1;
else if (tc->minutes > 0)
tim->seconds_flag = tim->minutes_flag = 1;
else if (tc->seconds > 0)
tim->seconds_flag = 1;
GST_LOG_OBJECT (h264parse,
"New time code value %02u:%02u:%02u:%02u",
tim->hours_value, tim->minutes_value, tim->seconds_value,
tim->n_frames);
i++;
}
for (j = i; j < 3; j++)
pic_timing->clock_timestamp_flag[j] = 0;
msg_array = g_array_new (FALSE, FALSE, sizeof (GstH264SEIMessage));
g_array_set_clear_func (msg_array, (GDestroyNotify) gst_h264_sei_clear);
g_array_append_val (msg_array, sei);
if (h264parse->format == GST_H264_PARSE_FORMAT_BYTE) {
sei_mem = gst_h264_create_sei_memory (3, msg_array);
} else {
sei_mem = gst_h264_create_sei_memory_avc (h264parse->nal_length_size,
msg_array);
}
g_array_unref (msg_array);
if (!sei_mem) {
GST_WARNING_OBJECT (h264parse, "Cannot create Picture Timing SEI memory");
return NULL;
}
out_buf = gst_buffer_new ();
gst_buffer_copy_into (out_buf, buffer, GST_BUFFER_COPY_METADATA, 0, -1);
if (h264parse->align == GST_H264_PARSE_ALIGN_NAL) {
gst_buffer_append_memory (out_buf, sei_mem);
} else {
gsize mem_size;
mem_size = gst_memory_get_sizes (sei_mem, NULL, NULL);
/* copy every data except for the SEI */
if (h264parse->pic_timing_sei_pos > 0) {
gst_buffer_copy_into (out_buf, buffer, GST_BUFFER_COPY_MEMORY, 0,
h264parse->pic_timing_sei_pos);
}
/* insert new SEI */
gst_buffer_append_memory (out_buf, sei_mem);
if (gst_buffer_get_size (buffer) >
h264parse->pic_timing_sei_pos + h264parse->pic_timing_sei_size) {
gst_buffer_copy_into (out_buf, buffer, GST_BUFFER_COPY_MEMORY,
h264parse->pic_timing_sei_pos + h264parse->pic_timing_sei_size, -1);
}
if (h264parse->idr_pos >= 0) {
h264parse->idr_pos += mem_size;
h264parse->idr_pos -= h264parse->pic_timing_sei_size;
}
}
return out_buf;
}
static GstFlowReturn
gst_h264_parse_pre_push_frame (GstBaseParse * parse, GstBaseParseFrame * frame)
{
GstH264Parse *h264parse;
GstBuffer *buffer;
GstBuffer *new_buf;
GstEvent *event;
GstBuffer *parse_buffer = NULL;
gboolean is_interlaced = FALSE;
h264parse = GST_H264_PARSE (parse);
2013-12-16 09:18:23 +00:00
if (h264parse->first_frame) {
2013-12-16 09:18:23 +00:00
GstTagList *taglist;
GstCaps *caps;
/* codec tag */
caps = gst_pad_get_current_caps (GST_BASE_PARSE_SRC_PAD (parse));
if (caps == NULL) {
if (GST_PAD_IS_FLUSHING (GST_BASE_PARSE_SRC_PAD (h264parse))) {
GST_INFO_OBJECT (h264parse, "Src pad is flushing");
return GST_FLOW_FLUSHING;
} else {
GST_INFO_OBJECT (h264parse, "Src pad is not negotiated!");
return GST_FLOW_NOT_NEGOTIATED;
}
}
taglist = gst_tag_list_new_empty ();
2013-12-16 09:18:23 +00:00
gst_pb_utils_add_codec_description_to_tag_list (taglist,
GST_TAG_VIDEO_CODEC, caps);
gst_caps_unref (caps);
gst_base_parse_merge_tags (parse, taglist, GST_TAG_MERGE_REPLACE);
gst_tag_list_unref (taglist);
2013-12-16 09:18:23 +00:00
/* also signals the end of first-frame processing */
h264parse->first_frame = FALSE;
2013-12-16 09:18:23 +00:00
}
2019-09-02 19:08:44 +00:00
/* In case of byte-stream, insert au delimiter by default
* if it doesn't exist */
if (h264parse->aud_insert && h264parse->format == GST_H264_PARSE_FORMAT_BYTE) {
GST_DEBUG_OBJECT (h264parse, "Inserting AUD into the stream.");
if (h264parse->align == GST_H264_PARSE_ALIGN_AU) {
GstMemory *mem =
gst_memory_new_wrapped (GST_MEMORY_FLAG_READONLY, (guint8 *) au_delim,
sizeof (au_delim), 0, sizeof (au_delim), NULL, NULL);
frame->out_buffer = gst_buffer_copy (frame->buffer);
gst_buffer_prepend_memory (frame->out_buffer, mem);
if (h264parse->idr_pos >= 0)
h264parse->idr_pos += sizeof (au_delim);
buffer = frame->out_buffer;
} else {
GstBuffer *aud_buffer = gst_buffer_new_allocate (NULL, 2, NULL);
gst_buffer_fill (aud_buffer, 0, (guint8 *) (au_delim + 4), 2);
buffer = frame->buffer;
gst_h264_parse_push_codec_buffer (h264parse, aud_buffer, buffer);
gst_buffer_unref (aud_buffer);
}
} else {
buffer = frame->buffer;
}
h264parse->aud_insert = FALSE;
if ((event = check_pending_key_unit_event (h264parse->force_key_unit_event,
&parse->segment, GST_BUFFER_TIMESTAMP (buffer),
GST_BUFFER_FLAGS (buffer), h264parse->pending_key_unit_ts))) {
gst_h264_parse_prepare_key_unit (h264parse, event);
}
/* handle timecode */
new_buf = gst_h264_parse_create_pic_timing_sei (h264parse, buffer);
if (new_buf) {
if (frame->out_buffer)
gst_buffer_unref (frame->out_buffer);
buffer = frame->out_buffer = new_buf;
}
/* periodic SPS/PPS sending */
if (h264parse->interval > 0 || h264parse->push_codec) {
GstClockTime timestamp = GST_BUFFER_TIMESTAMP (buffer);
guint64 diff;
gboolean initial_frame = FALSE;
/* init */
if (!GST_CLOCK_TIME_IS_VALID (h264parse->last_report)) {
h264parse->last_report = timestamp;
initial_frame = TRUE;
}
if (h264parse->idr_pos >= 0) {
GST_LOG_OBJECT (h264parse, "IDR nal at offset %d", h264parse->idr_pos);
if (timestamp > h264parse->last_report)
diff = timestamp - h264parse->last_report;
else
diff = 0;
GST_LOG_OBJECT (h264parse,
"now %" GST_TIME_FORMAT ", last SPS/PPS %" GST_TIME_FORMAT,
GST_TIME_ARGS (timestamp), GST_TIME_ARGS (h264parse->last_report));
GST_DEBUG_OBJECT (h264parse,
"interval since last SPS/PPS %" GST_TIME_FORMAT,
GST_TIME_ARGS (diff));
if (GST_TIME_AS_SECONDS (diff) >= h264parse->interval ||
initial_frame || h264parse->push_codec) {
GstClockTime new_ts;
/* avoid overwriting a perfectly fine timestamp */
new_ts = GST_CLOCK_TIME_IS_VALID (timestamp) ? timestamp :
h264parse->last_report;
if (gst_h264_parse_handle_sps_pps_nals (h264parse, buffer, frame)) {
h264parse->last_report = new_ts;
}
}
/* we pushed whatever we had */
h264parse->push_codec = FALSE;
h264parse->have_sps = FALSE;
h264parse->have_pps = FALSE;
h264parse->state &= GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS;
}
} else if (h264parse->interval == -1) {
if (h264parse->idr_pos >= 0) {
GST_LOG_OBJECT (h264parse, "IDR nal at offset %d", h264parse->idr_pos);
gst_h264_parse_handle_sps_pps_nals (h264parse, buffer, frame);
/* we pushed whatever we had */
h264parse->push_codec = FALSE;
h264parse->have_sps = FALSE;
h264parse->have_pps = FALSE;
h264parse->state &= GST_H264_PARSE_STATE_VALID_PICTURE_HEADERS;
}
}
2019-09-02 19:08:44 +00:00
/* Fixme: setting passthrough mode causing multiple issues:
* For nal aligned multiresoluton streams, passthrough mode make h264parse
* unable to advertise the new resolutions. Also causing issues while
* parsing MVC streams when it has two layers.
* Disabing passthourgh mode for now */
#if 0
/* If SPS/PPS and a keyframe have been parsed, and we're not converting,
* we might switch to passthrough mode now on the basis that we've seen
* the SEI packets and know optional caps params (such as multiview).
* This is an efficiency optimisation that relies on stream properties
* remaining uniform in practice. */
if (h264parse->can_passthrough) {
if (h264parse->keyframe && h264parse->have_sps && h264parse->have_pps) {
GST_LOG_OBJECT (parse, "Switching to passthrough mode");
gst_base_parse_set_passthrough (parse, TRUE);
}
}
#endif
if (frame->out_buffer) {
parse_buffer = frame->out_buffer =
gst_buffer_make_writable (frame->out_buffer);
} else {
parse_buffer = frame->buffer = gst_buffer_make_writable (frame->buffer);
}
if (!gst_buffer_get_video_time_code_meta (parse_buffer)) {
guint i = 0;
for (i = 0; i < 3 && h264parse->num_clock_timestamp; i++) {
GstH264ClockTimestamp *tim =
&h264parse->pic_timing_sei.clock_timestamp[i];
gint field_count = -1;
guint n_frames;
GstVideoTimeCodeFlags flags = 0;
if (!h264parse->pic_timing_sei.clock_timestamp_flag[i])
continue;
/* Table D-1 */
switch (h264parse->sei_pic_struct) {
case GST_H264_SEI_PIC_STRUCT_FRAME:
case GST_H264_SEI_PIC_STRUCT_TOP_FIELD:
case GST_H264_SEI_PIC_STRUCT_BOTTOM_FIELD:
field_count = h264parse->sei_pic_struct;
break;
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM:
field_count = i + 1;
break;
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP:
field_count = 2 - i;
break;
case GST_H264_SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
field_count = i % 2 ? 2 : 1;
break;
case GST_H264_SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
field_count = i % 2 ? 1 : 2;
break;
case GST_H264_SEI_PIC_STRUCT_FRAME_DOUBLING:
case GST_H264_SEI_PIC_STRUCT_FRAME_TRIPLING:
field_count = 0;
break;
}
if (field_count == -1) {
GST_WARNING_OBJECT (parse,
"failed to determine field count for timecode");
field_count = 0;
}
/* dropping of the two lowest (value 0 and 1) n_frames
* counts when seconds_value is equal to 0 and
* minutes_value is not an integer multiple of 10 */
if (tim->counting_type == 4)
flags |= GST_VIDEO_TIME_CODE_FLAGS_DROP_FRAME;
if (tim->ct_type == GST_H264_CT_TYPE_INTERLACED) {
flags |= GST_VIDEO_TIME_CODE_FLAGS_INTERLACED;
is_interlaced = TRUE;
}
n_frames =
gst_util_uint64_scale_int (tim->n_frames, 1,
2 - tim->nuit_field_based_flag);
GST_LOG_OBJECT (h264parse,
"Add time code meta %02u:%02u:%02u:%02u",
tim->hours_value, tim->minutes_value, tim->seconds_value, n_frames);
gst_buffer_add_video_time_code_meta_full (parse_buffer,
h264parse->parsed_fps_n,
h264parse->parsed_fps_d,
NULL,
flags,
tim->hours_flag ? tim->hours_value : 0,
tim->minutes_flag ? tim->minutes_value : 0,
tim->seconds_flag ? tim->seconds_value : 0, n_frames, field_count);
}
h264parse->num_clock_timestamp = 0;
}
if (is_interlaced) {
GST_BUFFER_FLAG_SET (parse_buffer, GST_VIDEO_BUFFER_FLAG_INTERLACED);
if (h264parse->sei_pic_struct == GST_H264_SEI_PIC_STRUCT_TOP_FIELD)
GST_BUFFER_FLAG_SET (parse_buffer, GST_VIDEO_BUFFER_FLAG_TFF);
}
gst_video_push_user_data ((GstElement *) h264parse, &h264parse->user_data,
parse_buffer);
gst_h264_parse_reset_frame (h264parse);
return GST_FLOW_OK;
}
static gboolean
gst_h264_parse_set_caps (GstBaseParse * parse, GstCaps * caps)
{
GstH264Parse *h264parse;
GstStructure *str;
const GValue *codec_data_value;
GstBuffer *codec_data = NULL;
gsize size;
guint format, align, off;
GstH264NalUnit nalu;
GstH264ParserResult parseres;
GstCaps *old_caps;
h264parse = GST_H264_PARSE (parse);
/* reset */
h264parse->push_codec = FALSE;
old_caps = gst_pad_get_current_caps (GST_BASE_PARSE_SINK_PAD (parse));
if (old_caps) {
if (!gst_caps_is_equal (old_caps, caps))
gst_h264_parse_reset_stream_info (h264parse);
gst_caps_unref (old_caps);
}
str = gst_caps_get_structure (caps, 0);
/* accept upstream info if provided */
gst_structure_get_int (str, "width", &h264parse->width);
gst_structure_get_int (str, "height", &h264parse->height);
gst_structure_get_fraction (str, "framerate", &h264parse->fps_num,
&h264parse->fps_den);
gst_structure_get_fraction (str, "pixel-aspect-ratio",
&h264parse->upstream_par_n, &h264parse->upstream_par_d);
/* get upstream format and align from caps */
gst_h264_parse_format_from_caps (caps, &format, &align);
codec_data_value = gst_structure_get_value (str, "codec_data");
/* fix up caps without stream-format for max. backwards compatibility */
if (format == GST_H264_PARSE_FORMAT_NONE) {
/* codec_data implies avc */
if (codec_data_value != NULL) {
GST_ERROR ("video/x-h264 caps with codec_data but no stream-format=avc");
format = GST_H264_PARSE_FORMAT_AVC;
} else {
/* otherwise assume bytestream input */
GST_ERROR ("video/x-h264 caps without codec_data or stream-format");
format = GST_H264_PARSE_FORMAT_BYTE;
}
}
/* avc caps sanity checks */
if (format == GST_H264_PARSE_FORMAT_AVC) {
/* AVC requires codec_data, AVC3 might have one and/or SPS/PPS inline */
if (codec_data_value == NULL)
goto avc_caps_codec_data_missing;
/* AVC implies alignment=au, everything else is not allowed */
if (align == GST_H264_PARSE_ALIGN_NONE)
align = GST_H264_PARSE_ALIGN_AU;
else if (align != GST_H264_PARSE_ALIGN_AU)
goto avc_caps_wrong_alignment;
}
/* bytestream caps sanity checks */
if (format == GST_H264_PARSE_FORMAT_BYTE) {
/* should have SPS/PSS in-band (and/or oob in streamheader field) */
if (codec_data_value != NULL)
goto bytestream_caps_with_codec_data;
}
/* packetized video has codec_data (required for AVC, optional for AVC3) */
if (codec_data_value != NULL) {
2012-01-25 15:20:41 +00:00
GstMapInfo map;
guint8 *data;
guint num_sps, num_pps;
#ifndef GST_DISABLE_GST_DEBUG
guint profile;
#endif
gint i;
GST_DEBUG_OBJECT (h264parse, "have packetized h264");
/* make note for optional split processing */
h264parse->packetized = TRUE;
/* codec_data field should hold a buffer */
if (!GST_VALUE_HOLDS_BUFFER (codec_data_value))
goto avc_caps_codec_data_wrong_type;
codec_data = gst_value_get_buffer (codec_data_value);
if (!codec_data)
goto avc_caps_codec_data_missing;
2012-01-25 15:20:41 +00:00
gst_buffer_map (codec_data, &map, GST_MAP_READ);
data = map.data;
size = map.size;
/* parse the avcC data */
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
if (size < 7) { /* when numSPS==0 and numPPS==0, length is 7 bytes */
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (codec_data, &map);
goto avcc_too_small;
}
/* parse the version, this must be 1 */
if (data[0] != 1) {
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (codec_data, &map);
goto wrong_version;
}
#ifndef GST_DISABLE_GST_DEBUG
/* AVCProfileIndication */
/* profile_compat */
/* AVCLevelIndication */
profile = (data[1] << 16) | (data[2] << 8) | data[3];
GST_DEBUG_OBJECT (h264parse, "profile %06x", profile);
#endif
/* 6 bits reserved | 2 bits lengthSizeMinusOne */
/* this is the number of bytes in front of the NAL units to mark their
* length */
h264parse->nal_length_size = (data[4] & 0x03) + 1;
GST_DEBUG_OBJECT (h264parse, "nal length size %u",
h264parse->nal_length_size);
num_sps = data[5] & 0x1f;
off = 6;
for (i = 0; i < num_sps; i++) {
parseres = gst_h264_parser_identify_nalu_avc (h264parse->nalparser,
data, off, size, 2, &nalu);
if (parseres != GST_H264_PARSER_OK) {
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (codec_data, &map);
goto avcc_too_small;
}
gst_h264_parse_process_nal (h264parse, &nalu);
off = nalu.offset + nalu.size;
}
if (off >= size) {
gst_buffer_unmap (codec_data, &map);
goto avcc_too_small;
}
num_pps = data[off];
off++;
for (i = 0; i < num_pps; i++) {
parseres = gst_h264_parser_identify_nalu_avc (h264parse->nalparser,
data, off, size, 2, &nalu);
if (parseres != GST_H264_PARSER_OK) {
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (codec_data, &map);
goto avcc_too_small;
}
gst_h264_parse_process_nal (h264parse, &nalu);
off = nalu.offset + nalu.size;
}
2012-01-25 15:20:41 +00:00
gst_buffer_unmap (codec_data, &map);
gst_buffer_replace (&h264parse->codec_data_in, codec_data);
/* don't confuse codec_data with inband sps/pps */
h264parse->have_sps_in_frame = FALSE;
h264parse->have_pps_in_frame = FALSE;
} else if (format == GST_H264_PARSE_FORMAT_BYTE) {
GST_DEBUG_OBJECT (h264parse, "have bytestream h264");
/* nothing to pre-process */
h264parse->packetized = FALSE;
/* we have 4 sync bytes */
h264parse->nal_length_size = 4;
} else {
/* probably AVC3 without codec_data field, anything to do here? */
}
{
GstCaps *in_caps;
/* prefer input type determined above */
in_caps = gst_caps_new_simple ("video/x-h264",
"parsed", G_TYPE_BOOLEAN, TRUE,
"stream-format", G_TYPE_STRING,
gst_h264_parse_get_string (h264parse, TRUE, format),
"alignment", G_TYPE_STRING,
gst_h264_parse_get_string (h264parse, FALSE, align), NULL);
/* negotiate with downstream, sets ->format and ->align */
gst_h264_parse_negotiate (h264parse, format, in_caps);
gst_caps_unref (in_caps);
}
if (format == h264parse->format && align == h264parse->align) {
/* we did parse codec-data and might supplement src caps */
gst_h264_parse_update_src_caps (h264parse, caps);
h264parse: Add support for stream-format=avc3 When outputting in AVC3 stream format, the codec_data should not contain any SPS or PPS, because they are embedded inside the stream. In case of avc->bytestream h264parse will push the SPS and PPS from codec_data downstream at the start of the stream, at intervals controlled by "config-interval" and when there is a codec_data change. In the case of avc3->bytstream h264parse detects that there is already SPS/PPS in the stream and sets h264parse->push_codec to FALSE. Therefore avc3->bytstream was already supported, except for the stream type. In the case of bystream->avc h264parse will generate codec_data caps from the parsed SPS/PPS in the stream. However it does not remove these SPS/PPS from the stream. bytestream->avc3 is the same as bytestream->avc except that the codec_data must not have any SPS/PPS in it. |--------------+-------------+-------------------| |stream-format | SPS in-band | SPS in codec_data | |--------------+-------------+-------------------| | avc | maybe | always | |--------------+-------------+-------------------| | avc3 | always | never | |--------------+-------------+-------------------| Amendment 2 of ISO/IEC 14496-15 (AVC file format) is defining a new structure for fragmented MP4 called "avc3". The principal difference between AVC1 and AVC3 is the location of the codec initialisation data (e.g. SPS, PPS). In AVC1 this data is placed in the initial MOOV box (moov.trak.mdia.minf.stbl.stsd.avc1) but in AVC3 this data goes in the first sample of every fragment. https://bugzilla.gnome.org/show_bug.cgi?id=702004
2013-08-30 12:54:40 +00:00
} else if (format == GST_H264_PARSE_FORMAT_AVC
|| format == GST_H264_PARSE_FORMAT_AVC3) {
/* if input != output, and input is avc, must split before anything else */
/* arrange to insert codec-data in-stream if needed.
* src caps are only arranged for later on */
h264parse->push_codec = TRUE;
h264parse->have_sps = FALSE;
h264parse->have_pps = FALSE;
if (h264parse->align == GST_H264_PARSE_ALIGN_NAL)
h264parse->split_packetized = TRUE;
h264parse->packetized = TRUE;
}
h264parse->in_align = align;
return TRUE;
/* ERRORS */
avc_caps_codec_data_wrong_type:
{
GST_WARNING_OBJECT (parse, "H.264 AVC caps, codec_data field not a buffer");
goto refuse_caps;
}
avc_caps_codec_data_missing:
{
GST_WARNING_OBJECT (parse, "H.264 AVC caps, but no codec_data");
goto refuse_caps;
}
avc_caps_wrong_alignment:
{
GST_WARNING_OBJECT (parse, "H.264 AVC caps with NAL alignment, must be AU");
goto refuse_caps;
}
bytestream_caps_with_codec_data:
{
GST_WARNING_OBJECT (parse, "H.264 bytestream caps with codec_data is not "
"expected, send SPS/PPS in-band with data or in streamheader field");
goto refuse_caps;
}
avcc_too_small:
{
GST_DEBUG_OBJECT (h264parse, "avcC size %" G_GSIZE_FORMAT " < 8", size);
goto refuse_caps;
}
wrong_version:
{
GST_DEBUG_OBJECT (h264parse, "wrong avcC version");
goto refuse_caps;
}
refuse_caps:
{
GST_WARNING_OBJECT (h264parse, "refused caps %" GST_PTR_FORMAT, caps);
return FALSE;
}
}
static void
remove_fields (GstCaps * caps, gboolean all)
{
guint i, n;
n = gst_caps_get_size (caps);
for (i = 0; i < n; i++) {
GstStructure *s = gst_caps_get_structure (caps, i);
if (all) {
gst_structure_remove_field (s, "alignment");
gst_structure_remove_field (s, "stream-format");
}
gst_structure_remove_field (s, "parsed");
}
}
static GstCaps *
gst_h264_parse_get_caps (GstBaseParse * parse, GstCaps * filter)
{
GstCaps *peercaps, *templ;
GstCaps *res, *tmp, *pcopy;
templ = gst_pad_get_pad_template_caps (GST_BASE_PARSE_SINK_PAD (parse));
if (filter) {
GstCaps *fcopy = gst_caps_copy (filter);
/* Remove the fields we convert */
remove_fields (fcopy, TRUE);
peercaps = gst_pad_peer_query_caps (GST_BASE_PARSE_SRC_PAD (parse), fcopy);
gst_caps_unref (fcopy);
} else
peercaps = gst_pad_peer_query_caps (GST_BASE_PARSE_SRC_PAD (parse), NULL);
pcopy = gst_caps_copy (peercaps);
remove_fields (pcopy, TRUE);
res = gst_caps_intersect_full (pcopy, templ, GST_CAPS_INTERSECT_FIRST);
gst_caps_unref (pcopy);
gst_caps_unref (templ);
if (filter) {
GstCaps *tmp = gst_caps_intersect_full (res, filter,
GST_CAPS_INTERSECT_FIRST);
gst_caps_unref (res);
res = tmp;
}
/* Try if we can put the downstream caps first */
pcopy = gst_caps_copy (peercaps);
remove_fields (pcopy, FALSE);
tmp = gst_caps_intersect_full (pcopy, res, GST_CAPS_INTERSECT_FIRST);
gst_caps_unref (pcopy);
if (!gst_caps_is_empty (tmp))
res = gst_caps_merge (tmp, res);
else
gst_caps_unref (tmp);
gst_caps_unref (peercaps);
return res;
}
static gboolean
gst_h264_parse_event (GstBaseParse * parse, GstEvent * event)
{
gboolean res;
GstH264Parse *h264parse = GST_H264_PARSE (parse);
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_CUSTOM_DOWNSTREAM:
{
GstClockTime timestamp, stream_time, running_time;
gboolean all_headers;
guint count;
if (gst_video_event_is_force_key_unit (event)) {
gst_video_event_parse_downstream_force_key_unit (event,
&timestamp, &stream_time, &running_time, &all_headers, &count);
GST_INFO_OBJECT (h264parse,
"received downstream force key unit event, "
"seqnum %d running_time %" GST_TIME_FORMAT
" all_headers %d count %d", gst_event_get_seqnum (event),
GST_TIME_ARGS (running_time), all_headers, count);
if (h264parse->force_key_unit_event) {
GST_INFO_OBJECT (h264parse, "ignoring force key unit event "
"as one is already queued");
} else {
h264parse->pending_key_unit_ts = running_time;
gst_event_replace (&h264parse->force_key_unit_event, event);
}
gst_event_unref (event);
res = TRUE;
} else {
res = GST_BASE_PARSE_CLASS (parent_class)->sink_event (parse, event);
break;
}
break;
}
case GST_EVENT_FLUSH_STOP:
case GST_EVENT_SEGMENT_DONE:
h264parse->dts = GST_CLOCK_TIME_NONE;
h264parse->ts_trn_nb = GST_CLOCK_TIME_NONE;
h264parse->push_codec = TRUE;
res = GST_BASE_PARSE_CLASS (parent_class)->sink_event (parse, event);
break;
case GST_EVENT_SEGMENT:
{
const GstSegment *segment;
gst_event_parse_segment (event, &segment);
/* don't try to mess with more subtle cases (e.g. seek) */
if (segment->format == GST_FORMAT_TIME &&
(segment->start != 0 || segment->rate != 1.0
|| segment->applied_rate != 1.0))
h264parse->do_ts = FALSE;
if (segment->flags & GST_SEEK_FLAG_TRICKMODE_FORWARD_PREDICTED) {
GST_DEBUG_OBJECT (h264parse, "Will discard bidirectional frames");
h264parse->discard_bidirectional = TRUE;
}
h264parse->last_report = GST_CLOCK_TIME_NONE;
res = GST_BASE_PARSE_CLASS (parent_class)->sink_event (parse, event);
break;
}
default:
res = GST_BASE_PARSE_CLASS (parent_class)->sink_event (parse, event);
break;
}
return res;
}
static gboolean
gst_h264_parse_src_event (GstBaseParse * parse, GstEvent * event)
{
gboolean res;
GstH264Parse *h264parse = GST_H264_PARSE (parse);
switch (GST_EVENT_TYPE (event)) {
case GST_EVENT_CUSTOM_UPSTREAM:
{
GstClockTime running_time;
gboolean all_headers;
guint count;
if (gst_video_event_is_force_key_unit (event)) {
gst_video_event_parse_upstream_force_key_unit (event,
&running_time, &all_headers, &count);
GST_INFO_OBJECT (h264parse, "received upstream force-key-unit event, "
"seqnum %d running_time %" GST_TIME_FORMAT
" all_headers %d count %d", gst_event_get_seqnum (event),
GST_TIME_ARGS (running_time), all_headers, count);
if (all_headers) {
h264parse->pending_key_unit_ts = running_time;
gst_event_replace (&h264parse->force_key_unit_event, event);
}
}
res = GST_BASE_PARSE_CLASS (parent_class)->src_event (parse, event);
break;
}
default:
res = GST_BASE_PARSE_CLASS (parent_class)->src_event (parse, event);
break;
}
return res;
}
static void
gst_h264_parse_set_property (GObject * object, guint prop_id,
const GValue * value, GParamSpec * pspec)
{
GstH264Parse *parse;
parse = GST_H264_PARSE (object);
switch (prop_id) {
case PROP_CONFIG_INTERVAL:
parse->interval = g_value_get_int (value);
break;
case PROP_UPDATE_TIMECODE:
parse->update_timecode = g_value_get_boolean (value);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}
static void
gst_h264_parse_get_property (GObject * object, guint prop_id,
GValue * value, GParamSpec * pspec)
{
GstH264Parse *parse;
parse = GST_H264_PARSE (object);
switch (prop_id) {
case PROP_CONFIG_INTERVAL:
g_value_set_int (value, parse->interval);
break;
case PROP_UPDATE_TIMECODE:
g_value_set_boolean (value, parse->update_timecode);
break;
default:
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
break;
}
}