GF_Err MCDec_InitMpeg4Decoder(MCDec *ctx)
{
char *dsi_data = NULL;
u32 dsi_data_size = 0;
if (!ctx->esd->decoderConfig->decoderSpecificInfo) {
ctx->esd->decoderConfig->decoderSpecificInfo = (GF_DefaultDescriptor *) gf_odf_desc_new(GF_ODF_DSI_TAG);
}
if (ctx->esd->decoderConfig->decoderSpecificInfo->data) {
GF_M4VDecSpecInfo vcfg;
GF_BitStream *bs;
gf_m4v_get_config(ctx->esd->decoderConfig->decoderSpecificInfo->data, ctx->esd->decoderConfig->decoderSpecificInfo->dataLength, &vcfg);
ctx->width = vcfg.width;
ctx->height = vcfg.height;
if (ctx->esd->slConfig) {
ctx->esd->slConfig->predefined = 2;
}
bs = gf_bs_new(NULL, 0, GF_BITSTREAM_WRITE);
gf_bs_write_u32(bs, 0);
gf_odf_desc_write_bs((GF_Descriptor *) ctx->esd, bs);
gf_bs_get_content(bs, &dsi_data, &dsi_data_size);
gf_bs_del(bs);
ctx->mime = "video/mp4v-es";
char *esds = (char *)malloc(dsi_data_size);
memcpy(esds, dsi_data, dsi_data_size);
esds[0] = 0x00;
esds[1] = 0x00;
esds[2] = 0x00;
esds[3] = 0x01;
AMediaFormat_setBuffer(ctx->format, "csd-0", esds, dsi_data_size);
gf_free(dsi_data);
return GF_OK;
}
return GF_NOT_SUPPORTED;
}
static GF_Err XVID_AttachStream(GF_BaseDecoder *ifcg, GF_ESD *esd)
{
GF_M4VDecSpecInfo dsi;
GF_Err e;
unsigned char *ptr;
unsigned long pitch;
XVIDCTX();
if (ctx->ES_ID && ctx->ES_ID!=esd->ESID) return GF_NOT_SUPPORTED;
if (!esd->decoderConfig->decoderSpecificInfo || !esd->decoderConfig->decoderSpecificInfo->data) return GF_NON_COMPLIANT_BITSTREAM;
/*decode DSI*/
e = gf_m4v_get_config((char *) esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength, &dsi);
if (e) return e;
if (!dsi.width || !dsi.height) return GF_NON_COMPLIANT_BITSTREAM;
GF_LOG(GF_LOG_DEBUG, GF_LOG_CODEC, ("[XviD] Attaching Stream %d - framesize %d x %d\n", esd->ESID, dsi.width, dsi.height ));
ctx->codec = InitCodec(dsi.width, dsi.height, GF_4CC('x', 'v', 'i', 'd'));
if (!ctx->codec) return GF_OUT_OF_MEM;
GF_LOG(GF_LOG_DEBUG, GF_LOG_CODEC, ("[XviD] Decoding DecoderSpecificInfo\n"));
DecodeFrame(ctx->codec, esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength, ptr, ptr, ptr, pitch);
/*note that this may be irrelevant when used through systems (FPS is driven by systems CTS)*/
ctx->FPS = dsi.clock_rate;
ctx->FPS /= 1000;
if (!ctx->FPS) ctx->FPS = 30.0f;
ctx->width = dsi.width;
ctx->height = dsi.height;
ctx->pixel_ar = (dsi.par_num<<16) | dsi.par_den;
ctx->pixel_ar = 0;
ctx->ES_ID = esd->ESID;
ctx->first_frame = 1;
/*output in YV12 only - let the player handle conversion*/
ctx->out_size = 3 * ctx->width * ctx->height / 2;
GF_LOG(GF_LOG_DEBUG, GF_LOG_CODEC, ("[XviD] Decoder setup - output size %d\n", ctx->out_size ));
return GF_OK;
}
GF_EXPORT
GF_RTPStreamer *gf_rtp_streamer_new_extended(u32 streamType, u32 oti, u32 timeScale,
const char *ip_dest, u16 port, u32 MTU, u8 TTL, const char *ifce_addr,
u32 flags, char *dsi, u32 dsi_len,
u32 PayloadType, u32 sample_rate, u32 nb_ch,
Bool is_crypted, u32 IV_length, u32 KI_length,
u32 MinSize, u32 MaxSize, u32 avgTS, u32 maxDTSDelta, u32 const_dur, u32 bandwidth, u32 max_ptime,
u32 au_sn_len
)
{
GF_SLConfig slc;
GF_RTPStreamer *stream;
u32 rtp_type, default_rtp_rate;
u8 OfficialPayloadType;
u32 required_rate, force_dts_delta, PL_ID;
char *mpeg4mode;
Bool has_mpeg4_mapping;
GF_Err e;
if (!timeScale) timeScale = 1000;
GF_SAFEALLOC(stream, GF_RTPStreamer);
/*by default NO PL signaled*/
PL_ID = 0;
OfficialPayloadType = 0;
force_dts_delta = 0;
mpeg4mode = NULL;
required_rate = 0;
nb_ch = 0;
has_mpeg4_mapping = 1;
rtp_type = 0;
/*for max compatibility with QT*/
default_rtp_rate = 90000;
/*timed-text is a bit special, we support multiple stream descriptions & co*/
switch (streamType) {
case GF_STREAM_TEXT:
if (oti!=GPAC_OTI_TEXT_MPEG4)
return NULL;
rtp_type = GF_RTP_PAYT_3GPP_TEXT;
/*fixme - this works cos there's only one PL for text in mpeg4 at the current time*/
PL_ID = 0x10;
break;
case GF_STREAM_AUDIO:
required_rate = sample_rate;
switch (oti) {
/*AAC*/
case GPAC_OTI_AUDIO_AAC_MPEG4:
case GPAC_OTI_AUDIO_AAC_MPEG2_MP:
case GPAC_OTI_AUDIO_AAC_MPEG2_LCP:
case GPAC_OTI_AUDIO_AAC_MPEG2_SSRP:
PL_ID = 0x01;
mpeg4mode = "AAC";
rtp_type = GF_RTP_PAYT_MPEG4;
required_rate = sample_rate;
#ifndef GPAC_DISABLE_AV_PARSERS
if (dsi) {
GF_M4ADecSpecInfo a_cfg;
gf_m4a_get_config(dsi, dsi_len, &a_cfg);
nb_ch = a_cfg.nb_chan;
sample_rate = a_cfg.base_sr;
PL_ID = a_cfg.audioPL;
switch (a_cfg.base_object_type) {
case GF_M4A_AAC_MAIN:
case GF_M4A_AAC_LC:
if (flags & GP_RTP_PCK_USE_LATM_AAC) {
rtp_type = GF_RTP_PAYT_LATM;
break;
}
case GF_M4A_AAC_SBR:
case GF_M4A_AAC_PS:
case GF_M4A_AAC_LTP:
case GF_M4A_AAC_SCALABLE:
case GF_M4A_ER_AAC_LC:
case GF_M4A_ER_AAC_LTP:
case GF_M4A_ER_AAC_SCALABLE:
mpeg4mode = "AAC";
break;
case GF_M4A_CELP:
case GF_M4A_ER_CELP:
mpeg4mode = "CELP";
break;
}
}
#endif
break;
/*MPEG1/2 audio*/
case GPAC_OTI_AUDIO_MPEG2_PART3:
case GPAC_OTI_AUDIO_MPEG1:
if (!is_crypted) {
rtp_type = GF_RTP_PAYT_MPEG12_AUDIO;
/*use official RTP/AVP payload type*/
OfficialPayloadType = 14;
required_rate = 90000;
}
/*encrypted MP3 must be sent through MPEG-4 generic to signal all ISMACryp stuff*/
else {
rtp_type = GF_RTP_PAYT_MPEG4;
}
break;
/*QCELP audio*/
case GPAC_OTI_AUDIO_13K_VOICE:
rtp_type = GF_RTP_PAYT_QCELP;
OfficialPayloadType = 12;
required_rate = 8000;
nb_ch = 1;
break;
/*EVRC/SVM audio*/
case GPAC_OTI_AUDIO_EVRC_VOICE:
case GPAC_OTI_AUDIO_SMV_VOICE:
rtp_type = GF_RTP_PAYT_EVRC_SMV;
required_rate = 8000;
nb_ch = 1;
}
break;
case GF_STREAM_VISUAL:
rtp_type = GF_RTP_PAYT_MPEG4;
required_rate = default_rtp_rate;
if (is_crypted) {
/*that's another pain with ISMACryp, even if no B-frames the DTS is signaled...*/
if (oti==GPAC_OTI_VIDEO_MPEG4_PART2) force_dts_delta = 22;
flags |= GP_RTP_PCK_SIGNAL_RAP | GP_RTP_PCK_SIGNAL_TS;
}
switch (oti) {
/*ISO/IEC 14496-2*/
case GPAC_OTI_VIDEO_MPEG4_PART2:
PL_ID = 1;
#ifndef GPAC_DISABLE_AV_PARSERS
if (dsi) {
GF_M4VDecSpecInfo vhdr;
gf_m4v_get_config(dsi, dsi_len, &vhdr);
PL_ID = vhdr.VideoPL;
}
#endif
break;
/*MPEG1/2 video*/
case GPAC_OTI_VIDEO_MPEG1:
case GPAC_OTI_VIDEO_MPEG2_SIMPLE:
case GPAC_OTI_VIDEO_MPEG2_MAIN:
case GPAC_OTI_VIDEO_MPEG2_SNR:
case GPAC_OTI_VIDEO_MPEG2_SPATIAL:
case GPAC_OTI_VIDEO_MPEG2_HIGH:
case GPAC_OTI_VIDEO_MPEG2_422:
if (!is_crypted) {
rtp_type = GF_RTP_PAYT_MPEG12_VIDEO;
OfficialPayloadType = 32;
}
break;
/*AVC/H.264*/
case GPAC_OTI_VIDEO_AVC:
required_rate = 90000; /* "90 kHz clock rate MUST be used"*/
rtp_type = GF_RTP_PAYT_H264_AVC;
PL_ID = 0x0F;
break;
/*H264-SVC*/
case GPAC_OTI_VIDEO_SVC:
required_rate = 90000; /* "90 kHz clock rate MUST be used"*/
rtp_type = GF_RTP_PAYT_H264_SVC;
PL_ID = 0x0F;
break;
/*HEVC*/
case GPAC_OTI_VIDEO_HEVC:
required_rate = 90000; /* "90 kHz clock rate MUST be used"*/
rtp_type = GF_RTP_PAYT_HEVC;
PL_ID = 0x0F;
break;
}
break;
case GF_STREAM_SCENE:
case GF_STREAM_OD:
if (oti == GPAC_OTI_SCENE_DIMS) {
rtp_type = GF_RTP_PAYT_3GPP_DIMS;
has_mpeg4_mapping = 0;
} else {
rtp_type = GF_RTP_PAYT_MPEG4;
}
break;
case GF_STREAM_4CC:
switch (oti) {
case GF_ISOM_SUBTYPE_3GP_H263:
rtp_type = GF_RTP_PAYT_H263;
required_rate = 90000;
streamType = GF_STREAM_VISUAL;
OfficialPayloadType = 34;
/*not 100% compliant (short header is missing) but should still work*/
oti = GPAC_OTI_VIDEO_MPEG4_PART2;
PL_ID = 0x01;
break;
case GF_ISOM_SUBTYPE_3GP_AMR:
required_rate = 8000;
rtp_type = GF_RTP_PAYT_AMR;
streamType = GF_STREAM_AUDIO;
has_mpeg4_mapping = 0;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_3GP_AMR_WB:
required_rate = 16000;
rtp_type = GF_RTP_PAYT_AMR_WB;
streamType = GF_STREAM_AUDIO;
has_mpeg4_mapping = 0;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_AC3:
rtp_type = GF_RTP_PAYT_AC3;
streamType = GF_STREAM_AUDIO;
has_mpeg4_mapping = 1;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_AVC_H264:
case GF_ISOM_SUBTYPE_AVC2_H264:
case GF_ISOM_SUBTYPE_AVC3_H264:
case GF_ISOM_SUBTYPE_AVC4_H264:
case GF_ISOM_SUBTYPE_SVC_H264:
{
required_rate = 90000; /* "90 kHz clock rate MUST be used"*/
rtp_type = GF_RTP_PAYT_H264_AVC;
streamType = GF_STREAM_VISUAL;
oti = GPAC_OTI_VIDEO_AVC;
PL_ID = 0x0F;
}
break;
case GF_ISOM_SUBTYPE_3GP_QCELP:
required_rate = 8000;
rtp_type = GF_RTP_PAYT_QCELP;
streamType = GF_STREAM_AUDIO;
oti = GPAC_OTI_AUDIO_13K_VOICE;
OfficialPayloadType = 12;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_3GP_EVRC:
case GF_ISOM_SUBTYPE_3GP_SMV:
required_rate = 8000;
rtp_type = GF_RTP_PAYT_EVRC_SMV;
streamType = GF_STREAM_AUDIO;
oti = (oti==GF_ISOM_SUBTYPE_3GP_EVRC) ? GPAC_OTI_AUDIO_EVRC_VOICE : GPAC_OTI_AUDIO_SMV_VOICE;
nb_ch = 1;
break;
}
break;
default:
GF_LOG(GF_LOG_ERROR, GF_LOG_RTP, ("[RTP Packetizer] Unsupported stream type %x\n", streamType));
return NULL;
}
/*not supported*/
if (!rtp_type) return NULL;
/*override hinter type if requested and possible*/
if (has_mpeg4_mapping && (flags & GP_RTP_PCK_FORCE_MPEG4)) {
rtp_type = GF_RTP_PAYT_MPEG4;
}
/*use static payload ID if enabled*/
else if (OfficialPayloadType && (flags & GP_RTP_PCK_USE_STATIC_ID) ) {
PayloadType = OfficialPayloadType;
}
/*systems carousel: we need at least IDX and RAP signaling*/
if (flags & GP_RTP_PCK_SYSTEMS_CAROUSEL) {
flags |= GP_RTP_PCK_SIGNAL_RAP;
}
/*update flags in MultiSL*/
if (flags & GP_RTP_PCK_USE_MULTI) {
if (MinSize != MaxSize) flags |= GP_RTP_PCK_SIGNAL_SIZE;
if (!const_dur) flags |= GP_RTP_PCK_SIGNAL_TS;
}
/*default SL for RTP */
memset(&slc, 0, sizeof(GF_SLConfig));
slc.tag = GF_ODF_SLC_TAG;
slc.useTimestampsFlag = 1;
slc.timestampLength = 32;
slc.timestampResolution = timeScale;
/*override clockrate if set*/
if (required_rate) {
Double sc = required_rate;
sc /= slc.timestampResolution;
maxDTSDelta = (u32) (maxDTSDelta*sc);
slc.timestampResolution = required_rate;
}
/*switch to RTP TS*/
max_ptime = (u32) (max_ptime * slc.timestampResolution / 1000);
slc.AUSeqNumLength = au_sn_len;
slc.CUDuration = const_dur;
if (flags & GP_RTP_PCK_SIGNAL_RAP) {
slc.useRandomAccessPointFlag = 1;
} else {
slc.useRandomAccessPointFlag = 0;
slc.hasRandomAccessUnitsOnlyFlag = 1;
}
stream->packetizer = gf_rtp_builder_new(rtp_type, &slc, flags,
stream,
rtp_stream_on_new_packet, rtp_stream_on_packet_done,
NULL, rtp_stream_on_data);
if (!stream->packetizer) {
GF_LOG(GF_LOG_ERROR, GF_LOG_RTP, ("[RTP Packetizer] Failed to create packetizer\n"));
gf_free(stream);
return NULL;
}
gf_rtp_builder_init(stream->packetizer, PayloadType, MTU, max_ptime,
streamType, oti, PL_ID, MinSize, MaxSize, avgTS, maxDTSDelta, IV_length, KI_length, mpeg4mode);
if (force_dts_delta) stream->packetizer->slMap.DTSDeltaLength = force_dts_delta;
e = rtp_stream_init_channel(stream, MTU + 12, ip_dest, port, TTL, ifce_addr);
if (e) {
GF_LOG(GF_LOG_ERROR, GF_LOG_RTP, ("[RTP Packetizer] Failed to create RTP channel - error %s\n", gf_error_to_string(e) ));
gf_free(stream);
return NULL;
}
stream->ts_scale = slc.timestampResolution;
stream->ts_scale /= timeScale;
stream->buffer_alloc = MTU+12;
stream->buffer = gf_malloc(sizeof(char) * stream->buffer_alloc);
return stream;
}
static GF_Err XVID_AttachStream(GF_BaseDecoder *ifcg, GF_ESD *esd)
{
GF_M4VDecSpecInfo dsi;
GF_Err e;
void **codec;
#ifdef XVID_USE_OLD_API
XVID_DEC_FRAME frame;
XVID_DEC_PARAM par;
#else
xvid_dec_frame_t frame;
xvid_dec_create_t par;
#endif
XVIDCTX();
if (!esd->decoderConfig->decoderSpecificInfo || !esd->decoderConfig->decoderSpecificInfo->data) return GF_NON_COMPLIANT_BITSTREAM;
/*locate any auxiliary video data descriptor on this stream*/
if (esd->dependsOnESID) {
u32 i = 0;
GF_Descriptor *d = NULL;
if (esd->dependsOnESID != ctx->base_ES_ID) return GF_NOT_SUPPORTED;
#ifdef XVID_USE_OLD_API
return GF_NOT_SUPPORTED;
#endif
while ((d = gf_list_enum(esd->extensionDescriptors, &i))) {
if (d->tag == GF_ODF_AUX_VIDEO_DATA) break;
}
if (!d) return GF_NOT_SUPPORTED;
codec = &ctx->depth_codec;
ctx->depth_ES_ID = esd->ESID;
} else {
if (ctx->base_ES_ID && ctx->base_ES_ID!=esd->ESID) return GF_NOT_SUPPORTED;
codec = &ctx->base_codec;
ctx->base_ES_ID = esd->ESID;
}
if (*codec) xvid_decore(*codec, XVID_DEC_DESTROY, NULL, NULL);
/*decode DSI*/
e = gf_m4v_get_config(esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength, &dsi);
if (e) return e;
if (!dsi.width || !dsi.height) return GF_NON_COMPLIANT_BITSTREAM;
memset(&par, 0, sizeof(par));
par.width = dsi.width;
par.height = dsi.height;
/*note that this may be irrelevant when used through systems (FPS is driven by systems CTS)*/
ctx->FPS = dsi.clock_rate;
ctx->FPS /= 1000;
if (!ctx->FPS) ctx->FPS = 30.0f;
ctx->pixel_ar = (dsi.par_num<<16) | dsi.par_den;
#ifndef XVID_USE_OLD_API
par.version = XVID_VERSION;
#endif
if (xvid_decore(NULL, XVID_DEC_CREATE, &par, NULL) < 0) return GF_NON_COMPLIANT_BITSTREAM;
ctx->width = par.width;
ctx->height = par.height;
*codec = par.handle;
/*init decoder*/
memset(&frame, 0, sizeof(frame));
frame.bitstream = (void *) esd->decoderConfig->decoderSpecificInfo->data;
frame.length = esd->decoderConfig->decoderSpecificInfo->dataLength;
#ifndef XVID_USE_OLD_API
frame.version = XVID_VERSION;
xvid_decore(*codec, XVID_DEC_DECODE, &frame, NULL);
#else
/*don't perform error check, XviD doesn't like DSI only frame ...*/
xvid_decore(*codec, XVID_DEC_DECODE, &frame, NULL);
#endif
ctx->first_frame = 1;
/*output in YV12 only - let the player handle conversion*/
if (ctx->depth_codec) {
ctx->out_size = ctx->width * ctx->height * 5 / 2;
ctx->temp_uv = gf_malloc(sizeof(char)*ctx->width * ctx->height / 2);
} else {
ctx->yuv_size = ctx->out_size = ctx->width * ctx->height * 3 / 2;
}
return GF_OK;
}
GF_EXPORT
GF_RTPHinter *gf_hinter_track_new(GF_ISOFile *file, u32 TrackNum,
u32 Path_MTU, u32 max_ptime, u32 default_rtp_rate, u32 flags, u8 PayloadID,
Bool copy_media, u32 InterleaveGroupID, u8 InterleaveGroupPriority, GF_Err *e)
{
GF_SLConfig my_sl;
u32 descIndex, MinSize, MaxSize, avgTS, streamType, oti, const_dur, nb_ch, maxDTSDelta;
u8 OfficialPayloadID;
u32 TrackMediaSubType, TrackMediaType, hintType, nbEdts, required_rate, force_dts_delta, avc_nalu_size, PL_ID, bandwidth, IV_length, KI_length;
const char *url, *urn;
char *mpeg4mode;
Bool is_crypted, has_mpeg4_mapping;
GF_RTPHinter *tmp;
GF_ESD *esd;
*e = GF_BAD_PARAM;
if (!file || !TrackNum || !gf_isom_get_track_id(file, TrackNum)) return NULL;
if (!gf_isom_get_sample_count(file, TrackNum)) {
*e = GF_OK;
return NULL;
}
*e = GF_NOT_SUPPORTED;
nbEdts = gf_isom_get_edit_segment_count(file, TrackNum);
if (nbEdts>1) {
u64 et, sd, mt;
u8 em;
gf_isom_get_edit_segment(file, TrackNum, 1, &et, &sd, &mt, &em);
if ((nbEdts>2) || (em!=GF_ISOM_EDIT_EMPTY)) {
GF_LOG(GF_LOG_ERROR, GF_LOG_RTP, ("[rtp hinter] Cannot hint track whith EditList\n"));
return NULL;
}
}
if (nbEdts) gf_isom_remove_edit_segments(file, TrackNum);
if (!gf_isom_is_track_enabled(file, TrackNum)) return NULL;
/*by default NO PL signaled*/
PL_ID = 0;
OfficialPayloadID = 0;
force_dts_delta = 0;
streamType = oti = 0;
mpeg4mode = NULL;
required_rate = 0;
is_crypted = 0;
IV_length = KI_length = 0;
oti = 0;
nb_ch = 0;
avc_nalu_size = 0;
has_mpeg4_mapping = 1;
TrackMediaType = gf_isom_get_media_type(file, TrackNum);
TrackMediaSubType = gf_isom_get_media_subtype(file, TrackNum, 1);
/*for max compatibility with QT*/
if (!default_rtp_rate) default_rtp_rate = 90000;
/*timed-text is a bit special, we support multiple stream descriptions & co*/
if ( (TrackMediaType==GF_ISOM_MEDIA_TEXT) || (TrackMediaType==GF_ISOM_MEDIA_SUBT)) {
hintType = GF_RTP_PAYT_3GPP_TEXT;
oti = GPAC_OTI_TEXT_MPEG4;
streamType = GF_STREAM_TEXT;
/*fixme - this works cos there's only one PL for text in mpeg4 at the current time*/
PL_ID = 0x10;
} else {
if (gf_isom_get_sample_description_count(file, TrackNum) > 1) return NULL;
TrackMediaSubType = gf_isom_get_media_subtype(file, TrackNum, 1);
switch (TrackMediaSubType) {
case GF_ISOM_SUBTYPE_MPEG4_CRYP:
is_crypted = 1;
case GF_ISOM_SUBTYPE_MPEG4:
esd = gf_isom_get_esd(file, TrackNum, 1);
hintType = GF_RTP_PAYT_MPEG4;
if (esd) {
streamType = esd->decoderConfig->streamType;
oti = esd->decoderConfig->objectTypeIndication;
if (esd->URLString) hintType = 0;
/*AAC*/
if ((streamType==GF_STREAM_AUDIO) && esd->decoderConfig->decoderSpecificInfo
/*(nb: we use mpeg4 for MPEG-2 AAC)*/
&& ((oti==GPAC_OTI_AUDIO_AAC_MPEG4) || (oti==GPAC_OTI_AUDIO_AAC_MPEG4) || (oti==GPAC_OTI_AUDIO_AAC_MPEG2_MP) || (oti==GPAC_OTI_AUDIO_AAC_MPEG2_LCP) || (oti==GPAC_OTI_AUDIO_AAC_MPEG2_SSRP)) ) {
u32 sample_rate;
GF_M4ADecSpecInfo a_cfg;
gf_m4a_get_config(esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength, &a_cfg);
nb_ch = a_cfg.nb_chan;
sample_rate = a_cfg.base_sr;
PL_ID = a_cfg.audioPL;
switch (a_cfg.base_object_type) {
case GF_M4A_AAC_MAIN:
case GF_M4A_AAC_LC:
if (flags & GP_RTP_PCK_USE_LATM_AAC) {
hintType = GF_RTP_PAYT_LATM;
break;
}
case GF_M4A_AAC_SBR:
case GF_M4A_AAC_PS:
case GF_M4A_AAC_LTP:
case GF_M4A_AAC_SCALABLE:
case GF_M4A_ER_AAC_LC:
case GF_M4A_ER_AAC_LTP:
case GF_M4A_ER_AAC_SCALABLE:
mpeg4mode = "AAC";
break;
case GF_M4A_CELP:
case GF_M4A_ER_CELP:
mpeg4mode = "CELP";
break;
}
required_rate = sample_rate;
}
/*MPEG1/2 audio*/
else if ((streamType==GF_STREAM_AUDIO) && ((oti==GPAC_OTI_AUDIO_MPEG2_PART3) || (oti==GPAC_OTI_AUDIO_MPEG1))) {
u32 sample_rate;
if (!is_crypted) {
GF_ISOSample *samp = gf_isom_get_sample(file, TrackNum, 1, NULL);
u32 hdr = GF_4CC((u8)samp->data[0], (u8)samp->data[1], (u8)samp->data[2], (u8)samp->data[3]);
nb_ch = gf_mp3_num_channels(hdr);
sample_rate = gf_mp3_sampling_rate(hdr);
gf_isom_sample_del(&samp);
hintType = GF_RTP_PAYT_MPEG12_AUDIO;
/*use official RTP/AVP payload type*/
OfficialPayloadID = 14;
required_rate = 90000;
}
/*encrypted MP3 must be sent through MPEG-4 generic to signal all ISMACryp stuff*/
else {
u8 bps;
gf_isom_get_audio_info(file, TrackNum, 1, &sample_rate, &nb_ch, &bps);
required_rate = sample_rate;
}
}
/*QCELP audio*/
else if ((streamType==GF_STREAM_AUDIO) && (oti==GPAC_OTI_AUDIO_13K_VOICE)) {
hintType = GF_RTP_PAYT_QCELP;
OfficialPayloadID = 12;
required_rate = 8000;
streamType = GF_STREAM_AUDIO;
nb_ch = 1;
}
/*EVRC/SVM audio*/
else if ((streamType==GF_STREAM_AUDIO) && ((oti==GPAC_OTI_AUDIO_EVRC_VOICE) || (oti==GPAC_OTI_AUDIO_SMV_VOICE)) ) {
hintType = GF_RTP_PAYT_EVRC_SMV;
required_rate = 8000;
streamType = GF_STREAM_AUDIO;
nb_ch = 1;
}
/*visual streams*/
else if (streamType==GF_STREAM_VISUAL) {
if (oti==GPAC_OTI_VIDEO_MPEG4_PART2) {
GF_M4VDecSpecInfo dsi;
gf_m4v_get_config(esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength, &dsi);
PL_ID = dsi.VideoPL;
}
/*MPEG1/2 video*/
if ( ((oti>=GPAC_OTI_VIDEO_MPEG2_SIMPLE) && (oti<=GPAC_OTI_VIDEO_MPEG2_422)) || (oti==GPAC_OTI_VIDEO_MPEG1)) {
if (!is_crypted) {
hintType = GF_RTP_PAYT_MPEG12_VIDEO;
OfficialPayloadID = 32;
}
}
/*for ISMA*/
if (is_crypted) {
/*that's another pain with ISMACryp, even if no B-frames the DTS is signaled...*/
if (oti==GPAC_OTI_VIDEO_MPEG4_PART2) force_dts_delta = 22;
else if (oti==GPAC_OTI_VIDEO_AVC) {
flags &= ~GP_RTP_PCK_USE_MULTI;
force_dts_delta = 22;
}
flags |= GP_RTP_PCK_SIGNAL_RAP | GP_RTP_PCK_SIGNAL_TS;
}
required_rate = default_rtp_rate;
}
/*systems streams*/
else if (gf_isom_has_sync_shadows(file, TrackNum) || gf_isom_has_sample_dependency(file, TrackNum)) {
flags |= GP_RTP_PCK_SYSTEMS_CAROUSEL;
}
gf_odf_desc_del((GF_Descriptor*)esd);
}
break;
case GF_ISOM_SUBTYPE_3GP_H263:
hintType = GF_RTP_PAYT_H263;
required_rate = 90000;
streamType = GF_STREAM_VISUAL;
OfficialPayloadID = 34;
/*not 100% compliant (short header is missing) but should still work*/
oti = GPAC_OTI_VIDEO_MPEG4_PART2;
PL_ID = 0x01;
break;
case GF_ISOM_SUBTYPE_3GP_AMR:
required_rate = 8000;
hintType = GF_RTP_PAYT_AMR;
streamType = GF_STREAM_AUDIO;
has_mpeg4_mapping = 0;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_3GP_AMR_WB:
required_rate = 16000;
hintType = GF_RTP_PAYT_AMR_WB;
streamType = GF_STREAM_AUDIO;
has_mpeg4_mapping = 0;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_AVC_H264:
case GF_ISOM_SUBTYPE_AVC2_H264:
case GF_ISOM_SUBTYPE_SVC_H264:
{
GF_AVCConfig *avcc = gf_isom_avc_config_get(file, TrackNum, 1);
required_rate = 90000; /* "90 kHz clock rate MUST be used"*/
hintType = GF_RTP_PAYT_H264_AVC;
streamType = GF_STREAM_VISUAL;
avc_nalu_size = avcc->nal_unit_size;
oti = GPAC_OTI_VIDEO_AVC;
PL_ID = 0x0F;
gf_odf_avc_cfg_del(avcc);
}
break;
case GF_ISOM_SUBTYPE_3GP_QCELP:
required_rate = 8000;
hintType = GF_RTP_PAYT_QCELP;
streamType = GF_STREAM_AUDIO;
oti = GPAC_OTI_AUDIO_13K_VOICE;
OfficialPayloadID = 12;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_3GP_EVRC:
case GF_ISOM_SUBTYPE_3GP_SMV:
required_rate = 8000;
hintType = GF_RTP_PAYT_EVRC_SMV;
streamType = GF_STREAM_AUDIO;
oti = (TrackMediaSubType==GF_ISOM_SUBTYPE_3GP_EVRC) ? GPAC_OTI_AUDIO_EVRC_VOICE : GPAC_OTI_AUDIO_SMV_VOICE;
nb_ch = 1;
break;
case GF_ISOM_SUBTYPE_3GP_DIMS:
hintType = GF_RTP_PAYT_3GPP_DIMS;
streamType = GF_STREAM_SCENE;
break;
case GF_ISOM_SUBTYPE_AC3:
hintType = GF_RTP_PAYT_AC3;
streamType = GF_STREAM_AUDIO;
gf_isom_get_audio_info(file, TrackNum, 1, NULL, &nb_ch, NULL);
break;
default:
/*ERROR*/
hintType = 0;
break;
}
}
/*not hintable*/
if (!hintType) return NULL;
/*we only support self-contained files for hinting*/
gf_isom_get_data_reference(file, TrackNum, 1, &url, &urn);
if (url || urn) return NULL;
*e = GF_OUT_OF_MEM;
GF_SAFEALLOC(tmp, GF_RTPHinter);
if (!tmp) return NULL;
/*override hinter type if requested and possible*/
if (has_mpeg4_mapping && (flags & GP_RTP_PCK_FORCE_MPEG4)) {
hintType = GF_RTP_PAYT_MPEG4;
avc_nalu_size = 0;
}
/*use static payload ID if enabled*/
else if (OfficialPayloadID && (flags & GP_RTP_PCK_USE_STATIC_ID) ) {
PayloadID = OfficialPayloadID;
}
tmp->file = file;
tmp->TrackNum = TrackNum;
tmp->avc_nalu_size = avc_nalu_size;
tmp->nb_chan = nb_ch;
/*spatial scalability check*/
tmp->has_ctts = gf_isom_has_time_offset(file, TrackNum);
/*get sample info*/
gf_media_get_sample_average_infos(file, TrackNum, &MinSize, &MaxSize, &avgTS, &maxDTSDelta, &const_dur, &bandwidth);
/*systems carousel: we need at least IDX and RAP signaling*/
if (flags & GP_RTP_PCK_SYSTEMS_CAROUSEL) {
flags |= GP_RTP_PCK_SIGNAL_RAP;
}
/*update flags in MultiSL*/
if (flags & GP_RTP_PCK_USE_MULTI) {
if (MinSize != MaxSize) flags |= GP_RTP_PCK_SIGNAL_SIZE;
if (!const_dur) flags |= GP_RTP_PCK_SIGNAL_TS;
}
if (tmp->has_ctts) flags |= GP_RTP_PCK_SIGNAL_TS;
/*default SL for RTP */
InitSL_RTP(&my_sl);
my_sl.timestampResolution = gf_isom_get_media_timescale(file, TrackNum);
/*override clockrate if set*/
if (required_rate) {
Double sc = required_rate;
sc /= my_sl.timestampResolution;
maxDTSDelta = (u32) (maxDTSDelta*sc);
my_sl.timestampResolution = required_rate;
}
/*switch to RTP TS*/
max_ptime = (u32) (max_ptime * my_sl.timestampResolution / 1000);
my_sl.AUSeqNumLength = gf_get_bit_size(gf_isom_get_sample_count(file, TrackNum));
my_sl.CUDuration = const_dur;
if (gf_isom_has_sync_points(file, TrackNum)) {
my_sl.useRandomAccessPointFlag = 1;
} else {
my_sl.useRandomAccessPointFlag = 0;
my_sl.hasRandomAccessUnitsOnlyFlag = 1;
}
if (is_crypted) {
Bool use_sel_enc;
gf_isom_get_ismacryp_info(file, TrackNum, 1, NULL, NULL, NULL, NULL, NULL, &use_sel_enc, &IV_length, &KI_length);
if (use_sel_enc) flags |= GP_RTP_PCK_SELECTIVE_ENCRYPTION;
}
// in case a different timescale was provided
tmp->OrigTimeScale = gf_isom_get_media_timescale(file, TrackNum);
tmp->rtp_p = gf_rtp_builder_new(hintType, &my_sl, flags, tmp,
MP4T_OnNewPacket, MP4T_OnPacketDone,
/*if copy, no data ref*/
copy_media ? NULL : MP4T_OnDataRef,
MP4T_OnData);
//init the builder
gf_rtp_builder_init(tmp->rtp_p, PayloadID, Path_MTU, max_ptime,
streamType, oti, PL_ID, MinSize, MaxSize, avgTS, maxDTSDelta, IV_length, KI_length, mpeg4mode);
/*ISMA compliance is a pain...*/
if (force_dts_delta) tmp->rtp_p->slMap.DTSDeltaLength = force_dts_delta;
/* Hint Track Setup */
tmp->TrackID = gf_isom_get_track_id(file, TrackNum);
tmp->HintID = tmp->TrackID + 65535;
while (gf_isom_get_track_by_id(file, tmp->HintID)) tmp->HintID++;
tmp->HintTrack = gf_isom_new_track(file, tmp->HintID, GF_ISOM_MEDIA_HINT, my_sl.timestampResolution);
gf_isom_setup_hint_track(file, tmp->HintTrack, GF_ISOM_HINT_RTP);
/*create a hint description*/
gf_isom_new_hint_description(file, tmp->HintTrack, -1, -1, 0, &descIndex);
gf_isom_rtp_set_timescale(file, tmp->HintTrack, descIndex, my_sl.timestampResolution);
if (hintType==GF_RTP_PAYT_MPEG4) {
tmp->rtp_p->slMap.ObjectTypeIndication = oti;
/*set this SL for extraction.*/
gf_isom_set_extraction_slc(file, TrackNum, 1, &my_sl);
}
tmp->bandwidth = bandwidth;
/*set interleaving*/
gf_isom_set_track_group(file, TrackNum, InterleaveGroupID);
if (!copy_media) {
/*if we don't copy data set hint track and media track in the same group*/
gf_isom_set_track_group(file, tmp->HintTrack, InterleaveGroupID);
} else {
gf_isom_set_track_group(file, tmp->HintTrack, InterleaveGroupID + OFFSET_HINT_GROUP_ID);
}
/*use user-secified priority*/
InterleaveGroupPriority*=2;
gf_isom_set_track_priority_in_group(file, TrackNum, InterleaveGroupPriority+1);
gf_isom_set_track_priority_in_group(file, tmp->HintTrack, InterleaveGroupPriority);
#if 0
/*QT FF: not setting these flags = server uses a random offset*/
gf_isom_rtp_set_time_offset(file, tmp->HintTrack, 1, 0);
/*we don't use seq offset for maintainance pruposes*/
gf_isom_rtp_set_time_sequence_offset(file, tmp->HintTrack, 1, 0);
#endif
*e = GF_OK;
return tmp;
}
static GF_Err MCDec_AttachStream(GF_BaseDecoder *ifcg, GF_ESD *esd)
{
MCDec *ctx = (MCDec *)ifcg->privateStack;
ctx->esd = esd;
GF_Err e;
//check AVC config
if (esd->decoderConfig->objectTypeIndication == GPAC_OTI_VIDEO_AVC) {
ctx->SPSs = gf_list_new();
ctx->PPSs = gf_list_new();
ctx->mime = "video/avc";
ctx->avc.sps_active_idx = -1;
ctx->active_sps = ctx->active_pps = -1;
if (!esd->decoderConfig->decoderSpecificInfo || !esd->decoderConfig->decoderSpecificInfo->data) {
ctx->width=ctx->height=128;
ctx->out_size = ctx->width*ctx->height*3/2;
ctx->pix_fmt = GF_PIXEL_NV12;
return GF_OK;
} else {
u32 i;
GF_AVCConfigSlot *slc;
GF_AVCConfig *cfg = gf_odf_avc_cfg_read(esd->decoderConfig->decoderSpecificInfo->data, esd->decoderConfig->decoderSpecificInfo->dataLength);
for (i = 0; i<gf_list_count(cfg->sequenceParameterSets); i++) {
slc = gf_list_get(cfg->sequenceParameterSets, i);
slc->id = -1;
MCDec_RegisterParameterSet(ctx, slc->data, slc->size, GF_TRUE);
}
for (i = 0; i<gf_list_count(cfg->pictureParameterSets); i++) {
slc = gf_list_get(cfg->pictureParameterSets, i);
slc->id = -1;
MCDec_RegisterParameterSet(ctx, slc->data, slc->size, GF_FALSE);
}
slc = gf_list_get(ctx->SPSs, 0);
if (slc) ctx->active_sps = slc->id;
slc = gf_list_get(ctx->PPSs, 0);
if (slc) ctx->active_pps = slc->id;
ctx->nalu_size_length = cfg->nal_unit_size;
if (gf_list_count(ctx->SPSs) && gf_list_count(ctx->PPSs)) {
e = MCDec_InitDecoder(ctx);
}
else {
e = GF_OK;
}
gf_odf_avc_cfg_del(cfg);
return e;
}
}
if (esd->decoderConfig->objectTypeIndication == GPAC_OTI_VIDEO_MPEG4_PART2) {
if (!esd->decoderConfig->decoderSpecificInfo || !esd->decoderConfig->decoderSpecificInfo->data) {
ctx->width=ctx->height=128;
ctx->out_size = ctx->width*ctx->height*3/2;
ctx->pix_fmt = GF_PIXEL_NV12;
} else {
GF_M4VDecSpecInfo vcfg;
gf_m4v_get_config(ctx->esd->decoderConfig->decoderSpecificInfo->data, ctx->esd->decoderConfig->decoderSpecificInfo->dataLength, &vcfg);
ctx->width = vcfg.width;
ctx->height = vcfg.height;
ctx->out_size = ctx->width*ctx->height*3/2;
ctx->pix_fmt = GF_PIXEL_NV12;
return MCDec_InitDecoder(ctx);
}
}
if (esd->decoderConfig->objectTypeIndication == GPAC_OTI_VIDEO_HEVC) {
ctx->esd= esd;
if (!esd->decoderConfig->decoderSpecificInfo || !esd->decoderConfig->decoderSpecificInfo->data) {
ctx->width=ctx->height=128;
ctx->out_size = ctx->width*ctx->height*3/2;
ctx->pix_fmt = GF_PIXEL_NV12;
} else {
return MCDec_InitDecoder(ctx);
}
}
return MCDec_InitDecoder(ctx);
}
static GF_Err VTBDec_ProcessData(GF_MediaDecoder *ifcg,
char *inBuffer, u32 inBufferLength,
u16 ES_ID, u32 *CTS,
char *outBuffer, u32 *outBufferLength,
u8 PaddingBits, u32 mmlevel)
{
OSStatus status;
CMSampleBufferRef sample = NULL;
CMBlockBufferRef block_buffer = NULL;
OSType type;
char *in_data;
u32 in_data_size;
GF_Err e;
VTBDec *ctx = (VTBDec *)ifcg->privateStack;
if (ctx->skip_mpeg4_vosh) {
GF_M4VDecSpecInfo dsi;
dsi.width = dsi.height = 0;
e = gf_m4v_get_config(inBuffer, inBufferLength, &dsi);
//found a vosh - remove it from payload, init decoder if needed
if ((e==GF_OK) && dsi.width && dsi.height) {
if (!ctx->vtb_session) {
ctx->vosh = inBuffer;
ctx->vosh_size = dsi.next_object_start;
e = VTBDec_InitDecoder(ctx, GF_FALSE);
if (e) return e;
//enfoce removal for all frames
ctx->skip_mpeg4_vosh = GF_TRUE;
if (ctx->out_size != *outBufferLength) {
*outBufferLength = ctx->out_size;
return GF_BUFFER_TOO_SMALL;
}
}
ctx->vosh_size = dsi.next_object_start;
} else if (!ctx->vtb_session) {
*outBufferLength=0;
return GF_OK;
}
}
if (ctx->init_mpeg12) {
GF_M4VDecSpecInfo dsi;
dsi.width = dsi.height = 0;
e = gf_mpegv12_get_config(inBuffer, inBufferLength, &dsi);
if ((e==GF_OK) && dsi.width && dsi.height) {
ctx->width = dsi.width;
ctx->height = dsi.height;
ctx->pixel_ar = dsi.par_num;
ctx->pixel_ar <<= 16;
ctx->pixel_ar |= dsi.par_den;
e = VTBDec_InitDecoder(ctx, GF_FALSE);
if (e) return e;
if (ctx->out_size != *outBufferLength) {
*outBufferLength = ctx->out_size;
return GF_BUFFER_TOO_SMALL;
}
}
if (!ctx->vtb_session) {
*outBufferLength=0;
return GF_OK;
}
}
if (ctx->is_annex_b || (!ctx->vtb_session && ctx->nalu_size_length) ) {
if (ctx->cached_annex_b) {
in_data = ctx->cached_annex_b;
in_data_size = ctx->cached_annex_b_size;
ctx->cached_annex_b = NULL;
} else {
e = VTB_RewriteNALs(ctx, inBuffer, inBufferLength, &in_data, &in_data_size);
if (e) return e;
}
if (ctx->out_size != *outBufferLength) {
*outBufferLength = ctx->out_size;
ctx->cached_annex_b = in_data;
ctx->cached_annex_b_size = in_data_size;
return GF_BUFFER_TOO_SMALL;
}
} else if (ctx->vosh_size) {
in_data = inBuffer + ctx->vosh_size;
in_data_size = inBufferLength - ctx->vosh_size;
ctx->vosh_size = 0;
} else {
in_data = inBuffer;
in_data_size = inBufferLength;
}
if (!ctx->vtb_session) {
*outBufferLength=0;
return GF_OK;
}
status = CMBlockBufferCreateWithMemoryBlock(kCFAllocatorDefault, in_data, in_data_size, kCFAllocatorNull, NULL, 0, in_data_size, 0, &block_buffer);
if (status) {
return GF_IO_ERR;
}
*outBufferLength=0;
if (block_buffer == NULL)
return GF_OK;
status = CMSampleBufferCreate(kCFAllocatorDefault, block_buffer, TRUE, NULL, NULL, ctx->fmt_desc, 1, 0, NULL, 0, NULL, &sample);
if (status || (sample==NULL)) {
if (block_buffer)
CFRelease(block_buffer);
return GF_IO_ERR;
}
ctx->last_error = GF_OK;
status = VTDecompressionSessionDecodeFrame(ctx->vtb_session, sample, 0, NULL, 0);
if (!status)
status = VTDecompressionSessionWaitForAsynchronousFrames(ctx->vtb_session);
CFRelease(block_buffer);
CFRelease(sample);
if (ctx->cached_annex_b)
gf_free(in_data);
if (ctx->last_error) return ctx->last_error;
if (status) return GF_NON_COMPLIANT_BITSTREAM;
if (!ctx->frame) {
*outBufferLength=0;
return ctx->last_error;
}
*outBufferLength = ctx->out_size;
status = CVPixelBufferLockBaseAddress(ctx->frame, kCVPixelBufferLock_ReadOnly);
if (status != kCVReturnSuccess) {
GF_LOG(GF_LOG_ERROR, GF_LOG_CODEC, ("[VTB] Error locking frame data\n"));
return GF_IO_ERR;
}
type = CVPixelBufferGetPixelFormatType(ctx->frame);
if (CVPixelBufferIsPlanar(ctx->frame)) {
u32 i, j, nb_planes = (u32) CVPixelBufferGetPlaneCount(ctx->frame);
char *dst = outBuffer;
Bool needs_stride=GF_FALSE;
if ((type==kCVPixelFormatType_420YpCbCr8Planar)
|| (type==kCVPixelFormatType_420YpCbCr8PlanarFullRange)
|| (type==kCVPixelFormatType_422YpCbCr8_yuvs)
|| (type==kCVPixelFormatType_444YpCbCr8)
|| (type=='444v')
) {
u32 stride = (u32) CVPixelBufferGetBytesPerRowOfPlane(ctx->frame, 0);
//TOCHECK - for now the 3 planes are consecutive in VideoToolbox
if (stride==ctx->width) {
char *data = CVPixelBufferGetBaseAddressOfPlane(ctx->frame, 0);
memcpy(dst, data, sizeof(char)*ctx->out_size);
} else {
for (i=0; i<nb_planes; i++) {
char *data = CVPixelBufferGetBaseAddressOfPlane(ctx->frame, i);
u32 stride = (u32) CVPixelBufferGetBytesPerRowOfPlane(ctx->frame, i);
u32 w, h = (u32) CVPixelBufferGetHeightOfPlane(ctx->frame, i);
w = ctx->width;
if (i) {
switch (ctx->pix_fmt) {
case GF_PIXEL_YUV444:
break;
case GF_PIXEL_YUV422:
case GF_PIXEL_YV12:
w /= 2;
break;
}
}
if (stride != w) {
needs_stride=GF_TRUE;
for (j=0; j<h; j++) {
memcpy(dst, data, sizeof(char)*w);
dst += w;
data += stride;
}
} else {
memcpy(dst, data, sizeof(char)*h*stride);
dst += sizeof(char)*h*stride;
}
}
}
} else if ((type==kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange) || (type==kCVPixelFormatType_420YpCbCr8BiPlanarFullRange)) {
char *dst_v;
char *data = CVPixelBufferGetBaseAddressOfPlane(ctx->frame, 0);
u32 stride = (u32) CVPixelBufferGetBytesPerRowOfPlane(ctx->frame, 0);
u32 i, h = (u32) CVPixelBufferGetHeightOfPlane(ctx->frame, 0);
if (stride==ctx->width) {
memcpy(dst, data, sizeof(char)*h*stride);
dst += sizeof(char)*h*stride;
} else {
for (i=0; i<h; i++) {
memcpy(dst, data, sizeof(char)*ctx->width);
dst += ctx->width;
data += stride;
}
needs_stride=GF_TRUE;
}
data = CVPixelBufferGetBaseAddressOfPlane(ctx->frame, 1);
stride = (u32) CVPixelBufferGetBytesPerRowOfPlane(ctx->frame, 1);
h = (u32) CVPixelBufferGetHeightOfPlane(ctx->frame, 1);
dst_v = dst+sizeof(char) * h*stride/2;
for (i=0; i<ctx->width * h / 2; i++) {
*dst = data[0];
*dst_v = data[1];
data += 2;
dst_v++;
dst++;
if (!(i%ctx->width)) data += (stride - ctx->width);
}
}
}
CVPixelBufferUnlockBaseAddress(ctx->frame, kCVPixelBufferLock_ReadOnly);
return GF_OK;
}
static GF_Err VTBDec_InitDecoder(VTBDec *ctx, Bool force_dsi_rewrite)
{
CFMutableDictionaryRef dec_dsi, dec_type;
CFMutableDictionaryRef dsi;
VTDecompressionOutputCallbackRecord cbacks;
CFDictionaryRef buffer_attribs;
OSStatus status;
OSType kColorSpace;
CFDataRef data = NULL;
char *dsi_data=NULL;
u32 dsi_data_size=0;
dec_dsi = CFDictionaryCreateMutable(kCFAllocatorDefault, 1, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks);
kColorSpace = kCVPixelFormatType_420YpCbCr8Planar;
ctx->pix_fmt = GF_PIXEL_YV12;
switch (ctx->esd->decoderConfig->objectTypeIndication) {
case GPAC_OTI_VIDEO_AVC :
if (ctx->sps && ctx->pps) {
AVCState avc;
s32 idx;
memset(&avc, 0, sizeof(AVCState));
avc.sps_active_idx = -1;
idx = gf_media_avc_read_sps(ctx->sps, ctx->sps_size, &avc, 0, NULL);
ctx->vtb_type = kCMVideoCodecType_H264;
assert(ctx->sps);
ctx->width = avc.sps[idx].width;
ctx->height = avc.sps[idx].height;
if (avc.sps[idx].vui.par_num && avc.sps[idx].vui.par_den) {
ctx->pixel_ar = avc.sps[idx].vui.par_num;
ctx->pixel_ar <<= 16;
ctx->pixel_ar |= avc.sps[idx].vui.par_den;
}
ctx->chroma_format = avc.sps[idx].chroma_format;
ctx->luma_bit_depth = 8 + avc.sps[idx].luma_bit_depth_m8;
ctx->chroma_bit_depth = 8 + avc.sps[idx].chroma_bit_depth_m8;
switch (ctx->chroma_format) {
case 2:
//422 decoding doesn't seem supported ...
if (ctx->luma_bit_depth>8) {
kColorSpace = kCVPixelFormatType_422YpCbCr10;
ctx->pix_fmt = GF_PIXEL_YUV422_10;
} else {
kColorSpace = kCVPixelFormatType_422YpCbCr8;
ctx->pix_fmt = GF_PIXEL_YUV422;
}
break;
case 3:
if (ctx->luma_bit_depth>8) {
kColorSpace = kCVPixelFormatType_444YpCbCr10;
ctx->pix_fmt = GF_PIXEL_YUV444_10;
} else {
kColorSpace = kCVPixelFormatType_444YpCbCr8;
ctx->pix_fmt = GF_PIXEL_YUV444;
}
break;
default:
if (ctx->luma_bit_depth>8) {
kColorSpace = kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
ctx->pix_fmt = GF_PIXEL_YV12_10;
}
break;
}
if (!ctx->esd->decoderConfig->decoderSpecificInfo || force_dsi_rewrite || !ctx->esd->decoderConfig->decoderSpecificInfo->data) {
GF_AVCConfigSlot *slc_s, *slc_p;
GF_AVCConfig *cfg = gf_odf_avc_cfg_new();
cfg->configurationVersion = 1;
cfg->profile_compatibility = avc.sps[idx].prof_compat;
cfg->AVCProfileIndication = avc.sps[idx].profile_idc;
cfg->AVCLevelIndication = avc.sps[idx].level_idc;
cfg->chroma_format = avc.sps[idx].chroma_format;
cfg->luma_bit_depth = 8 + avc.sps[idx].luma_bit_depth_m8;
cfg->chroma_bit_depth = 8 + avc.sps[idx].chroma_bit_depth_m8;
cfg->nal_unit_size = 4;
GF_SAFEALLOC(slc_s, GF_AVCConfigSlot);
slc_s->data = ctx->sps;
slc_s->size = ctx->sps_size;
gf_list_add(cfg->sequenceParameterSets, slc_s);
GF_SAFEALLOC(slc_p, GF_AVCConfigSlot);
slc_p->data = ctx->pps;
slc_p->size = ctx->pps_size;
gf_list_add(cfg->pictureParameterSets , slc_p);
gf_odf_avc_cfg_write(cfg, &dsi_data, &dsi_data_size);
slc_s->data = slc_p->data = NULL;
gf_odf_avc_cfg_del((cfg));
} else {
dsi_data = ctx->esd->decoderConfig->decoderSpecificInfo->data;
dsi_data_size = ctx->esd->decoderConfig->decoderSpecificInfo->dataLength;
}
dsi = CFDictionaryCreateMutable(kCFAllocatorDefault, 1, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks);
data = CFDataCreate(kCFAllocatorDefault, dsi_data, dsi_data_size);
if (data) {
CFDictionarySetValue(dsi, CFSTR("avcC"), data);
CFDictionarySetValue(dec_dsi, kCMFormatDescriptionExtension_SampleDescriptionExtensionAtoms, dsi);
CFRelease(data);
}
CFRelease(dsi);
if (!ctx->esd->decoderConfig->decoderSpecificInfo || !ctx->esd->decoderConfig->decoderSpecificInfo->data) {
gf_free(ctx->sps);
ctx->sps = NULL;
gf_free(ctx->pps);
ctx->pps = NULL;
gf_free(dsi_data);
}
}
break;
case GPAC_OTI_VIDEO_MPEG2_SIMPLE:
case GPAC_OTI_VIDEO_MPEG2_MAIN:
case GPAC_OTI_VIDEO_MPEG2_SNR:
case GPAC_OTI_VIDEO_MPEG2_SPATIAL:
case GPAC_OTI_VIDEO_MPEG2_HIGH:
case GPAC_OTI_VIDEO_MPEG2_422:
ctx->vtb_type = kCMVideoCodecType_MPEG2Video;
if (!ctx->width || !ctx->height) {
ctx->init_mpeg12 = GF_TRUE;
return GF_OK;
}
ctx->init_mpeg12 = GF_FALSE;
break;
case GPAC_OTI_VIDEO_MPEG1:
ctx->vtb_type = kCMVideoCodecType_MPEG1Video;
if (!ctx->width || !ctx->height) {
ctx->init_mpeg12 = GF_TRUE;
return GF_OK;
}
ctx->init_mpeg12 = GF_FALSE;
break;
case GPAC_OTI_VIDEO_MPEG4_PART2 :
{
Bool reset_dsi = GF_FALSE;
ctx->vtb_type = kCMVideoCodecType_MPEG4Video;
if (!ctx->esd->decoderConfig->decoderSpecificInfo) {
ctx->esd->decoderConfig->decoderSpecificInfo = (GF_DefaultDescriptor *) gf_odf_desc_new(GF_ODF_DSI_TAG);
}
if (!ctx->esd->decoderConfig->decoderSpecificInfo->data) {
reset_dsi = GF_TRUE;
ctx->esd->decoderConfig->decoderSpecificInfo->data = ctx->vosh;
ctx->esd->decoderConfig->decoderSpecificInfo->dataLength = ctx->vosh_size;
}
if (ctx->esd->decoderConfig->decoderSpecificInfo->data) {
GF_M4VDecSpecInfo vcfg;
GF_BitStream *bs;
gf_m4v_get_config(ctx->esd->decoderConfig->decoderSpecificInfo->data, ctx->esd->decoderConfig->decoderSpecificInfo->dataLength, &vcfg);
ctx->width = vcfg.width;
ctx->height = vcfg.height;
if (ctx->esd->slConfig) {
ctx->esd->slConfig->predefined = 2;
}
bs = gf_bs_new(NULL, 0, GF_BITSTREAM_WRITE);
gf_bs_write_u32(bs, 0);
gf_odf_desc_write_bs((GF_Descriptor *) ctx->esd, bs);
gf_bs_get_content(bs, &dsi_data, &dsi_data_size);
gf_bs_del(bs);
dsi = CFDictionaryCreateMutable(kCFAllocatorDefault, 1, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks);
data = CFDataCreate(kCFAllocatorDefault, dsi_data, dsi_data_size);
gf_free(dsi_data);
if (data) {
CFDictionarySetValue(dsi, CFSTR("esds"), data);
CFDictionarySetValue(dec_dsi, kCMFormatDescriptionExtension_SampleDescriptionExtensionAtoms, dsi);
CFRelease(data);
}
CFRelease(dsi);
if (reset_dsi) {
ctx->esd->decoderConfig->decoderSpecificInfo->data = NULL;
ctx->esd->decoderConfig->decoderSpecificInfo->dataLength = 0;
}
ctx->skip_mpeg4_vosh = GF_FALSE;
} else {
ctx->skip_mpeg4_vosh = GF_TRUE;
return GF_OK;
}
break;
}
case GPAC_OTI_MEDIA_GENERIC:
if (ctx->esd->decoderConfig->decoderSpecificInfo && ctx->esd->decoderConfig->decoderSpecificInfo->dataLength) {
char *dsi = ctx->esd->decoderConfig->decoderSpecificInfo->data;
if (ctx->esd->decoderConfig->decoderSpecificInfo->dataLength<8) return GF_NON_COMPLIANT_BITSTREAM;
if (strnicmp(dsi, "s263", 4)) return GF_NOT_SUPPORTED;
ctx->width = ((u8) dsi[4]); ctx->width<<=8; ctx->width |= ((u8) dsi[5]);
ctx->height = ((u8) dsi[6]); ctx->height<<=8; ctx->height |= ((u8) dsi[7]);
ctx->vtb_type = kCMVideoCodecType_H263;
}
break;
default :
return GF_NOT_SUPPORTED;
}
if (! ctx->width || !ctx->height) return GF_NOT_SUPPORTED;
status = CMVideoFormatDescriptionCreate(kCFAllocatorDefault, ctx->vtb_type, ctx->width, ctx->height, dec_dsi, &ctx->fmt_desc);
if (!ctx->fmt_desc) {
if (dec_dsi) CFRelease(dec_dsi);
return GF_NON_COMPLIANT_BITSTREAM;
}
buffer_attribs = VTBDec_CreateBufferAttributes(ctx->width, ctx->height, kColorSpace);
cbacks.decompressionOutputCallback = VTBDec_on_frame;
cbacks.decompressionOutputRefCon = ctx;
dec_type = CFDictionaryCreateMutable(kCFAllocatorDefault, 1, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks);
CFDictionarySetValue(dec_type, kVTVideoDecoderSpecification_RequireHardwareAcceleratedVideoDecoder, kCFBooleanTrue);
ctx->is_hardware = GF_TRUE;
status = VTDecompressionSessionCreate(NULL, ctx->fmt_desc, dec_type, NULL, &cbacks, &ctx->vtb_session);
//if HW decoder not available, try soft one
if (status) {
status = VTDecompressionSessionCreate(NULL, ctx->fmt_desc, NULL, buffer_attribs, &cbacks, &ctx->vtb_session);
ctx->is_hardware = GF_FALSE;
}
if (dec_dsi)
CFRelease(dec_dsi);
if (dec_type)
CFRelease(dec_type);
if (buffer_attribs)
CFRelease(buffer_attribs);
switch (status) {
case kVTVideoDecoderNotAvailableNowErr:
case kVTVideoDecoderUnsupportedDataFormatErr:
return GF_NOT_SUPPORTED;
case kVTVideoDecoderMalfunctionErr:
return GF_IO_ERR;
case kVTVideoDecoderBadDataErr :
return GF_BAD_PARAM;
case kVTPixelTransferNotSupportedErr:
case kVTCouldNotFindVideoDecoderErr:
return GF_NOT_SUPPORTED;
case 0:
break;
default:
return GF_SERVICE_ERROR;
}
//good to go !
if (ctx->pix_fmt == GF_PIXEL_YUV422) {
ctx->out_size = ctx->width*ctx->height*2;
} else if (ctx->pix_fmt == GF_PIXEL_YUV444) {
ctx->out_size = ctx->width*ctx->height*3;
} else {
// (ctx->pix_fmt == GF_PIXEL_YV12)
ctx->out_size = ctx->width*ctx->height*3/2;
}
if (ctx->luma_bit_depth>8) {
ctx->out_size *= 2;
}
return GF_OK;
}
