static int dts_importer_probe( importer_t *importer )
{
dts_importer_t *dts_imp = create_dts_importer( importer );
if( !dts_imp )
return LSMASH_ERR_MEMORY_ALLOC;
lsmash_bits_t *bits = dts_imp->info.bits;
lsmash_bs_t *bs = bits->bs;
bs->buffer.max_size = DTS_MAX_EXSS_SIZE;
importer->info = dts_imp;
int err = dts_importer_get_next_accessunit_internal( importer );
if( err < 0 )
goto fail;
lsmash_audio_summary_t *summary = dts_create_summary( &dts_imp->info );
if( !summary )
{
err = LSMASH_ERR_NAMELESS;
goto fail;
}
if( importer->status != IMPORTER_EOF )
importer->status = IMPORTER_OK;
dts_imp->au_number = 0;
if( lsmash_add_entry( importer->summaries, summary ) < 0 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
err = LSMASH_ERR_MEMORY_ALLOC;
goto fail;
}
return 0;
fail:
remove_dts_importer( dts_imp );
importer->info = NULL;
return err;
}
static lsmash_video_summary_t *vc1_create_summary( vc1_info_t *info, vc1_sequence_header_t *sequence, uint32_t max_au_length )
{
if( !info->sequence.present || !info->entry_point.present )
return NULL;
lsmash_video_summary_t *summary = (lsmash_video_summary_t *)lsmash_create_summary( LSMASH_SUMMARY_TYPE_VIDEO );
if( !summary )
return NULL;
lsmash_codec_specific_t *specific = lsmash_create_codec_specific_data( LSMASH_CODEC_SPECIFIC_DATA_TYPE_ISOM_VIDEO_VC_1,
LSMASH_CODEC_SPECIFIC_FORMAT_UNSTRUCTURED );
specific->data.unstructured = lsmash_create_vc1_specific_info( &info->dvc1_param, &specific->size );
if( !specific->data.unstructured
|| lsmash_add_entry( &summary->opaque->list, specific ) < 0 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
lsmash_destroy_codec_specific_data( specific );
return NULL;
}
summary->sample_type = ISOM_CODEC_TYPE_VC_1_VIDEO;
summary->max_au_length = max_au_length;
summary->timescale = sequence->framerate_numerator;
summary->timebase = sequence->framerate_denominator;
summary->vfr = !sequence->framerate_flag;
summary->sample_per_field = 0;
summary->width = sequence->disp_horiz_size;
summary->height = sequence->disp_vert_size;
summary->par_h = sequence->aspect_width;
summary->par_v = sequence->aspect_height;
summary->color.primaries_index = sequence->color_prim;
summary->color.transfer_index = sequence->transfer_char;
summary->color.matrix_index = sequence->matrix_coef;
return summary;
}
static int vc1_importer_probe( importer_t *importer )
{
/* Find the first start code. */
vc1_importer_t *vc1_imp = create_vc1_importer( importer );
if( !vc1_imp )
return LSMASH_ERR_MEMORY_ALLOC;
lsmash_bs_t *bs = importer->bs;
uint64_t first_ebdu_head_pos = 0;
int err;
while( 1 )
{
/* The first EBDU in decoding order of the stream shall have start code (0x000001). */
if( 0x000001 == lsmash_bs_show_be24( bs, first_ebdu_head_pos ) )
break;
/* Invalid if encountered any value of non-zero before the first start code. */
if( lsmash_bs_show_byte( bs, first_ebdu_head_pos ) )
{
err = LSMASH_ERR_INVALID_DATA;
goto fail;
}
++first_ebdu_head_pos;
}
/* OK. It seems the stream has a sequence header of VC-1. */
importer->info = vc1_imp;
vc1_info_t *info = &vc1_imp->info;
lsmash_bs_read_seek( bs, first_ebdu_head_pos, SEEK_SET );
info->ebdu_head_pos = first_ebdu_head_pos;
if( (err = vc1_analyze_whole_stream( importer )) < 0 )
goto fail;
lsmash_video_summary_t *summary = vc1_create_summary( info, &vc1_imp->first_sequence, vc1_imp->max_au_length );
if( !summary )
{
err = LSMASH_ERR_NAMELESS;
goto fail;
}
if( lsmash_add_entry( importer->summaries, summary ) < 0 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
err = LSMASH_ERR_MEMORY_ALLOC;
goto fail;
}
/* Go back to layer of the first EBDU. */
importer->status = IMPORTER_OK;
lsmash_bs_read_seek( bs, first_ebdu_head_pos, SEEK_SET );
info->prev_bdu_type = 0xFF; /* 0xFF is a forbidden value. */
info->ebdu_head_pos = first_ebdu_head_pos;
uint8_t *temp_access_unit = info->access_unit.data;
uint8_t *temp_incomplete_access_unit = info->access_unit.incomplete_data;
memset( &info->access_unit, 0, sizeof(vc1_access_unit_t) );
info->access_unit.data = temp_access_unit;
info->access_unit.incomplete_data = temp_incomplete_access_unit;
memset( &info->picture, 0, sizeof(vc1_picture_info_t) );
return 0;
fail:
remove_vc1_importer( vc1_imp );
importer->info = NULL;
lsmash_remove_entries( importer->summaries, lsmash_cleanup_summary );
return err;
}
static void remove_mp4_hnd( hnd_t handle )
{
mp4_hnd_t *p_mp4 = handle;
if( !p_mp4 )
return;
lsmash_cleanup_summary( (lsmash_summary_t *)p_mp4->summary );
lsmash_close_file( &p_mp4->file_param );
lsmash_destroy_root( p_mp4->p_root );
free( p_mp4->p_sei_buffer );
free( p_mp4 );
}
static int wave_importer_probe( importer_t *importer )
{
wave_importer_t *wave_imp = create_wave_importer( importer );
if( !wave_imp )
return LSMASH_ERR_MEMORY_ALLOC;
int err = 0;
uint32_t filesize;
lsmash_bs_t *bs = importer->bs;
if( lsmash_bs_get_be32( bs ) != LSMASH_4CC( 'R', 'I', 'F', 'F' )
|| ((filesize = lsmash_bs_get_le32( bs ) + 8) < WAVE_MIN_FILESIZE && filesize > 8)
|| lsmash_bs_get_be32( bs ) != LSMASH_4CC( 'W', 'A', 'V', 'E' ) )
{
err = LSMASH_ERR_INVALID_DATA;
goto fail;
}
int fmt_chunk_present = 0;
int data_chunk_present = 0;
while( !bs->eob && !(fmt_chunk_present && data_chunk_present) )
{
uint32_t ckID = lsmash_bs_get_be32( bs );
uint32_t ckSize = lsmash_bs_get_le32( bs );
lsmash_bs_reset_counter( bs );
switch( ckID )
{
case LSMASH_4CC( 'f', 'm', 't', ' ' ) :
if( ckSize < 16 )
{
err = LSMASH_ERR_INVALID_DATA;
goto fail;
}
if( (err = wave_parse_fmt_chunk( wave_imp, bs )) < 0 )
goto fail;
fmt_chunk_present = 1;
break;
case LSMASH_4CC( 'd', 'a', 't', 'a' ) :
if( !fmt_chunk_present )
{
/* The 'fmt ' chunk must be present before the 'data' chunk. */
err = LSMASH_ERR_INVALID_DATA;
goto fail;
}
wave_imp->chunk.data_offset = lsmash_bs_get_stream_pos( bs );
wave_imp->chunk.length = ckSize;
wave_imp->chunk.number = 1;
wave_imp->chunk.file = importer->file;
wave_imp->number_of_samples = ckSize / wave_imp->fmt.wfx.nBlockAlign;
data_chunk_present = 1;
break;
default :
break;
}
if( !data_chunk_present )
{
/* Skip the rest of this chunk.
* Note that ckData is word-aligned even if ckSize is an odd number. */
uint32_t skip_size = ckSize;
if( skip_size & 1 )
skip_size++;
if( skip_size > lsmash_bs_count( bs ) )
{
skip_size -= lsmash_bs_count( bs );
lsmash_bs_read_seek( bs, skip_size, SEEK_CUR );
}
}
}
if( !(fmt_chunk_present && data_chunk_present) )
{
err = LSMASH_ERR_INVALID_DATA;
goto fail;
}
/* Make fake movie.
* Treat WAVE file format as if it's QuickTime file format. */
uint32_t track_ID;
lsmash_movie_parameters_t movie_param = { 0 };
lsmash_track_parameters_t track_param = { 0 };
lsmash_media_parameters_t media_param = { 0 };
importer->file->qt_compatible = 1;
if( (err = lsmash_importer_make_fake_movie( importer )) < 0
|| (err = lsmash_importer_make_fake_track( importer, ISOM_MEDIA_HANDLER_TYPE_AUDIO_TRACK, &track_ID )) < 0
|| (err = lsmash_get_movie_parameters( importer->root, &movie_param )) < 0
|| (err = lsmash_get_track_parameters( importer->root, track_ID, &track_param )) < 0
|| (err = lsmash_get_media_parameters( importer->root, track_ID, &media_param )) < 0 )
goto fail;
movie_param.timescale = wave_imp->fmt.wfx.nSamplesPerSec;
media_param.timescale = wave_imp->fmt.wfx.nSamplesPerSec;
if( (err = lsmash_set_movie_parameters( importer->root, &movie_param )) < 0
|| (err = lsmash_set_track_parameters( importer->root, track_ID, &track_param )) < 0
|| (err = lsmash_set_media_parameters( importer->root, track_ID, &media_param )) < 0 )
goto fail;
lsmash_audio_summary_t *summary = wave_create_summary( &wave_imp->fmt );
if( !summary || lsmash_add_sample_entry( importer->root, track_ID, summary ) != 1 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
err = LSMASH_ERR_NAMELESS;
goto fail;
}
if( (err = lsmash_add_entry( importer->summaries, summary )) < 0 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
goto fail;
}
importer->info = wave_imp;
importer->status = IMPORTER_OK;
return 0;
fail:
lsmash_importer_break_fake_movie( importer );
remove_wave_importer( wave_imp );
importer->file->qt_compatible = 0;
importer->info = NULL;
return err;
}
static lsmash_audio_summary_t *wave_create_summary( waveformat_extensible_t *fmt )
{
lsmash_audio_summary_t *summary = (lsmash_audio_summary_t *)lsmash_create_summary( LSMASH_SUMMARY_TYPE_AUDIO );
if( !summary )
return NULL;
waveformat_extended_t *wfx = &fmt->wfx;
summary->sample_type = QT_CODEC_TYPE_LPCM_AUDIO;
summary->aot = MP4A_AUDIO_OBJECT_TYPE_NULL;
summary->frequency = wfx->nSamplesPerSec;
summary->channels = wfx->nChannels;
summary->sample_size = wfx->wFormatTag == WAVE_FORMAT_TYPE_ID_EXTENSIBLE
? fmt->Samples.wValidBitsPerSample
: wfx->wBitsPerSample;
summary->samples_in_frame = 1000; /* arbitrary */
summary->sbr_mode = MP4A_AAC_SBR_NOT_SPECIFIED;
summary->bytes_per_frame = wfx->nBlockAlign * summary->samples_in_frame;
summary->max_au_length = summary->bytes_per_frame;
lsmash_codec_specific_t *cs = lsmash_create_codec_specific_data( LSMASH_CODEC_SPECIFIC_DATA_TYPE_QT_AUDIO_FORMAT_SPECIFIC_FLAGS,
LSMASH_CODEC_SPECIFIC_FORMAT_STRUCTURED );
if( !cs )
goto fail;
lsmash_qt_audio_format_specific_flags_t *lpcm = (lsmash_qt_audio_format_specific_flags_t *)cs->data.structured;
if( (summary->sample_size & 7) == 0 )
lpcm->format_flags |= QT_AUDIO_FORMAT_FLAG_PACKED;
else
lpcm->format_flags |= QT_AUDIO_FORMAT_FLAG_ALIGNED_HIGH;
if( summary->sample_size > 8 )
lpcm->format_flags |= QT_AUDIO_FORMAT_FLAG_SIGNED_INTEGER;
if( lsmash_add_entry( &summary->opaque->list, cs ) < 0 )
{
lsmash_destroy_codec_specific_data( cs );
goto fail;
}
if( wfx->wFormatTag == WAVE_FORMAT_TYPE_ID_EXTENSIBLE || wfx->nChannels > 2 )
{
cs = lsmash_create_codec_specific_data( LSMASH_CODEC_SPECIFIC_DATA_TYPE_QT_AUDIO_CHANNEL_LAYOUT,
LSMASH_CODEC_SPECIFIC_FORMAT_STRUCTURED );
if( !cs )
goto fail;
lsmash_qt_audio_channel_layout_t *layout = (lsmash_qt_audio_channel_layout_t *)cs->data.structured;
if( wfx->wFormatTag == WAVE_FORMAT_TYPE_ID_EXTENSIBLE )
{
layout->channelLayoutTag = QT_CHANNEL_LAYOUT_USE_CHANNEL_BITMAP;
layout->channelBitmap = fmt->dwChannelMask;
}
else
{
layout->channelLayoutTag = QT_CHANNEL_LAYOUT_UNKNOWN | wfx->nChannels;
layout->channelBitmap = 0;
}
if( lsmash_add_entry( &summary->opaque->list, cs ) < 0 )
{
lsmash_destroy_codec_specific_data( cs );
goto fail;
}
}
return summary;
fail:
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
return NULL;
}
bool fcMP4Muxer::mux(const Params ¶ms)
{
lsmash_root_t *root;
lsmash_file_parameters_t mp4_stream;
lsmash_file_parameters_t h264_stream;
lsmash_brand_type major_brand = ISOM_BRAND_TYPE_MP42;
lsmash_brand_type compatible_brands[2] = { ISOM_BRAND_TYPE_MP42, ISOM_BRAND_TYPE_ISOM };
uint32_t fps_num = params.frame_rate;
uint32_t fps_den = 1;
// 出力 mp4
root = lsmash_create_root();
if (lsmash_open_file(params.out_mp4_path, 0, &mp4_stream) != 0) {
return false;
}
mp4_stream.major_brand = major_brand;
mp4_stream.brands = compatible_brands;
mp4_stream.brand_count = sizeof(compatible_brands) / sizeof(compatible_brands[0]);
mp4_stream.minor_version = 0;
lsmash_set_file(root, &mp4_stream);
lsmash_movie_parameters_t movie_param;
lsmash_initialize_movie_parameters(&movie_param);
lsmash_set_movie_parameters(root, &movie_param);
int track_number = 1;
MP4TrackData track_data[2];
int num_track_data = 0;
if (params.in_h264_path) {
importer_t *h264_importer = lsmash_importer_open(params.in_h264_path, "H.264");
if (h264_importer == nullptr) { return false; }
int h264_track_id = lsmash_create_track(root, ISOM_MEDIA_HANDLER_TYPE_VIDEO_TRACK);
lsmash_track_parameters_t track_param;
lsmash_initialize_track_parameters(&track_param);
(int&)track_param.mode = ISOM_TRACK_IN_MOVIE | ISOM_TRACK_IN_PREVIEW;
int sample_entry = 0;
lsmash_summary_t *summary = lsmash_duplicate_summary(h264_importer, track_number);
auto video_summary = (lsmash_video_summary_t*)summary;
track_param.display_width = video_summary->width << 16;
track_param.display_height = video_summary->height << 16;
lsmash_set_track_parameters(root, h264_track_id, &track_param);
sample_entry = lsmash_add_sample_entry(root, h264_track_id, summary);
auto& td = track_data[num_track_data++];
td.importer = h264_importer;
td.summary = summary;
td.sample_entry = sample_entry;
td.track_number = track_number;
td.track_id = h264_track_id;
td.timescale = video_summary->timescale;
td.timebase = video_summary->timebase;
lsmash_media_parameters_t media_param;
lsmash_initialize_media_parameters(&media_param);
media_param.timescale = td.timescale;
lsmash_set_media_parameters(root, h264_track_id, &media_param);
}
if (params.in_aac_path) {
importer_t *aac_importer = lsmash_importer_open(params.in_aac_path, "adts");
if (aac_importer == nullptr) { return false; }
int aac_track_id = lsmash_create_track(root, ISOM_MEDIA_HANDLER_TYPE_AUDIO_TRACK);
lsmash_track_parameters_t track_param;
lsmash_initialize_track_parameters(&track_param);
(int&)track_param.mode = ISOM_TRACK_IN_MOVIE | ISOM_TRACK_IN_PREVIEW;
int sample_entry = 0;
lsmash_summary_t *summary = lsmash_duplicate_summary(aac_importer, track_number);
auto audio_summary = (lsmash_audio_summary_t*)summary;
lsmash_set_track_parameters(root, aac_track_id, &track_param);
sample_entry = lsmash_add_sample_entry(root, aac_track_id, summary);
auto& td = track_data[num_track_data++];
td.importer = aac_importer;
td.summary = summary;
td.sample_entry = sample_entry;
td.track_number = track_number;
td.track_id = aac_track_id;
td.timescale = audio_summary->frequency;
td.timebase = 1;
lsmash_media_parameters_t media_param;
lsmash_initialize_media_parameters(&media_param);
media_param.timescale = td.timescale;
lsmash_set_media_parameters(root, aac_track_id, &media_param);
}
double largest_dts = 0.0;
uint32_t num_consecutive_sample_skip = 0;
for (int ti = 0;;) {
auto &td = track_data[ti];
if (!td.sample) {
int ret = lsmash_importer_get_access_unit(td.importer, td.track_number, &td.sample);
if (ret <= -1) // error
{
lsmash_delete_sample(td.sample);
break;
}
else if (ret == 1) /* a change of stream's properties */
{
lsmash_cleanup_summary(td.summary);
td.summary = lsmash_duplicate_summary(td.importer, td.track_number);
td.sample_entry = lsmash_add_sample_entry(root, td.track_id, td.summary);
if (!td.sample_entry) { break; }
}
else if (ret == 2) /* EOF */
{
lsmash_delete_sample(td.sample);
break;
}
if (td.sample)
{
td.sample->index = td.sample_entry;
td.sample->dts *= td.timebase;
td.sample->cts *= td.timebase;
td.dts = (double)td.sample->dts / td.timescale;
}
}
if (td.sample) {
if (td.dts <= largest_dts || num_consecutive_sample_skip == num_track_data)
{
uint64_t sample_size = td.sample->length;
uint64_t sample_dts = td.sample->dts;
uint64_t sample_cts = td.sample->cts;
if (lsmash_append_sample(root, td.track_id, td.sample)) {
return false;
}
td.prev_dts = sample_dts;
td.sample = nullptr;
largest_dts = std::max<double>(largest_dts, td.dts);
num_consecutive_sample_skip = 0;
}
else {
// skip
++num_consecutive_sample_skip;
}
}
ti = (ti + 1) % num_track_data;
}
for (int i = 0; i < num_track_data; ++i) {
lsmash_flush_pooled_samples(root, track_data[i].track_id, fps_den);
lsmash_cleanup_summary(track_data[i].summary);
lsmash_importer_close(track_data[i].importer);
}
lsmash_finish_movie(root, nullptr);
lsmash_close_file(&mp4_stream);
lsmash_destroy_root(root);
return true;
}
static lsmash_audio_summary_t *dts_create_summary( dts_info_t *info )
{
lsmash_audio_summary_t *summary = (lsmash_audio_summary_t *)lsmash_create_summary( LSMASH_SUMMARY_TYPE_AUDIO );
if( !summary )
return NULL;
lsmash_dts_specific_parameters_t *param = &info->ddts_param;
lsmash_codec_specific_t *specific = lsmash_create_codec_specific_data( LSMASH_CODEC_SPECIFIC_DATA_TYPE_ISOM_AUDIO_DTS,
LSMASH_CODEC_SPECIFIC_FORMAT_UNSTRUCTURED );
specific->data.unstructured = lsmash_create_dts_specific_info( param, &specific->size );
if( !specific->data.unstructured
|| lsmash_add_entry( &summary->opaque->list, specific ) < 0 )
{
lsmash_cleanup_summary( (lsmash_summary_t *)summary );
lsmash_destroy_codec_specific_data( specific );
return NULL;
}
/* The CODEC identifiers probably should not be the combination of 'mp4a' and
* the objectTypeIndications for DTS audio since there is no public specification
* which defines the encapsulation of the stream as the MPEG-4 Audio context yet.
* In the world, there are muxers which is using such doubtful implementation.
* The objectTypeIndications are registered at MP4RA, but this does not always
* mean we can mux by using those objectTypeIndications.
* If available, there shall be the specification which defines the existence of
* DecoderSpecificInfo and its semantics, and what access unit consists of. */
summary->sample_type = lsmash_dts_get_codingname( param );
summary->aot = MP4A_AUDIO_OBJECT_TYPE_NULL; /* make no sense */
summary->sbr_mode = MP4A_AAC_SBR_NOT_SPECIFIED; /* make no sense */
switch( param->DTSSamplingFrequency )
{
case 12000 : /* Invalid? (No reference in the spec) */
case 24000 :
case 48000 :
case 96000 :
case 192000 :
case 384000 : /* Invalid? (No reference in the spec) */
summary->frequency = 48000;
break;
case 22050 :
case 44100 :
case 88200 :
case 176400 :
case 352800 : /* Invalid? (No reference in the spec) */
summary->frequency = 44100;
break;
case 8000 : /* Invalid? (No reference in the spec) */
case 16000 :
case 32000 :
case 64000 :
case 128000 :
summary->frequency = 32000;
break;
default :
summary->frequency = 0;
break;
}
summary->samples_in_frame = (summary->frequency * info->frame_duration) / param->DTSSamplingFrequency;
summary->max_au_length = DTS_MAX_CORE_SIZE + DTS_MAX_NUM_EXSS * DTS_MAX_EXSS_SIZE;
summary->sample_size = param->pcmSampleDepth;
summary->channels = dts_get_max_channel_count( info );
return summary;
}
