[FFmpeg-devel] [PATCH 01/11] avformat/mov: add support for lhvC box parsing

[James Almer]

On 7/7/2024 12:43 PM, Andreas Rheinhardt wrote:
> James Almer:
>> Signed-off-by: James Almer <jamrial at gmail.com>
>> ---
>>   libavformat/mov.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 48 insertions(+)
>>
>> diff --git a/libavformat/mov.c b/libavformat/mov.c
>> index a3951a6942..30e8086855 100644
>> --- a/libavformat/mov.c
>> +++ b/libavformat/mov.c
>> @@ -8157,6 +8157,53 @@ static int mov_read_dvcc_dvvc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
>>       return ff_isom_parse_dvcc_dvvc(c->fc, st, buf, read_size);
>>   }
>>   
>> +static int mov_read_lhvc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
>> +{
>> +    AVStream *st;
>> +    uint8_t *buf;
>> +    int ret, old_size, num_arrays;
>> +
>> +    if (c->fc->nb_streams < 1)
>> +        return 0;
>> +    st = c->fc->streams[c->fc->nb_streams-1];
>> +
>> +    if (!st->codecpar->extradata_size)
>> +        // TODO: handle lhvC when present before hvcC
>> +        return 0;
>> +
>> +    if (atom.size < 6 || st->codecpar->extradata_size < 23)
>> +        return AVERROR_INVALIDDATA;
>> +
>> +    buf = av_malloc(atom.size + AV_INPUT_BUFFER_PADDING_SIZE);
>> +    if (!buf)
>> +        return AVERROR(ENOMEM);
>> +    memset(buf + atom.size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
>> +
>> +    ret = ffio_read_size(pb, buf, atom.size);
>> +    if (ret < 0) {
>> +        av_free(buf);
>> +        av_log(c->fc, AV_LOG_WARNING, "lhvC atom truncated\n");
>> +        return 0;
>> +    }
>> +
>> +    num_arrays = buf[5];
>> +    old_size = st->codecpar->extradata_size;
>> +    atom.size -= 8 /* account for mov_realloc_extradata offseting */
>> +               + 6 /* lhvC bytes before the arrays*/;
>> +
>> +    ret = mov_realloc_extradata(st->codecpar, atom);
>> +    if (ret < 0) {
>> +        av_free(buf);
>> +        return ret;
>> +    }
>> +
>> +    st->codecpar->extradata[22] += num_arrays;
>> +    memcpy(st->codecpar->extradata + old_size, buf + 6, atom.size + 8);
>> +
>> +    av_free(buf);
>> +    return 0;
>> +}
>> +
>>   static int mov_read_kind(MOVContext *c, AVIOContext *pb, MOVAtom atom)
>>   {
>>       AVFormatContext *ctx = c->fc;
>> @@ -8943,6 +8990,7 @@ static const MOVParseTableEntry mov_default_parse_table[] = {
>>   { MKTAG('i','p','r','p'), mov_read_iprp },
>>   { MKTAG('i','i','n','f'), mov_read_iinf },
>>   { MKTAG('a','m','v','e'), mov_read_amve }, /* ambient viewing environment box */
>> +{ MKTAG('l','h','v','C'), mov_read_lhvc },
>>   #if CONFIG_IAMFDEC
>>   { MKTAG('i','a','c','b'), mov_read_iacb },
>>   #endif
> 
> Could this also explain what this box is supposed to be good for?

It contains the global nuh_layer_id > 0 PS NALUs to handle L-HEVC 
(layered) streams. It's described in ISO 14496-15 section 9.4.3.