Vrftjkry

I have written a library that could be used with python to extract frames from a video in a buffer for some processing. I used the examples from the ffmpeg to build my library using libavcodec.
This and seems to be working fine for most of the videos. But I see on some videos where my library does not extract all the frames (way less than the FPS).

It seems like somehow the packets might be out of order and docoder is not able to handle it. I keep getting a lot of the following errors

pid=100 pes_code=0x1e0

nal_unit_type: 9, nal_ref_idc: 0

nal_unit_type: 1, nal_ref_idc: 2

deblocking_filter_idc 7 out of range

decode_slice_header error

no frame!

I use the following steps to setup the decoder.

    //open input file, allocate context

if ((ret = avformat_open_input(&format_ctx, in_filename, 0, 0)) < 0) {

    PyErr_SetString(ExtractorError, "Could not open input file!");

    goto end;

}



if ((ret = avformat_find_stream_info(format_ctx, 0)) < 0) {

    PyErr_SetString(ExtractorError, "Failed to retrieve input stream information!");

    goto end;

}



av_dump_format(format_ctx, 0, in_filename, 0);



// Get the video index from the stream

for(int i=0; i<format_ctx->nb_streams ;i++ )

{

    if( format_ctx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_VIDEO )

    {

        video_stream_index = i;

        break;

    }

}



/* if video stream not availabe */

if((video_stream_index) == -1)

{

    PyErr_SetString(ExtractorError, "video stream to retreive fps is not found!");

    return NULL;

}



long duration = format_ctx->duration + (format_ctx->duration <= INT64_MAX - 5000 ? 5000 : 0);



float duration_in_secs = (float)duration / AV_TIME_BASE;



stream_mapping_size = format_ctx->nb_streams;

stream_mapping = av_mallocz_array(stream_mapping_size, sizeof(*stream_mapping));

if (!stream_mapping) {

    ret = AVERROR(ENOMEM);

    goto end;

}



AVCodec *pCodec = NULL;

AVCodecParameters *in_codecpar = NULL;

for (i = 0; i < format_ctx->nb_streams; i++) {

    AVStream *in_stream = format_ctx->streams[i];

    in_codecpar = in_stream->codecpar;



    if (in_codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&

        in_codecpar->codec_type != AVMEDIA_TYPE_VIDEO &&

        in_codecpar->codec_type != AVMEDIA_TYPE_SUBTITLE) {

        stream_mapping[i] = -1;

        continue;

    }



    if (in_codecpar->codec_type == AVMEDIA_TYPE_VIDEO){

        pCodec = avcodec_find_decoder(in_codecpar->codec_id);

        stream_mapping[i] = stream_index++;

        break;

    }

}



if(!pCodec){

    PyErr_SetString(ExtractorError, "error, no pCodec!");

    return NULL;

}



// convert CodecParam to CodecCtx

AVCodecContext *pCodecCtx = avcodec_alloc_context3(pCodec);

if (!pCodecCtx) {

    PyErr_SetString(ExtractorError, "Failed to convert codecParam to CodecCtx!");

    return NULL;

}



ret = avcodec_parameters_to_context(pCodecCtx, in_codecpar);

if (ret < 0)

    goto end;



//open video decoder

if (avcodec_open2(pCodecCtx, pCodec, NULL) < 0)

{

    logging("EXTRACTOR: failed to open codec through avcodec_open2n");

    return NULL;

}

In order to test it I used the following command from ffmpeg to extract frames from the same video and it worked fine.

ffmpeg -i ~/thuuz/extractor/03000.ts -f image2 -qscale:v 7 ffmpeg-detect--%03d.jpg -hide_banner -v quiet

Am I not passing the right codec params? Please let me know some inputs on how to debug this issue.
Thanks a lot.