docs-7.0.0/utils/video_demuxer.h Source File

docs-7.0.0/utils/video_demuxer.h Source File#

rocDecode: docs-7.0.0/utils/video_demuxer.h Source File
 /*
 Copyright (c) 2023 - 2025 Advanced Micro Devices, Inc. All rights reserved.
  
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
 in the Software without restriction, including without limitation the rights
 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 copies of the Software, and to permit persons to whom the Software is
 furnished to do so, subject to the following conditions:
  
 The above copyright notice and this permission notice shall be included in
 all copies or substantial portions of the Software.
  
 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 THE SOFTWARE.
 */
  
 #pragma once
  
 #include <iostream>
 extern "C" {
     #include <libavcodec/avcodec.h>
     #include <libavformat/avformat.h>
     #if USE_AVCODEC_GREATER_THAN_58_134
         #include <libavcodec/bsf.h>
     #endif
 }
  
 #include "rocdecode/rocdecode.h"
  
 typedef enum SeekModeEnum {
     SEEK_MODE_EXACT_FRAME = 0,
     SEEK_MODE_PREV_KEY_FRAME = 1,
     SEEK_MODE_NUM,
 } SeekMode;
  
 typedef enum SeekCriteriaEnum {
     SEEK_CRITERIA_FRAME_NUM = 0,
     SEEK_CRITERIA_TIME_STAMP = 1,
     SEEK_CRITERIA_NUM,
 } SeekCriteria;
  
 struct PacketData {
     int32_t key;
     int64_t pts;
     int64_t dts;
     uint64_t pos;
     uintptr_t bsl_data;
     uint64_t bsl;
     uint64_t duration;
 };
  
 class VideoSeekContext {
 public:
     VideoSeekContext()
         : use_seek_(false), seek_frame_(0), seek_mode_(SEEK_MODE_PREV_KEY_FRAME), seek_crit_(SEEK_CRITERIA_FRAME_NUM),
         out_frame_pts_(0), out_frame_duration_(0), num_frames_decoded_(0U) {}
  
     VideoSeekContext(uint64_t frame_id)
         : use_seek_(true), seek_frame_(frame_id), seek_mode_(SEEK_MODE_PREV_KEY_FRAME),
         seek_crit_(SEEK_CRITERIA_FRAME_NUM), out_frame_pts_(0), out_frame_duration_(0), num_frames_decoded_(0U) {}
  
     VideoSeekContext& operator=(const VideoSeekContext& other) {
         use_seek_ = other.use_seek_;
         seek_frame_ = other.seek_frame_;
         seek_mode_ = other.seek_mode_;
         seek_crit_ = other.seek_crit_;
         out_frame_pts_ = other.out_frame_pts_;
         out_frame_duration_ = other.out_frame_duration_;
         num_frames_decoded_ = other.num_frames_decoded_;
         return *this;
     }
  
     /* Will be set to false when not seeking, true otherwise;
      */
     bool use_seek_;
  
     /* Frame we want to get. Set by user.
      * Shall be set to frame timestamp in case seek is done by time.
      */
     uint64_t seek_frame_;
  
     /* Mode in which we seek. */
     SeekMode seek_mode_;
  
     /* Criteria by which we seek. */
     SeekCriteria seek_crit_;
  
     /* PTS of frame found after seek. */
     int64_t out_frame_pts_;
  
     /* Duration of frame found after seek. */
     int64_t out_frame_duration_;
  
     /* Number of frames that were decoded during seek. */
     uint64_t num_frames_decoded_;
  
     /* PTS of frame to seek as set by the user in seek_frame_. */
     int64_t requested_frame_pts_;
 };
  
  
 // Video Demuxer Interface class
 class VideoDemuxer {
     public:
         class StreamProvider {
             public:
                 virtual ~StreamProvider() {}
                 virtual int GetData(uint8_t *buf, int buf_size) = 0;
                 virtual size_t GetBufferSize() = 0;
         };
         AVCodecID GetCodecID() { return av_video_codec_id_; };
         VideoDemuxer(const char *input_file_path) : VideoDemuxer(CreateFmtContextUtil(input_file_path)) {}
         VideoDemuxer(StreamProvider *stream_provider) : VideoDemuxer(CreateFmtContextUtil(stream_provider)) {av_io_ctx_ = av_fmt_input_ctx_->pb;}
         ~VideoDemuxer() {
             if (!av_fmt_input_ctx_) {
                 return;
             }
             if (packet_) {
                 av_packet_free(&packet_);
             }
             if (packet_filtered_) {
                 av_packet_free(&packet_filtered_);
             }
             if (av_bsf_ctx_) {
                 av_bsf_free(&av_bsf_ctx_);
             }
             avformat_close_input(&av_fmt_input_ctx_);
             if (av_io_ctx_) {
                 av_freep(&av_io_ctx_->buffer);
                 av_freep(&av_io_ctx_);
             }
             if (data_with_header_) {
                 av_free(data_with_header_);
             }
         }
         bool Demux(uint8_t **video, int *video_size, int64_t *pts = nullptr) {
             if (!av_fmt_input_ctx_) {
                 return false;
             }
             *video_size = 0;
             if (packet_->data) {
                 av_packet_unref(packet_);
             }
             int ret = 0;
             while ((ret = av_read_frame(av_fmt_input_ctx_, packet_)) >= 0 && packet_->stream_index != av_stream_) {
                 av_packet_unref(packet_);
             }
             if (ret < 0) {
                 return false;
             }
             if (is_h264_ || is_hevc_) {
                 if (packet_filtered_->data) {
                     av_packet_unref(packet_filtered_);
                 }
                 if (av_bsf_send_packet(av_bsf_ctx_, packet_) != 0) {
                     std::cerr << "ERROR: av_bsf_send_packet failed!" << std::endl;
                     return false;
                 }
                 if (av_bsf_receive_packet(av_bsf_ctx_, packet_filtered_) != 0) {
                     std::cerr << "ERROR: av_bsf_receive_packet failed!" << std::endl;
                     return false;
                 }
                 *video = packet_filtered_->data;
                 *video_size = packet_filtered_->size;
                 if (packet_filtered_->dts != AV_NOPTS_VALUE) {
                     pkt_dts_ = packet_filtered_->dts;
                 } else {
                     pkt_dts_ = packet_filtered_->pts;
                 }
                 if (pts) {
                     *pts = (int64_t) (packet_filtered_->pts * default_time_scale_ * time_base_);
                     pkt_duration_ = packet_filtered_->duration;
                 }
             } else {
                 if (is_mpeg4_ && (frame_count_ == 0)) {
                     int ext_data_size = av_fmt_input_ctx_->streams[av_stream_]->codecpar->extradata_size;
                     if (ext_data_size > 0) {
                         data_with_header_ = (uint8_t *)av_malloc(ext_data_size + packet_->size - 3 * sizeof(uint8_t));
                         if (!data_with_header_) {
                             std::cerr << "ERROR: av_malloc failed!" << std::endl;
                             return false;
                         }
                         memcpy(data_with_header_, av_fmt_input_ctx_->streams[av_stream_]->codecpar->extradata, ext_data_size);
                         memcpy(data_with_header_ + ext_data_size, packet_->data + 3, packet_->size - 3 * sizeof(uint8_t));
                         *video = data_with_header_;
                         *video_size = ext_data_size + packet_->size - 3 * sizeof(uint8_t);
                     }
                 } else {
                     *video = packet_->data;
                     *video_size = packet_->size;
                 }
                 if (packet_->dts != AV_NOPTS_VALUE) {
                     pkt_dts_ = packet_->dts;
                 } else {
                     pkt_dts_ = packet_->pts;
                 }
                 if (pts) {
                     *pts = (int64_t)(packet_->pts * default_time_scale_ * time_base_);
                     pkt_duration_ = packet_->duration;
                 }
             }
             frame_count_++;
             return true;
         }
         bool Seek(VideoSeekContext& seek_ctx, uint8_t** pp_video, int* video_size) {
             /* !!! IMPORTANT !!!
                 * Across this function, packet decode timestamp (DTS) values are used to
                 * compare given timestamp against. This is done because DTS values shall
                 * monotonically increase during the course of decoding unlike PTS values
                 * which may be affected by frame reordering due to B frames.
                 */
  
             if (!is_seekable_) {
                 std::cerr << "ERROR: Seek isn't supported for this input." << std::endl;
                 return false;
             }
  
             if (IsVFR() && (SEEK_CRITERIA_FRAME_NUM == seek_ctx.seek_crit_)) {
                 std::cerr << "ERROR: Can't seek by frame number in VFR sequences. Seek by timestamp instead." << std::endl;
                 return false;
             }
             int64_t timestamp = 0;
             // Seek for single frame;
             auto seek_frame = [&](VideoSeekContext const& seek_ctx, int flags) {
                 bool seek_backward = true;
                 int ret = 0;
  
                 switch (seek_ctx.seek_crit_) {
                     case SEEK_CRITERIA_FRAME_NUM:
                         timestamp = TsFromFrameNumber(seek_ctx.seek_frame_);
                         ret = av_seek_frame(av_fmt_input_ctx_, av_stream_, timestamp, seek_backward ? AVSEEK_FLAG_BACKWARD | flags : flags);
                         break;
                     case SEEK_CRITERIA_TIME_STAMP:
                         timestamp = TsFromTime(seek_ctx.seek_frame_);
                         ret = av_seek_frame(av_fmt_input_ctx_, av_stream_, timestamp, seek_backward ? AVSEEK_FLAG_BACKWARD | flags : flags);
                         break;
                     default:
                         std::cerr << "ERROR: Invalid seek mode" << std::endl;
                         ret = -1;
                 }
  
                 if (ret < 0) {
                     throw std::runtime_error("ERROR: seeking for frame");
                 }
             };
  
             // Check if frame satisfies seek conditions;
             auto is_seek_done = [&](PacketData& pkt_data, VideoSeekContext const& seek_ctx) {
                 int64_t target_ts = 0;
  
                 switch (seek_ctx.seek_crit_) {
                     case SEEK_CRITERIA_FRAME_NUM:
                         target_ts = TsFromFrameNumber(seek_ctx.seek_frame_);
                         break;
                     case SEEK_CRITERIA_TIME_STAMP:
                         target_ts = TsFromTime(seek_ctx.seek_frame_);
                         break;
                     default:
                         std::cerr << "ERROR::Invalid seek criteria" << std::endl;
                         return -1;
                 }
  
                 if (pkt_dts_ == target_ts) {
                     return 0;
                 } else if (pkt_dts_ > target_ts) {
                     return 1;
                 } else {
                     return -1;
                 };
             };
  
             /* This will seek for exact frame number;
                 * Note that decoder may not be able to decode such frame; */
             auto seek_for_exact_frame = [&](PacketData& pkt_data, VideoSeekContext& seek_ctx) {
                 // Repetititive seek until seek condition is satisfied;
                 VideoSeekContext tmp_ctx(seek_ctx.seek_frame_);
                 seek_frame(tmp_ctx, AVSEEK_FLAG_ANY);
  
                 int seek_done = 0;
                 do {
                     if (!Demux(pp_video, video_size, &pkt_data.pts)) {
                         throw std::runtime_error("ERROR: Demux failed trying to seek for specified frame number/timestamp");
                     }
                     seek_done = is_seek_done(pkt_data, seek_ctx);
                     //TODO: one last condition, check for a target too high than available for timestamp
                     if (seek_done > 0) { // We've gone too far and need to seek backwards;
                         if ((tmp_ctx.seek_frame_--) >= 0) {
                             seek_frame(tmp_ctx, AVSEEK_FLAG_ANY);
                         }
                     } else if (seek_done < 0) { // Need to read more frames until we reach requested number;
                         tmp_ctx.seek_frame_++;
                         seek_frame(tmp_ctx, AVSEEK_FLAG_ANY);
                     }
                     if (tmp_ctx.seek_frame_ == seek_ctx.seek_frame_) // if frame 'N' is too far and frame 'N-1' is too less from target. Avoids infinite loop between N & N-1 
                         break;
                 } while (seek_done != 0);
  
                 seek_ctx.out_frame_pts_ = pkt_data.pts;
                 seek_ctx.out_frame_duration_ = pkt_data.duration = pkt_duration_;
                 seek_ctx.requested_frame_pts_ = (int64_t) (timestamp * default_time_scale_ * time_base_);
             };
  
             // Seek for closest key frame in the past;
             auto seek_for_prev_key_frame = [&](PacketData& pkt_data, VideoSeekContext& seek_ctx) {
                 seek_frame(seek_ctx, AVSEEK_FLAG_BACKWARD);
                 Demux(pp_video, video_size, &pkt_data.pts);
                 seek_ctx.num_frames_decoded_ = static_cast<uint64_t>(pkt_data.pts / 1000 * frame_rate_);
                 seek_ctx.out_frame_pts_ = pkt_data.pts;
                 seek_ctx.out_frame_duration_ = pkt_data.duration = pkt_duration_;
                 seek_ctx.requested_frame_pts_ = (int64_t) (timestamp * default_time_scale_ * time_base_);
             };
  
             PacketData pktData;
             pktData.bsl_data = size_t(*pp_video);
             pktData.bsl = *video_size;
  
             switch (seek_ctx.seek_mode_) {
             case SEEK_MODE_EXACT_FRAME:
                 seek_for_exact_frame(pktData, seek_ctx);
                 break;
             case SEEK_MODE_PREV_KEY_FRAME:
                 seek_for_prev_key_frame(pktData, seek_ctx);
                 break;
             default:
                 throw std::runtime_error("ERROR::Unsupported seek mode");
                 break;
             }
  
             return true;
         }
         const uint32_t GetWidth() const { return width_;}
         const uint32_t GetHeight() const { return height_;}
         const uint32_t GetChromaHeight() const { return chroma_height_;}
         const uint32_t GetBitDepth() const { return bit_depth_;}
         const uint32_t GetBytePerPixel() const { return byte_per_pixel_;}
         const uint32_t GetBitRate() const { return bit_rate_;}
         const double GetFrameRate() const {return frame_rate_;};
         bool IsVFR() const { return frame_rate_ != avg_frame_rate_; };
         int64_t TsFromTime(double ts_sec) {
             // Convert integer timestamp representation to AV_TIME_BASE and switch to fixed_point
             auto const ts_tbu = llround(ts_sec * AV_TIME_BASE);
             // Rescale the timestamp to value represented in stream base units;
             AVRational time_factor = {1, AV_TIME_BASE};
             return av_rescale_q(ts_tbu, time_factor, av_fmt_input_ctx_->streams[av_stream_]->time_base);
         }
  
         int64_t TsFromFrameNumber(int64_t frame_num) {
             auto const ts_sec = static_cast<double>(frame_num) / frame_rate_;
             return TsFromTime(ts_sec);
         }
  
     private:
         VideoDemuxer(AVFormatContext *av_fmt_input_ctx) : av_fmt_input_ctx_(av_fmt_input_ctx) {
             av_log_set_level(AV_LOG_QUIET);
             if (!av_fmt_input_ctx_) {
                 std::cerr << "ERROR: av_fmt_input_ctx_ is not vaild!" << std::endl;
                 return;
             }
             packet_ = av_packet_alloc();
             packet_filtered_ = av_packet_alloc();
             if (!packet_ || !packet_filtered_) {
                 std::cerr << "ERROR: av_packet_alloc failed!" << std::endl;
                 return;
             }
             if (avformat_find_stream_info(av_fmt_input_ctx_, nullptr) < 0) {
                 std::cerr << "ERROR: avformat_find_stream_info failed!" << std::endl;
                 return;
             }
             av_stream_ = av_find_best_stream(av_fmt_input_ctx_, AVMEDIA_TYPE_VIDEO, -1, -1, nullptr, 0);
             if (av_stream_ < 0) {
                 std::cerr << "ERROR: av_find_best_stream failed!" << std::endl;
                 av_packet_free(&packet_);
                 av_packet_free(&packet_filtered_);
                 return;
             }
             av_video_codec_id_ = av_fmt_input_ctx_->streams[av_stream_]->codecpar->codec_id;
             width_ = av_fmt_input_ctx_->streams[av_stream_]->codecpar->width;
             height_ = av_fmt_input_ctx_->streams[av_stream_]->codecpar->height;
             chroma_format_ = (AVPixelFormat)av_fmt_input_ctx_->streams[av_stream_]->codecpar->format;
             bit_rate_ = av_fmt_input_ctx_->streams[av_stream_]->codecpar->bit_rate;
             if (av_fmt_input_ctx_->streams[av_stream_]->r_frame_rate.den != 0)
                 frame_rate_ = static_cast<double>(av_fmt_input_ctx_->streams[av_stream_]->r_frame_rate.num) / static_cast<double>(av_fmt_input_ctx_->streams[av_stream_]->r_frame_rate.den);
             if (av_fmt_input_ctx_->streams[av_stream_]->avg_frame_rate.den != 0)
                 avg_frame_rate_ = static_cast<double>(av_fmt_input_ctx_->streams[av_stream_]->avg_frame_rate.num) / static_cast<double>(av_fmt_input_ctx_->streams[av_stream_]->avg_frame_rate.den);
  
             switch (chroma_format_) {
                 case AV_PIX_FMT_YUV420P10LE:
                 case AV_PIX_FMT_GRAY10LE:
                     bit_depth_ = 10;
                     chroma_height_ = (height_ + 1) >> 1;
                     byte_per_pixel_ = 2;
                     break;
                 case AV_PIX_FMT_YUV420P12LE:
                     bit_depth_ = 12;
                     chroma_height_ = (height_ + 1) >> 1;
                     byte_per_pixel_ = 2;
                     break;
                 case AV_PIX_FMT_YUV444P10LE:
                     bit_depth_ = 10;
                     chroma_height_ = height_ << 1;
                     byte_per_pixel_ = 2;
                     break;
                 case AV_PIX_FMT_YUV444P12LE:
                     bit_depth_ = 12;
                     chroma_height_ = height_ << 1;
                     byte_per_pixel_ = 2;
                     break;
                 case AV_PIX_FMT_YUV444P:
                     bit_depth_ = 8;
                     chroma_height_ = height_ << 1;
                     byte_per_pixel_ = 1;
                     break;
                 case AV_PIX_FMT_YUV420P:
                 case AV_PIX_FMT_YUVJ420P:
                 case AV_PIX_FMT_YUVJ422P:
                 case AV_PIX_FMT_YUVJ444P:
                 case AV_PIX_FMT_GRAY8:
                     bit_depth_ = 8;
                     chroma_height_ = (height_ + 1) >> 1;
                     byte_per_pixel_ = 1;
                     break;
                 default:
                     chroma_format_ = AV_PIX_FMT_YUV420P;
                     bit_depth_ = 8;
                     chroma_height_ = (height_ + 1) >> 1;
                     byte_per_pixel_ = 1;
                 }
  
             AVRational time_base = av_fmt_input_ctx_->streams[av_stream_]->time_base;
             time_base_ = av_q2d(time_base);
  
             is_h264_ = av_video_codec_id_ == AV_CODEC_ID_H264 && (!strcmp(av_fmt_input_ctx_->iformat->long_name, "QuickTime / MOV") 
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "FLV (Flash Video)") 
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "Matroska / WebM"));
             is_hevc_ = av_video_codec_id_ == AV_CODEC_ID_HEVC && (!strcmp(av_fmt_input_ctx_->iformat->long_name, "QuickTime / MOV")
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "FLV (Flash Video)")
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "Matroska / WebM"));
             is_mpeg4_ = av_video_codec_id_ == AV_CODEC_ID_MPEG4 && (!strcmp(av_fmt_input_ctx_->iformat->long_name, "QuickTime / MOV")
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "FLV (Flash Video)")
                         || !strcmp(av_fmt_input_ctx_->iformat->long_name, "Matroska / WebM"));
  
             // Check if the input file allow seek functionality.
 #if USE_AVCODEC_GREATER_THAN_58_134
             is_seekable_ = true;    //for latest version of FFMPeg, read_seek and read_seek2 is not exposed in AVFormatContext
 #else            
             is_seekable_ = av_fmt_input_ctx_->iformat->read_seek || av_fmt_input_ctx_->iformat->read_seek2;
 #endif            
  
             if (is_h264_) {
                 const AVBitStreamFilter *bsf = av_bsf_get_by_name("h264_mp4toannexb");
                 if (!bsf) {
                     std::cerr << "ERROR: av_bsf_get_by_name() failed" << std::endl;
                     av_packet_free(&packet_);
                     av_packet_free(&packet_filtered_);
                     return;
                 }
                 if (av_bsf_alloc(bsf, &av_bsf_ctx_) != 0) {
                     std::cerr << "ERROR: av_bsf_alloc failed!" << std::endl;
                         return;
                 }
                 avcodec_parameters_copy(av_bsf_ctx_->par_in, av_fmt_input_ctx_->streams[av_stream_]->codecpar);
                 if (av_bsf_init(av_bsf_ctx_) < 0) {
                     std::cerr << "ERROR: av_bsf_init failed!" << std::endl;
                     return;
                 }
             }
             if (is_hevc_) {
                 const AVBitStreamFilter *bsf = av_bsf_get_by_name("hevc_mp4toannexb");
                 if (!bsf) {
                     std::cerr << "ERROR: av_bsf_get_by_name() failed" << std::endl;
                     av_packet_free(&packet_);
                     av_packet_free(&packet_filtered_);
                     return;
                 }
                 if (av_bsf_alloc(bsf, &av_bsf_ctx_) != 0 ) {
                     std::cerr << "ERROR: av_bsf_alloc failed!" << std::endl;
                     return;
                 }
                 avcodec_parameters_copy(av_bsf_ctx_->par_in, av_fmt_input_ctx_->streams[av_stream_]->codecpar);
                 if (av_bsf_init(av_bsf_ctx_) < 0) {
                     std::cerr << "ERROR: av_bsf_init failed!" << std::endl;
                     return;
                 }
             }
         }
         AVFormatContext *CreateFmtContextUtil(StreamProvider *stream_provider) {
             AVFormatContext *ctx = nullptr;
             if (!(ctx = avformat_alloc_context())) {
                 std::cerr << "ERROR: avformat_alloc_context failed" << std::endl;
                 return nullptr;
             }
             uint8_t *avioc_buffer = nullptr;
             int avioc_buffer_size = stream_provider->GetBufferSize();
             avioc_buffer = (uint8_t *)av_malloc(avioc_buffer_size);
             if (!avioc_buffer) {
                 std::cerr << "ERROR: av_malloc failed!" << std::endl;
                 return nullptr;
             }
             av_io_ctx_ = avio_alloc_context(avioc_buffer, avioc_buffer_size,
                     0, stream_provider, &ReadPacket, nullptr, nullptr);
             if (!av_io_ctx_) {
                 std::cerr << "ERROR: avio_alloc_context failed!" << std::endl;
                 return nullptr;
             }
             ctx->pb = av_io_ctx_;
  
             if (avformat_open_input(&ctx, nullptr, nullptr, nullptr) != 0) {
                 std::cerr << "ERROR: avformat_open_input failed!" << std::endl;
                 return nullptr;
             }
             return ctx;
         }
         AVFormatContext *CreateFmtContextUtil(const char *input_file_path) {
             avformat_network_init();
             AVFormatContext *ctx = nullptr;
             if (avformat_open_input(&ctx, input_file_path, nullptr, nullptr) != 0 ) {
                 std::cerr << "ERROR: avformat_open_input failed!" << std::endl;
                 return nullptr;
             }
             return ctx;
         }
         static int ReadPacket(void *data, uint8_t *buf, int buf_size) {
             return ((StreamProvider *)data)->GetData(buf, buf_size);
         }
         AVFormatContext *av_fmt_input_ctx_ = nullptr;
         AVIOContext *av_io_ctx_ = nullptr;
         AVPacket* packet_ = nullptr;
         AVPacket* packet_filtered_ = nullptr;
         AVBSFContext *av_bsf_ctx_ = nullptr;
         AVCodecID av_video_codec_id_;
         AVPixelFormat chroma_format_;
         double frame_rate_ = 0.0;
         double avg_frame_rate_ = 0.0;
         uint8_t *data_with_header_ = nullptr;
         int av_stream_ = 0;
         bool is_h264_ = false; 
         bool is_hevc_ = false;
         bool is_mpeg4_ = false;
         bool is_seekable_ = false;
         int64_t default_time_scale_ = 1000;
         double time_base_ = 0.0;
         uint32_t frame_count_ = 0;
         uint32_t width_ = 0;
         uint32_t height_ = 0;
         uint32_t chroma_height_ = 0;
         uint32_t bit_depth_ = 0;
         uint32_t byte_per_pixel_ = 0;
         uint32_t bit_rate_ = 0;
         // used for Seek Exact frame
         int64_t pkt_dts_ = 0;
         int64_t pkt_duration_ = 0;
 };
  
 static inline rocDecVideoCodec AVCodec2RocDecVideoCodec(AVCodecID av_codec) {
     switch (av_codec) {
         case AV_CODEC_ID_MPEG1VIDEO : return rocDecVideoCodec_MPEG1;
         case AV_CODEC_ID_MPEG2VIDEO : return rocDecVideoCodec_MPEG2;
         case AV_CODEC_ID_MPEG4      : return rocDecVideoCodec_MPEG4;
         case AV_CODEC_ID_H264       : return rocDecVideoCodec_AVC;
         case AV_CODEC_ID_HEVC       : return rocDecVideoCodec_HEVC;
         case AV_CODEC_ID_VP8        : return rocDecVideoCodec_VP8;
         case AV_CODEC_ID_VP9        : return rocDecVideoCodec_VP9;
         case AV_CODEC_ID_MJPEG      : return rocDecVideoCodec_JPEG;
         case AV_CODEC_ID_AV1        : return rocDecVideoCodec_AV1;
         default                     : return rocDecVideoCodec_NumCodecs;
     }
 }