# How transform the c++ code to julia?

**URL:** <https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392>\
**Category:** General Usage\
**Tags:** question, c\
**Created:** [July 18, 2022, 9:38am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392 "2022-07-18T09:38:52Z")\
**Posts on this page:** 13\
**Page:** 1

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 18, 2022, 9:38am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/1 "2022-07-18T09:38:52Z")

</div>

how transform the c++ code to julia ??  
How modify the julia code so that it works as the c++ code?  
ignore the part of visualization (opencv).

code in :

> **[使用ffmpeg绘制运动向量MV\_mb60e54a17708bc的技术博客\_51CTO博客](https://blog.51cto.com/u_15296426/3029796)**
>
> 使用ffmpeg绘制运动向量MV，使用ffmpeg绘制运动向量MV本文将使用ffmpeg提取每帧视频的运动向量MV，并使用opencv将其绘制出来。运动向量MV了解视频编码的人都对运动向量不陌生，它是在进行帧间预测时标记当前块和参考块位置关系的一个向量。帧间预测包括单向预测（P帧）和双向预测（B帧），单向预测只需要一个MV，双向预测需要两个MV。MV在ffmpeg中的定义如下：typedefstructAVMotionVector{/\*\*\*Wherethecurrentm...

use ffmpeg to get and draw motion vector.

c++ code

```julia
typedef struct AVMotionVector {
       /**
        * Where the current macroblock comes from; negative value when it comes
        * from the past, positive value when it comes from the future.
        * XXX: set exact relative ref frame reference instead of a +/- 1 "direction".
        */
       //表明参考块在前面帧（负）还是后面帧（正）
       int32_t source;
        /**
        * Width and height of the block.
         */
       //所属块的宽和高
       uint8_t w, h;
       /**
        * Absolute source position. Can be outside the frame area.
         */
       int16_t src_x, src_y;
       /**
         * Absolute destination position. Can be outside the frame area.
         */
        int16_t dst_x, dst_y;
        /**
         * Extra flag information.
         * Currently unused.
         */
        uint64_t flags;
        /**
         * Motion vector
         * src_x = dst_x + motion_x / motion_scale
         * src_y = dst_y + motion_y / motion_scale
         */
       int32_t motion_x, motion_y;
        uint16_t motion_scale;
    } AVMotionVector;

extern "C"
{
#include <libavutil/motion_vector.h>
#include <libavformat/avformat.h>
}
#include <opencv.hpp>
using namespace cv;
​
static AVFormatContext *fmt_ctx = NULL;
static AVCodecContext *video_dec_ctx = NULL;
static AVStream *video_stream = NULL;
static const char *src_filename = NULL;
​
static int video_stream_idx = -1;
static AVFrame *frame = NULL;
static int video_frame_count = 0;
​
FILE *fout;
VideoWriter out;
​
static int decode_packet(const AVPacket *pkt)
{
    int ret = avcodec_send_packet(video_dec_ctx, pkt);
    if (ret < 0) {
        printf("Error while sending a packet to the decoder: %s\n");
        return ret;
    }
​
    while (ret >= 0) {
        ret = avcodec_receive_frame(video_dec_ctx, frame);
        if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
            break;
        }
        else if (ret < 0) {
            printf("Error while receiving a frame from the decoder: %s\n");
            return ret;
        }
​
        if (ret >= 0) {
            int i;
            AVFrameSideData *sd;
​
            video_frame_count++;
            sd = av_frame_get_side_data(frame, AV_FRAME_DATA_MOTION_VECTORS);
            //获取每帧数据
            cv::Mat yuvImg;
            yuvImg.create(frame->height * 3 / 2, frame->width, CV_8UC1);
            memcpy(yuvImg.data, frame->data[0], frame->linesize[0] * frame->height*sizeof(uint8_t));
            memcpy(yuvImg.data + frame->linesize[0] * frame->height*sizeof(uint8_t), frame->data[1], frame->linesize[1] * frame->height/2*sizeof(uint8_t));
            memcpy(yuvImg.data + (frame->linesize[0] * frame->height + frame->linesize[1] * frame->height / 2)*sizeof(uint8_t), frame->data[2], frame->linesize[2] * frame->height / 2 * sizeof(uint8_t));
            cv::Mat rgbImg;
            cv::cvtColor(yuvImg, rgbImg, CV_YUV2BGR_I420);
            if (sd) {
                const AVMotionVector *mvs = (const AVMotionVector *)sd->data;
                for (i = 0; i < sd->size / sizeof(*mvs); i++) {
                    const AVMotionVector *mv = &mvs[i];
                    //绘制mv
                    line(rgbImg, Point(mv->src_x, mv->src_y), Point(mv->dst_x, mv->dst_y), Scalar(0, 0, 255));
                }
            }
            //将带mv的帧写入文件
            out << rgbImg;
            av_frame_unref(frame);
        }
    }
​
    return 0;
}
​
static int open_codec_context(AVFormatContext *fmt_ctx, enum AVMediaType type)
{
    int ret;
    AVStream *st;
    AVCodecContext *dec_ctx = NULL;
    AVCodec *dec = NULL;
    AVDictionary *opts = NULL;
​
    ret = av_find_best_stream(fmt_ctx, type, -1, -1, &dec, 0);
    if (ret < 0) {
        fprintf(stderr, "Could not find %s stream in input file '%s'\n",
            av_get_media_type_string(type), src_filename);
        return ret;
    }
    else {
        int stream_idx = ret;
        st = fmt_ctx->streams[stream_idx];
​
        dec_ctx = avcodec_alloc_context3(dec);
        if (!dec_ctx) {
            fprintf(stderr, "Failed to allocate codec\n");
            return AVERROR(EINVAL);
        }
​
        ret = avcodec_parameters_to_context(dec_ctx, st->codecpar);
        if (ret < 0) {
            fprintf(stderr, "Failed to copy codec parameters to codec context\n");
            return ret;
        }
​
        /* Init the video decoder */
        av_dict_set(&opts, "flags2", "+export_mvs", 0);
        if ((ret = avcodec_open2(dec_ctx, dec, &opts)) < 0) {
            fprintf(stderr, "Failed to open %s codec\n",
                av_get_media_type_string(type));
            return ret;
        }
​
        video_stream_idx = stream_idx;
        video_stream = fmt_ctx->streams[video_stream_idx];
        video_dec_ctx = dec_ctx;
    }
​
    return 0;
}
​
int main(int argc, char **argv)
{
    fout = fopen("out.yuv","wb");
    //out.open("out.avi", CV_FOURCC('X', 'V', 'I', 'D'),25, Size(640, 272));
    out.open("out.mp4", CV_FOURCC('D', 'I', 'V', 'X'), 25, Size(640, 272));
    int ret = 0;
    AVPacket pkt = { 0 };
​
    if (argc != 2) {
        fprintf(stderr, "Usage: %s <video>\n", argv[0]);
        exit(1);
    }
    src_filename = argv[1];
​
    if (avformat_open_input(&fmt_ctx, src_filename, NULL, NULL) < 0) {
        fprintf(stderr, "Could not open source file %s\n", src_filename);
        exit(1);
    }
​
    if (avformat_find_stream_info(fmt_ctx, NULL) < 0) {
        fprintf(stderr, "Could not find stream information\n");
        exit(1);
    }
​
    open_codec_context(fmt_ctx, AVMEDIA_TYPE_VIDEO);
​
    av_dump_format(fmt_ctx, 0, src_filename, 0);
​
    if (!video_stream) {
        fprintf(stderr, "Could not find video stream in the input, aborting\n");
        ret = 1;
        goto end;
    }
​
    frame = av_frame_alloc();
    if (!frame) {
        fprintf(stderr, "Could not allocate frame\n");
        ret = AVERROR(ENOMEM);
        goto end;
    }
​
    printf("framenum,source,blockw,blockh,srcx,srcy,dstx,dsty,flags\n");
​
    /* read frames from the file */
    while (av_read_frame(fmt_ctx, &pkt) >= 0) {
        if (pkt.stream_index == video_stream_idx)
            ret = decode_packet(&pkt);
        av_packet_unref(&pkt);
        if (ret < 0)
            break;
    }
​
    /* flush cached frames */
    decode_packet(NULL);
​
end:
    avcodec_free_context(&video_dec_ctx);
    avformat_close_input(&fmt_ctx);
    av_frame_free(&frame);
    fclose(fout);
    system("pause");
    return ret < 0;
}

```

julia code (unfinished, How modify it so that it works?)

```julia
using Printf
using VideoIO
using VideoIO.libffmpeg
# using OpenCV

function decode_packet(pkt::AVPacket)::Int
    ret = avcodec_send_packet(video_dec_ctx, pkt)
    if (ret < 0) 
        println("Error while sending a packet to the decoder: %s\n")
        return ret
    end
    while ret >= 0
        ret = avcodec_receive_frame(video_dec_ctx, frame)
        if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
            break
        elseif ret < 0
            println("Error while receiving a frame from the decoder: %s\n")
            return ret
        end

        if ret >= 0
            # sd = Ptr{AVFrameSideData}
            sd = av_frame_get_side_data(frame, AV_FRAME_DATA_MOTION_VECTORS)
            # 获取每帧数据
            # cv::Mat yuvImg
            # yuvImg.create(frame->height * 3 / 2, frame->width, CV_8UC1)
            # memcpy(yuvImg.data, frame->data[0], frame->linesize[0] * frame->height*sizeof(uint8_t))
            # memcpy(yuvImg.data + frame->linesize[0] * frame->height*sizeof(uint8_t), frame->data[1], frame->linesize[1] * frame->height/2*sizeof(uint8_t))
            # memcpy(yuvImg.data + (frame->linesize[0] * frame->height + frame->linesize[1] * frame->height / 2)*sizeof(uint8_t), frame->data[2], frame->linesize[2] * frame->height / 2 * sizeof(uint8_t))
            # # cv::Mat rgbImg
            # cv::cvtColor(yuvImg, rgbImg, CV_YUV2BGR_I420)
            # if sd
            # const AVMotionVector *mvs = (const AVMotionVector *)sd->data
            # for i in sd->size / sizeof(*mvs) 
            # AVMotionVector *mv = &mvs[i]
            # # 绘制mv
            # line(rgbImg, Point(mv->src_x, mv->src_y), Point(mv->dst_x, mv->dst_y), Scalar(0, 0, 255))
            # end
            # end
            # # 将带mv的帧写入文件
            # # out << rgbImg
            # av_frame_unref(frame)
        end

    end
    return 0
end

function open_codec_context(fmt_ctx::AVFormatContext, type::AVMediaType)::Int
    dec = AVCodec[]
    opts = AVDictionary[]
    ret = av_find_best_stream(fmt_ctx, type, -1, -1, dec, 0)
    if (ret < 0) 
        @printf(stderr, "Could not find %s stream in input file '%s'\n", av_get_media_type_string(type), src_filename)
        return ret
    else
        stream_idx = ret
        st = fmt_ctx->streams[stream_idx]
        dec_ctx = avcodec_alloc_context3(dec)
        if (!dec_ctx)
            @printf(stderr, "Failed to allocate codec\n")
            return AVERROR(EINVAL)
        end
        ret = avcodec_parameters_to_context(dec_ctx, st->codecpar)
        if (ret < 0) 
            @printf(stderr, "Failed to copy codec parameters to codec context\n")
            return ret
        end
        # Init the video decoder 
        av_dict_set(opts, "flags2", "+export_mvs", 0)
        if ((ret = avcodec_open2(dec_ctx, dec, opts)) < 0)
            @printf(stderr, "Failed to open %s codec\n", av_get_media_type_string(type))
            return ret
        end
        video_stream_idx = stream_idx
        video_stream = fmt_ctx->streams[video_stream_idx]
        video_dec_ctx = dec_ctx
    end

    return 0
end

function main(video_path)
    # fout = fopen("out.yuv","wb")
    # out = VideoIO.open("out.mp4", CV_FOURCC('D', 'I', 'V', 'X'), 25, Size(640, 272))
    ret = 0
    # Create a `Ref{}` pointing to type T, we'll unpack into that. its unpacked contents:pkt[]
    pkt = Ref{AVPacket}() # 引用,类似指针

    src_filename = video_path
    # fmt_ctx = Ref{AVFormatContext}() # AVFormatContext[]
    fmt_ctx = Ptr{AVFormatContext}() # Ptr<:Ref pointer_from_objref() 取指针/地址

    if avformat_open_input(pointer_from_objref(fmt_ctx), src_filename, C_NULL, C_NULL) < 0 # C_NULL
        @printf(stderr, "Could not open source file %s\n", src_filename)
        exit(1)
    end

    if avformat_find_stream_info(fmt_ctx, C_NULL) < 0
        @printf(stderr, "Could not find stream information\n")
        exit(1)
    end

    open_codec_context(fmt_ctx, AVMEDIA_TYPE_VIDEO)

    av_dump_format(fmt_ctx, 0, src_filename, 0)

    if !video_stream
        @printf(stderr, "Could not find video stream in the input, aborting\n")
        ret = 1
        @goto aa
    end

    frame = av_frame_alloc()
    if (!frame)
        @printf(stderr, "Could not allocate frame\n")
        ret = AVERROR(ENOMEM)
        @goto aa
    end

    @printf("framenum,source,blockw,blockh,srcx,srcy,dstx,dsty,flags\n")

    # read frames from the file
    while av_read_frame(fmt_ctx, pkt) >= 0
        if pkt.stream_index == video_stream_idx
            ret = decode_packet(pkt)
        end
        av_packet_unref(pkt)
        if ret < 0
            break
        end
    end

    # flush cached frames
    decode_packet(C_NULL)

    @label aa
    avcodec_free_context(video_dec_ctx)
    avformat_close_input(fmt_ctx)
    av_frame_free(frame)
    system("pause")
    return ret < 0

end

video_path = "/data/exps/merge.mp4"
main(video_path)

```

you can ignore the part of visualization (opencv).

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 18, 2022, 2:35pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/3 "2022-07-18T14:35:44Z")

</div>

> [@zsz00](#):
>
> ```julia
> AVFrameSideData *sd
> 
> ```

You’ll have to define the `AVFrameSideData` type.

```julia
sd = Ptr{AVFrameSideData}(C_NULL)

```

Let’s port a struct from C++ to Julia:

> [@zsz00](#):
>
> ```julia
> typedef struct AVMotionVector {
> int32_t source;
> uint8_t w, h;
> int16_t src_x, src_y;
> int16_t dst_x, dst_y;
> uint64_t flags;
> int32_t motion_x, motion_y;
> uint16_t motion_scale;
> } AVMotionVector;
> 
> ```

```julia
struct AVMotionVector
    source::Int32
    w::UInt8
    h::UInt8
    src_x::UInt16
    src_y::UInt16
    dst_x::UInt16
    dst_y::UInt16
    flags::UInt64
    motion_x::Int32
    motion_y::Int32
    motion_scale::Uint16
end

```

Rather than porting the entire code base, perhaps you should consider using CxxWrap.jl to wrap the library and produce Julia bindings rather than porting the entire program to Julia:

> **[GitHub - JuliaInterop/CxxWrap.jl: Package to make C++ libraries available in...](https://github.com/JuliaInterop/CxxWrap.jl)**
>
> Package to make C++ libraries available in Julia. Contribute to JuliaInterop/CxxWrap.jl development by creating an account on GitHub.

Via [binarybuilder.org](http://binarybuilder.org) we have the capability of distributing binaries.

---

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 18, 2022, 2:45pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/4 "2022-07-18T14:45:56Z")

</div>

> [@mkitti](#):
>
> CxxWrap

All functions in ffmpeg are considered using CxxWrap.jl. there only 2 custom function need to transform to julia. using the VideoIO.libffmpeg can get AVMotionVector and other functions or Types.

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 18, 2022, 2:47pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/5 "2022-07-18T14:47:54Z")

</div>

> [@zsz00](#):
>
> ```julia
> video_frame_count++
> 
> ```

Julia translation

```julia
            video_frame_count += 1

```

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 18, 2022, 2:50pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/6 "2022-07-18T14:50:24Z")

</div>

> [@mkitti](#):
>
> You’ll have to define the `AVFrameSideData` type.
> 
> ```julia
> sd = Ptr{AVFrameSideData}(C_NULL)
> 
> ```

Alternatively, you could just treat this as a void pointer and just use `Ptr{Nothing}` or initialize with `C_NULL`.

---

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 18, 2022, 2:51pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/7 "2022-07-18T14:51:22Z")

</div>

```julia
static AVFormatContext *fmt_ctx = NULL;

if (avformat_open_input(&fmt_ctx, src_filename, NULL, NULL) < 0) {

```

to julia :

```julia
fmt_ctx = Ptr{AVFormatContext}(C_NULL) # Ptr<:Ref pointer_from_objref() 取指针/地址

if avformat_open_input(pointer_from_objref(fmt_ctx), src_filename, C_NULL, C_NULL) < 0 # C_NULL
    @printf(stderr, "Could not open source file %s\n", src_filename)
    exit(1)
end

```

ERROR: LoadError: pointer\_from\_objref cannot be used on immutable objects

 ![image](https://global.discourse-cdn.com/julialang/original/3X/1/d/1d26539e493f09bfc355cef6b4e93d16a51e92ed.png)

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 18, 2022, 3:06pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/8 "2022-07-18T15:06:41Z")

</div>

For reference, here is the documentation for `avformat_open_input`.

[https://www.ffmpeg.org/doxygen/2.8/group\_\_lavf\_\_decoding.html#ga10a404346c646e4ab58f4ed798baca32](https://www.ffmpeg.org/doxygen/2.8/group __lavf__ decoding.html#ga10a404346c646e4ab58f4ed798baca32)

```nohighlight
int avformat_open_input (AVFormatContext **ps, const char *filename, AVInputFormat *fmt, AVDictionary** options)

```

Over in VideoIO.jl, the authors have already written a very nice wrapper around `avformat_open_input` called `open_avinput`:

> **[GitHub Code Search (Preview)](https://cs.github.com/JuliaIO/VideoIO.jl/blob/7c8cc7f55fc23d73ad7c0b2f6eae796db5ac3ac8/src/avio.jl#L203)**
>
> GitHub Code Search (Preview)

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 18, 2022, 3:12pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/10 "2022-07-18T15:12:02Z")

</div>

Yes, but you seem to be trying to use the low level API while someone has already done a lot of hard work for you.

---

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 18, 2022, 3:14pm UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/11 "2022-07-18T15:14:05Z")

</div>

ok, i will try use open\_avinput to replace avformat\_open\_input… thank you

---

<div class="post-metadata">

**Author:** ![mkitti](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mkitti/32/12459_2.png) [@mkitti](https://discourse.julialang.org/u/mkitti)\
**Post date:** [July 19, 2022, 4:24am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/12 "2022-07-19T04:24:09Z")

</div>

I am going to ping @ianshmean and @BioTurboNick for you since I think they are more familiar with VideoIO.jl.

---

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 19, 2022, 4:29am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/13 "2022-07-19T04:29:22Z")

</div>

thank you, and I find only use high API not enough, I want to get motion vector data, this need modificat level function decode …  
I still don’t know how to modificat to julia.

---

<div class="post-metadata">

**Author:** ![ianshmean](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ianshmean/32/216042_2.png) [@ianshmean](https://discourse.julialang.org/u/ianshmean)\
**Post date:** [July 19, 2022, 4:58am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/14 "2022-07-19T04:58:49Z")

</div>

I’m afraid I can’t help greatly here. When I’ve converted c++ code across like this before it’s taken a lot of trial and error and learning as I go, most of which has left my memory.

I’m going to forward the ping to @Gnimuc or perhaps @melonedo

---

<div class="post-metadata">

**Author:** ![zsz00](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zsz00/32/6314_2.png) [@zsz00](https://discourse.julialang.org/u/zsz00)\
**Post date:** [July 19, 2022, 5:12am UTC](https://discourse.julialang.org/t/how-transform-the-c-code-to-julia/84392/15 "2022-07-19T05:12:49Z")

</div>

I think Is it possible to join

> sd = av\_frame\_get\_side\_data(frame, AV\_FRAME\_DATA\_MOTION\_VECTORS)

around here ?

> <https://github.com/JuliaIO/VideoIO.jl/blob/7c8cc7f55fc23d73ad7c0b2f6eae796db5ac3ac8/src/avio.jl#L445>