Recherche avancée

Sur d’autres sites (10572)

• After upgrade ffmpeg code doesn't working clip build

  11 février, par Tchoune

  I have a problem after upgrading ffmpeg from 4.2.2 to 5.2.2, my code no longer works. When I upload a video to my React-Native application I get a file corruption error on my FFmpeg python agent.
-> sends to Laravel which stores the video on the minio storage ; the video is available -> sends http to the minio key to download locally the mp4 video is corrupted on the minio too...
I have the impression that it's an error downloading the video locally that makes the video corrupt, but I have no idea how I can debug this problem.
If I upload the video directly from my web interface I don't have this problem. The only difference is processClipSynchronously which is set to True on mobile and False on web.

  


  Laravel Agent send to python microservice :

  


   // Store uploaded video file
        $videoFilePath = $this->storeVideoFile($learningGoal, $videoFile);

        // Add video to storyboard
        $agentResponse =  Http::post($this->agentUrl . 'learning-goals/' . $learningGoal->id . '/storyboards/' . $storyboardId .  '/chapters/' . $chapterId . '/videos',
            [
                'clip' => $videoFilePath,
                'processClipSynchronously' => $processClipSynchronously
            ]);


  


  Python agent video :

  


  @app.route('/learning-goals//storyboards//chapters//videos',
           methods=['post'])
def post_storyboard_videos(learning_goal_id, storyboard_id, chapter_id):
    storyboard = get_storyboard(learning_goal_id, storyboard_id)
    chapter, position = get_chapter(storyboard, chapter_id)

    if 'clip' in request.get_json():
        chapter['clip'] = request.get_json()['clip']
        if 'duration' in storyboard:
            del chapter['duration']
        if 'thumbnail' in storyboard:
            del chapter['thumbnail']
        if 'ncAudioPreviewPath' in chapter:
            del chapter['ncAudioPreviewPath']
        if 'trim_start' in chapter:
            del chapter['trim_start']
        if 'trim_end' in chapter:
            del chapter['trim_end']
        if 'perform_nc' in chapter:
            del chapter['perform_nc']
    else:
        abort(400)

    new_storyboard = create_new_version_storyboard(storyboard)

    if 'processClipSynchronously' in request.get_json() and request.get_json()['processClipSynchronously']:
        treat_clip(new_storyboard, chapter) #Mobile trigger here
    else:
        thread = StoppableThread(target=treat_clip, args=(new_storyboard, chapter))
        thread.daemon = True
        thread.start()

    chapter, position = get_chapter(new_storyboard, chapter_id)

    return json.loads(dumps(chapter))

def treat_clip(storyboard, chapter):
    logging.info(
        'start treating clip (' + chapter['clip'] + ') for learning goal : ' + str(storyboard['learningGoalId']))
    file = app.config['VOLUME_PATH'] + chapter['clip']
    os.makedirs(dirname(file), exist_ok=True)
    temp_files_to_remove = []

    if not os.path.exists(file):
        # Download file from S3 storage.
        s3.download_file(chapter['clip'], file)
        # Clean the file at the end (it's already in S3).
        temp_files_to_remove.append(file)
    else:
        logging.warn(f'Not downloading {chapter["clip"]} from S3 as it already exists on the filesystem')

    resolution_width, resolution_height = get_resolution(file)
    is_rotated_video = is_rotated(file)
    sample_aspect_ratio = get_sample_aspect_ratio(file)
    frame_rate = get_frame_rate(file)
    if not file.endswith(
            '.mp4') or resolution_width != 1920 or resolution_height != 1080 or is_rotated_video or sample_aspect_ratio != '1:1' or frame_rate > 60:
        chapter['clip'] = format_video(chapter['clip'], resolution_width, resolution_height, frame_rate,
                                       is_rotated_video, str(storyboard['learningGoalId']), 1920, 1080)
        file = app.config['VOLUME_PATH'] + chapter['clip']

        # Update file to S3 storage
        s3.upload_file(file, chapter['clip'])

        # Clean the new file at the end.
        temp_files_to_remove.append(file)

    clip = VideoFileClip(file)
    chapter['duration'] = float(clip.duration)
    thumbnail_relative_path = create_video_thumbnail(storyboard, clip, 0)
    ....


  


  It's VideoFileClip from moviepy who generate error : Moov atom not found
I think S3 not have time to download file, and corrumpt, but I don't know how to test or fix that

  


  thanks in advance

  


• avcodec/cbs_h266 : Fix regression in DVB clip introduced by 93281630a71c06642adfebebb0...

  10 novembre 2024, par Nuo Mi

  avcodec/cbs_h266 : Fix regression in DVB clip introduced by 93281630a71c06642adfebebb0d4b105a4e02e91
  This commit introduced a regression to VVC_HDR_UHDTV1_OpenGOP_3840x2160_50fps_HLG10_mosaic.ts.
  Root Cause :
  
  The AV_CEIL_RSHIFT(a, b) macro uses bit tricks that work only when -a is a negative value.
  
  However, due to integer promotion rules, this behavior does not extend to the unsigned int type.
  See "6.3.1.1 Boolean, characters, and integers" in the "ISO/IEC 9899" for details.
  Reported-by : Frank Plowman <post@frankplowman.com>
  

  • [DH] libavcodec/cbs_h266_syntax_template.c

• avformat/concatdec : clip outpoint - inpoint overflow in get_best_effort_duration()

  26 décembre 2023, par Michael Niedermayer

  avformat/concatdec : clip outpoint - inpoint overflow in get_best_effort_duration()
  An alternative would be to limit all time/duration fields to below 64bit
  Fixes : signed integer overflow : -93000000 - 9223372036839000000 cannot be represented in type 'long long'
  
  Fixes : 64546/clusterfuzz-testcase-minimized-ffmpeg_dem_CONCAT_fuzzer-5110813828186112
  Found-by : continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavformat/concatdec.c