Recherche avancée

Sur d’autres sites (7938)

• How Piwik uses Travis CI to deliver a reliable analytics platform to the community

  26 mai 2014, par Matthieu Aubry — Development, Meta

  In this post, we will explain how the Piwik project uses continuous integration to deliver a quality software platform to dozens of thousands of users worldwide. Read this post if you are interested in Piwik project, Quality Assurance or Automated testing.

  Why do we care about tests ?

  Continuous Integration brings us agility and peace of mind. From the very beginning of the Piwik project, it was clear to us that writing and maintaining automated tests was a necessity, in order to create a successful open source software platform.

  Over the years we have invested a lot of time into writing and maintaining our tests suites. This work has paid off in so many ways ! Piwik platform has fewer bugs, fewer regressions, and we are able to release new minor and major versions frequently.

  Which parts of Piwik software are automatically tested ?

  • Piwik back-end in PHP5 : we use PHPUnit to write and run our PHP tests : unit tests, integration tests, and plugin tests.
  • piwik.js Tracker : the JS tracker is included into all websites that use Piwik. For this reason, it is critical that piwik.js JavaScript tracker always works without any issue or regression. Our Javascript Tracker tests includes both unit and integration tests.
  • Piwik front-end : more recently we’ve started to write JavaScript tests for the user interface partially written in AngularJS.
  • Piwik front-end screenshots tests : after each change to Piwik, more than 150 different screenshots are automatically taken. For example, we take screenshots of each of the 8-step installation process, we take screenshots of the password reset workflow, etc. Each of these screenshot is then compared pixel by pixel, with the “expected” screenshot, and we can automatically detect whether the last code change has introduced an undesired visual change. Learn more about Piwik screenshot tests.

  How often do we run the tests ?

  The tests are executed by Travis CI after each change to the Piwik source code. On average all our tests run 20 times per day. Whenever a Piwik developer pushes some code to Github, or when a community member issues a Pull request, Travis CI automatically runs the tests. In case some of the automated tests started failing after a change, the developer that has made the change is notified by email.

  Should I use Travis CI ?

  Over the last six years, we have used various Continuous Integration servers such as Bamboo, Hudson, Jenkins… and have found that the Travis CI is the ideal continuous integration service for open source projects that are hosted on Github. Travis CI is free for open source projects and the Travis CI team is very friendly and reactive ! If you work on commercial closed source software, you may also use Travis by signing up to Travis CI Pro.

  Summary

  Tests make the Piwik analytics platform better. Writing tests make Piwik contributors better developers. We save a lot of time and effort, and we are not afraid of change !

  Here is the current status of our builds :
  Main build :
  Screenshot tests build :

  PS : If you are a developer looking for a challenge, Piwik is hiring a software developer to join our engineering team in New Zealand or Poland.

• Sound playback using FFmpeg and libsoundio in c++

  11 juillet 2021, par ldall03

  I am trying to make a video player desktop application in c++ using primarily FFmpeg and Qt6. As of for now, I can decode and play video frames correctly at the right speed, that is not a problem. I am now trying to get to playback audio, which is much harder than I expected it to be. I am using libsoundio for my audio library but the documentation is really poor and there are not many examples/tutorials on it. I am also a beginner when it comes to audio programming, although I understand the basics. First off, if anyone can recommend an audio library for this type of job let me know, but I would like to use open source libraries. Anyways, here is how I decode my audio data with FFmpeg. I'm not sure if I am doing it correctly as I could barely find documentation on that as well...
I have a struct that contains all the information which is initiated through a function :

  


  struct VideoReader
{
    bool valid;
    int width, height;
    int video_stream_index;
    int audio_stream_index;
    AVRational time_base;

    AVFormatContext* av_format_ctx;
    AVCodecContext* av_vi_codec_ctx;
    AVCodecContext* av_au_codec_ctx;
    AVPacket* packet;
    AVFrame* frame;
    SwsContext* sws_ctx;
    SwrContext* swr_ctx;
};


  


  The function that initiates it is quite long and is not necessary to share but it populates all those values except for the sws_ctx and the swr_ctx.

  


  Here is how I decode packets, this function is simplified, I left the video decoding out of it, ill take care of syncing once I can properly playback audio :

  


  bool video_reader_read_au_frame(VideoReader *video_reader, unsigned char **frame_buffer)
{
    // Unpack video_reader
    auto& av_format_ctx         = video_reader->av_format_ctx;
    auto& av_codec_ctx          = video_reader->av_au_codec_ctx;
    auto& av_packet             = video_reader->packet;
    auto& av_frame              = video_reader->frame;
    auto& swr_ctx               = video_reader->swr_ctx;
    int&  audio_stream_index    = video_reader->audio_stream_index;

    // Decode the video frame data
    int response;
    while (av_read_frame(av_format_ctx, av_packet) >= 0)
    {
        last_frame = false;
        if (av_packet->stream_index != audio_stream_index)
        {
            av_packet_unref(av_packet);
            continue;
        }

        response = avcodec_send_packet(av_codec_ctx, av_packet);
        if (response < 0)
        {
            Logger::error("Could not decode packet.");
            return false;
        }

        response = avcodec_receive_frame(av_codec_ctx, av_frame);
        if (response == AVERROR(EAGAIN) || response == AVERROR_EOF)
        {
            av_packet_unref(av_packet);
            continue;
        }
        else if (response < 0)
        {
            Logger::error("Could not decode packet.");
            return false;
        }
        av_packet_unref(av_packet);
        break;
    }

    // Initialize SwrContext
    if (!swr_ctx) {
        swr_ctx = swr_alloc_set_opts(nullptr,
                                     av_codec_ctx->channel_layout, AV_SAMPLE_FMT_FLT,
                                     av_codec_ctx->sample_rate, av_codec_ctx->channel_layout,
                                     av_codec_ctx->sample_fmt,  av_codec_ctx->sample_rate,
                                     0, nullptr);
        if (!swr_ctx)
        {
            Logger::error("Could not create SwrContext.");
            return false;
        }

        if (swr_init(swr_ctx) < 0)
        {
            Logger::error("Could not initialize SwrContext.");
            return false;
        }
    }


    const int MAX_BUFFER_SIZE = av_samples_get_buffer_size(nullptr, av_frame->channels, av_frame->nb_samples, AV_SAMPLE_FMT_FLT, 1);
    *frame_buffer = (unsigned char*)av_malloc(MAX_BUFFER_SIZE);
    swr_convert(swr_ctx, frame_buffer, av_frame->nb_samples,
                (const unsigned char**)av_frame->data, av_frame->nb_samples);

    av_frame_unref(av_frame);


    return true;
}


  


  Here is how I would normally call this function :

  


  VideoReader vr{};
if(!video_reader_open(&vr, "C:/Path/to/file.mp4"))
{
    Logger::error("Could not initialize VideoReader.");
    return 1;
}
unsigned char* buffer;
if(!video_reader_read_au_frame(&vr, &buffer))
{
    Logger::error("Could not read audio data.");
    return 1;
}

play_audio(&buffer);  <-- Find a way to play audio once buffer has data in it

video_reader_close(&vr);
return 0;


  


  Obviously I will loop over video_reader_read_au_frame(&vr, &buffer) to playback the whole video.

  


  I believe my code puts the samples from the decoded frame in buffer, but I am really not sure.. I am unsure as well if I need to convert to AV_SAMPLE_FMT_FLT audio format or something else or just leave it as it is. For libsoundio, I kind of understand this example : http://libsound.io/ but I'm not sure I fully understand how this library works, especially the callback function. I know I have to pass buffer in outstream->userdata as a void pointer, but I don't know how to use it in the callback function. Any help or guidance would be greatly appreciated. Note that later on in this project I might want to send this data over a network to play the video on another computer in sync.

  


• RaspberryPi HLS streaming with nginx and ffmpeg ; v4l2 error : ioctl(VIDIOC_STREAMON) : Protocol error

  22 janvier 2021, par Mirco Weber

  I'm trying to realize a baby monitoring with a Raspberry Pi (Model 4B, 4GB RAM) and an ordinary Webcam (with integrated Mic).
I followed this Tutorial : https://github.com/DeTeam/webcam-stream/blob/master/Tutorial.md

  


  Shortly described :

  


  

  1. I installed and configured an nginx server with rtmp module enabled.

  


  2. I installed ffmpeg with this configuration —enable-gpl —enable-nonfree —enable-mmal —enable-omx-rpi

  


  3. I tried to stream ;)

  


  


  The configuration of nginx seems to be working (sometimes streaming works, the server starts without any complication and when the server is up and running, the webpage is displayed).
The configuration of ffmpeg seems to be fine as well, since streaming sometimes works...

  


  I was trying a couple of different ffmpeg-commands ; all of them are sometimes working and sometimes resulting in an error.
The command looks like following :

  


  ffmpeg -re
-f v4l2
-i /dev/video0
-f alsa
-ac 1
-thread_queue_size 4096
-i hw:CARD=Camera,DEV=0
-profile:v high
-level:v 4.1
-vcodec h264_omx
-r 10
-b:v 512k
-s 640x360
-acodec aac
-strict
-2
-ac 2
-ab 32k
-ar 44100
-f flv
rtmp://localhost/show/stream;


  


  Note : I rearranged the code to make it easier to read. In the terminal, it is all in one line.
Note : There is no difference when using -f video4linux2 instead of -f v4l2

  


  The camera is recognized by the system :

  


  pi@raspberrypi:~ $ v4l2-ctl --list-devices
bcm2835-codec-decode (platform:bcm2835-codec):
    /dev/video10
    /dev/video11
    /dev/video12

bcm2835-isp (platform:bcm2835-isp):
    /dev/video13
    /dev/video14
    /dev/video15
    /dev/video16

HD Web Camera: HD Web Camera (usb-0000:01:00.0-1.2):
    /dev/video0
    /dev/video1


  


  When only using -i /dev/video0, audio transmission never worked.
The output of arecord -L was :

  


  pi@raspberrypi:~ $ arecord -L
default
    Playback/recording through the PulseAudio sound server
null
    Discard all samples (playback) or generate zero samples (capture)
jack
    JACK Audio Connection Kit
pulse
    PulseAudio Sound Server
usbstream:CARD=Headphones
    bcm2835 Headphones
    USB Stream Output
sysdefault:CARD=Camera
    HD Web Camera, USB Audio
    Default Audio Device
front:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    Front speakers
surround21:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    2.1 Surround output to Front and Subwoofer speakers
surround40:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    4.0 Surround output to Front and Rear speakers
surround41:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    4.1 Surround output to Front, Rear and Subwoofer speakers
surround50:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    5.0 Surround output to Front, Center and Rear speakers
surround51:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    5.1 Surround output to Front, Center, Rear and Subwoofer speakers
surround71:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    7.1 Surround output to Front, Center, Side, Rear and Woofer speakers
iec958:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    IEC958 (S/PDIF) Digital Audio Output
dmix:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    Direct sample mixing device
dsnoop:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    Direct sample snooping device
hw:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    Direct hardware device without any conversions
plughw:CARD=Camera,DEV=0
    HD Web Camera, USB Audio
    Hardware device with all software conversions
usbstream:CARD=Camera
    HD Web Camera
    USB Stream Output


  


  that's why i added -i hw:CARD=Camera,DEV=0.

  


  As mentioned above, it worked very well a couple of times with this configuration and commands.
But very often, i get the following error message when starting to stream :

  


  pi@raspberrypi:~ $ ffmpeg -re -f video4linux2 -i /dev/video0 -f alsa -ac 1 -thread_queue_size 4096 -i hw:CARD=Camera,DEV=0 -profile:v high -level:v 4.1 -vcodec h264_omx -r 10 -b:v 512k -s 640x360 -acodec aac -strict -2 -ac 2 -ab 32k -ar 44100 -f flv rtmp://localhost/show/stream
ffmpeg version N-100673-g553eb07737 Copyright (c) 2000-2021 the FFmpeg developers
  built with gcc 8 (Raspbian 8.3.0-6+rpi1)
  configuration: --enable-gpl --enable-nonfree --enable-mmal --enable-omx-rpi --extra-ldflags=-latomic
  libavutil      56. 63.101 / 56. 63.101
  libavcodec     58.117.101 / 58.117.101
  libavformat    58. 65.101 / 58. 65.101
  libavdevice    58. 11.103 / 58. 11.103
  libavfilter     7. 96.100 /  7. 96.100
  libswscale      5.  8.100 /  5.  8.100
  libswresample   3.  8.100 /  3.  8.100
  libpostproc    55.  8.100 / 55.  8.100
[video4linux2,v4l2 @ 0x2ea4600] ioctl(VIDIOC_STREAMON): Protocol error
/dev/video0: Protocol error


  


  And when I'm swithing to /dev/video1 (since this was also an output for v4l2-ctl --list-devices), I get the following error message :

  


  pi@raspberrypi:~ $ ffmpeg -re -f v4l2 -i /dev/video1 -f alsa -ac 1 -thread_queue_size 4096 -i hw:CARD=Camera,DEV=0 -profile:v high -level:v 4.1 -vcodec h264_omx -r 10 -b:v 512k -s 640x360 -acodec aac -strict -2 -ac 2 -ab 32k -ar 44100 -f flv rtmp://localhost/show/stream
ffmpeg version N-100673-g553eb07737 Copyright (c) 2000-2021 the FFmpeg developers
  built with gcc 8 (Raspbian 8.3.0-6+rpi1)
  configuration: --enable-gpl --enable-nonfree --enable-mmal --enable-omx-rpi --extra-ldflags=-latomic
  libavutil      56. 63.101 / 56. 63.101
  libavcodec     58.117.101 / 58.117.101
  libavformat    58. 65.101 / 58. 65.101
  libavdevice    58. 11.103 / 58. 11.103
  libavfilter     7. 96.100 /  7. 96.100
  libswscale      5.  8.100 /  5.  8.100
  libswresample   3.  8.100 /  3.  8.100
  libpostproc    55.  8.100 / 55.  8.100
[video4linux2,v4l2 @ 0x1aa4610] ioctl(VIDIOC_G_INPUT): Inappropriate ioctl for device
/dev/video1: Inappropriate ioctl for device


  


  When using the video0 input, the webcam's LED that recognizes an access is constantly on. When using video1not.

  


  After hours and days of googling and tears and whiskey, for the sake of my liver, my marriage and my physical and mental health, I'm very sincerly asking for your help...
What the f**k is happening and what can I do to make it work ???

  


  Thanks everybody :)

  


  UPDATE 1 :

  


  

  1. using the full path to ffmpeg does not change anything...

  


  2. /dev/video0 and /dev/video1 have access rights for everybody

  


  3. sudo ffmpeg ... does not change anything as well

  


  4. the problem seems to be at an "early stage". Stripping the command down to ffmpeg -i /dev/video0 results in the same problem

  


  


  UPDATE 2 :
  
It seems that everything is working when I first start another Application that needs access to the webcam and then ffmpeg...
Might be some driver issue, but when I'm looking for loaded modules with lsmod, there is absolutely no change before and after I started the application...
Any help still appreciated...

  


  UPDATE 3 :
  
I was checking the output of dmesg.
  
When I started the first application I received this message :
  
uvcvideo: Failed to query (GET_DEF) UVC control 12 on unit 2: -32 (exp. 4).

  And when I started ffmpeg, nothing happend but everything worked...