Recherche avancée

Sur d’autres sites (7431)

• Evolution #4256 : Faire un signalement des mise à jour de sécu

  11 janvier 2019, par Franck D

  Salut marcimat, Oui, c’est pour cela que je disais que cela allait un peu avec 3509, mais en même temps, simplement le signalement d’une mise à jour de secu ne demande pas forcément d’avoir le changelog, cela ne serait qu’un plus !
  Celui voulant savoir, pourrait toujours faire comme actuellement, une lecture des logs de commit (ce n’est pas génial, mais bon...)
  Il ne faut pas que les gens se rendent compte qu’une mise à jour de secu est dispo uniquement car ils vont voir la page de gestion des plugins, car les administrateurs ou webmestres n’y vont pas systématiquement à chaque fois qu’ils vont sur un site.
  C’est pour cela qu’il faut qu’ils reçoivent l’info dès qu’ils arrivent sur le site (une fenêtre qui s’ouvre, une couleur de fond qui change, l’envoi d’un mail, un message par pigeon voyageur, ????), quelque chose qui fasse que les gens se disent "mon site est vulnérable, il faut que je regarde les plugins qui ne sont pas à jour"

  C’est pour ça que je pensais à, un génie de plus dans spip qui regarderait un simple fichier archivelist_secu.txt (après reste à bien réfléchir comment doit être indiquer qu’un plug est vulnérable (je ne sais pas si ce que je proposais était suffisant (car j’ai peur que l’on indique que des versions ont des problèmes alors qu’il ne sont pas concernés), possible qu’il faudrait plutôt un truc du genre :
  Nom_du_prefix_du_plugin/Version_du_X_Y_Z_qui_contient_le_problème/Version_du_plugin_sans_problème

  Sachant que si la personne écrit par exemple : saisie / version 1.x.x / 1.42.11 (alors cela voudrait dire que toutes les versions 1.x.x ont un problème)

• Revision 118399 : paquet.xml et plugin.xml doivent avoir le même numéro de version car ...

  3 novembre 2019, par spip.franck@… — Log

  paquet.xml et plugin.xml doivent avoir le même numéro de version car il s’agit du même plug, j’en profite pour monter la version de nospam car il n’y a pas eu de changement de "x" https://zone.spip.org/trac/spip-zone/changeset/118075/spip-zone

• send h264 video to nginx-rtmp server using ffmpeg API

  11 décembre 2019, par Glen

  I have C++ code that grabs frames from a GigE camera and writes them out to a file. I’m using the libx264 codec and ffmpeg version 4.0.

  Writing to the file works fine, however I would also like to send the video to nginx configured with the nginx-rtmp plug-in to make the video available live via HLS.

  I can use the ffmpeg command line program to stream one of my previously captured files to my nginx server and rebroadcast as HLS, however if I try to stream from my C++ code the nginx server closes the connection after one or two frames are sent.

  To test further, I used the ffmpeg command line program to receive a rtmp stream and write it out to a file. I am able to send video to ffmpeg from my C++ program with rtmp, however every frame generates a warning like this :

  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1771, current: 53; changing to 1772. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1772, current: 53; changing to 1773. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1773, current: 53; changing to 1774. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1774, current: 53; changing to 1775. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1775, current: 53; changing to 1776. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1776, current: 53; changing to 1777. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1777, current: 53; changing to 1778. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1778, current: 53; changing to 1779. This may result in incorrect timestamps in the output file.
  
  [avi @ 0x1b6b6f0] Non-monotonous DTS in output stream 0:0; previous: 1779, current: 53; changing to 1780. This may result in incorrect timestamps in the output file.

  I printed PTS and DTS for my packet before writing it, and the numbers were monotonous (for example, in this last frame the pts and dts printed from my code were 1780, not the ’current : 53’ that ffmpeg reports>

  also, unless I tell ffmpeg what the output framerate should be I end up with a file that plays 2x speed.

  After ffmpeg receives the rtmp stream and writes it to the file, I am then able to successfully send that file to my nginx server using ffmpeg.

  here is some relevant code :

  //configuring the codec context
  
  // make sure that config.codec is something we support
  
  // for now we are only supporting LIBX264
  
  if (config.codec() != codecs::LIBX264) {
  
      throw std::invalid_argument("currently only libx264 codec is supported");
  
  }
  
  
  
  // lookup specified codec
  
  ffcodec_ = avcodec_find_encoder_by_name(config.codec().c_str());
  
  if (!ffcodec_) {
  
      throw std::invalid_argument("unable to get codec " + config.codec());
  
  }
  
  
  
  // unique_ptr to manage the codec_context
  
  codec_context_ = av_pointer::codec_context(avcodec_alloc_context3(ffcodec_));
  
  
  
  if (!codec_context_) {
  
      throw std::runtime_error("unable to initialize AVCodecContext");
  
  }
  
  
  
  // setup codec_context_
  
  codec_context_->width = frame_width;
  
  codec_context_->height = frame_height;
  
  codec_context_->time_base = (AVRational){1, config.target_fps()};
  
  codec_context_->framerate = (AVRational){config.target_fps(), 1};
  
  codec_context_->global_quality = 0;
  
  codec_context_->compression_level = 0;
  
  codec_context_->bits_per_raw_sample = 8;
  
  codec_context_->gop_size = 1;
  
  codec_context_->max_b_frames = 1;
  
  codec_context_->pix_fmt = AV_PIX_FMT_YUV420P;
  
  
  
  // x264 only settings
  
  if (config.codec() == codecs::LIBX264) {
  
      av_opt_set(codec_context_->priv_data, "preset", config.compression_target().c_str(), 0);
  
      av_opt_set(codec_context_->priv_data, "crf", std::to_string(config.crf()).c_str(), 0);
  
  }
  
  
  
  // Open up the codec
  
  if (avcodec_open2(codec_context_.get(), ffcodec_, NULL) < 0) {
  
      throw std::runtime_error("unable to open ffmpeg codec");
  
  }
  
  
  
  // setup the output format context and stream for RTMP
  
  AVFormatContext *tmp_f_context;
  
  avformat_alloc_output_context2(&tmp_f_context, NULL, "flv", uri.c_str());
  
  rtmp_format_context_ = av_pointer::format_context(tmp_f_context);
  
  rtmp_stream_ = avformat_new_stream(rtmp_format_context_.get(), ffcodec_);
  
  avcodec_parameters_from_context(rtmp_stream_->codecpar, codec_context_.get());
  
  rtmp_stream_->time_base = codec_context_->time_base;
  
  rtmp_stream_->r_frame_rate = codec_context_->framerate;
  
  
  
  /* open the output file */
  
  if (!(rtmp_format_context_->flags & AVFMT_NOFILE)) {
  
      int r = avio_open(&rtmp_format_context_->pb, uri.c_str(), AVIO_FLAG_WRITE);
  
      if (r < 0) {
  
          throw std::runtime_error("unable to open " + uri + " : " + av_err2str(r));
  
      }
  
  }
  
  
  
  if (avformat_write_header(rtmp_format_context_.get(), NULL) < 0) {
  
      throw std::runtime_error("unable to write header");
  
  }
  
  
  
  av_dump_format(rtmp_format_context_.get(), 0,uri.c_str() , 1);

  at this point the av_dump_format produces this output :

  Output #0, flv, to 'rtmp://[MY URI]':
  
    Metadata:
  
      encoder         : Lavf58.12.100
  
      Stream #0:0, 0, 1/1000: Video: h264 (libx264), 1 reference frame ([7][0][0][0] / 0x0007), yuv420p, 800x800 (0x0), 0/1, q=-1--1, 30 tbr, 1k tbn

  encoding and writing the frame :

  // send the frame to the encoder, filtering first if necessary
  
  void VideoWriter::Encode(AVFrame *frame)
  
  {
  
      int rval;
  
      if (!apply_filter_) {
  
          //send frame to encoder
  
          rval = avcodec_send_frame(codec_context_.get(), frame);
  
          if (rval < 0) {
  
              throw std::runtime_error("error sending frame for encoding");
  
          }
  
      } else {
  
          // push frame to filter 
  
          // REMOVED, currently testing without filtering
  
      }
  
  
  
      // get packets from encoder
  
      while (rval >= 0) {
  
          // create smart pointer to allocated packet
  
          av_pointer::packet pkt(av_packet_alloc());
  
          if (!pkt) {
  
              throw std::runtime_error("unable to allocate packet");
  
          }
  
  
  
          rval = avcodec_receive_packet(codec_context_.get(), pkt.get());
  
          if (rval == AVERROR(EAGAIN) || rval == AVERROR_EOF) {
  
              return;
  
          } else if (rval < 0) {
  
              throw std::runtime_error("error during encoding");
  
          }
  
  
  
          // if I print pkt->pts and pkt->dts here, I see sequential numbers
  
  
  
          // write packet
  
          rval = av_interleaved_write_frame(rtmp_format_context_.get(), pkt.get());
  
          if (rval < 0 ) {
  
              std::cerr << av_err2str(rval) << std::endl;
  
          }
  
      }
  
  }

  Since I am able to send video from a previously recorded file to nginx with the ffmpeg command line program, I believe the problem is in my code and not my nginx configuration.

  EDIT : I think it may have to do with SPS/PPS as I see a bunch of these error messages in the nginx log before it closes the stream

  2019/12/11 11:11:31 [error] 10180#0: *4 hls: failed to read 5 byte(s), client: XXX, server: 0.0.0.0:1935
  
  2019/12/11 11:11:31 [error] 10180#0: *4 hls: error appenging SPS/PPS NALs, client: XXX, server: 0.0.0.0:1935

  As I mentioned, this code works fine if I set it up to write to an avi file rather stream to rtmp, and I can stream to ffmpeg listening for rtmp but with lots of warnings about the DTS but if I try to send to nginx, it closes the connection almost immediately. My first thought was that there was something wrong with the frame timestamps, but when I print pts and dts prior to writing the packet to the stream they look okay to me.

  My end goal is to capture video to a file, and also be able to turn on the rtmp stream on demand — but for now I’m just trying to get the rtmp stream working continuously (without writing to a file)

  Thanks for any insights.