Exemplo n.º 1
0
  public boolean recordSamples(int sampleRate, int audioChannels, Buffer... samples)
      throws Exception {
    if (audio_st == null) {
      throw new Exception(
          "No audio output stream (Is audioChannels > 0 and has start() been called?)");
    }
    int ret;

    if (sampleRate <= 0) {
      sampleRate = audio_c.sample_rate();
    }
    if (audioChannels <= 0) {
      audioChannels = audio_c.channels();
    }
    int inputSize = samples != null ? samples[0].limit() - samples[0].position() : 0;
    int inputFormat = AV_SAMPLE_FMT_NONE;
    int inputChannels = samples != null && samples.length > 1 ? 1 : audioChannels;
    int inputDepth = 0;
    int outputFormat = audio_c.sample_fmt();
    int outputChannels = samples_out.length > 1 ? 1 : audio_c.channels();
    int outputDepth = av_get_bytes_per_sample(outputFormat);
    if (samples != null && samples[0] instanceof ByteBuffer) {
      inputFormat = samples.length > 1 ? AV_SAMPLE_FMT_U8P : AV_SAMPLE_FMT_U8;
      inputDepth = 1;
      for (int i = 0; i < samples.length; i++) {
        ByteBuffer b = (ByteBuffer) samples[i];
        if (samples_in[i] instanceof BytePointer
            && samples_in[i].capacity() >= inputSize
            && b.hasArray()) {
          ((BytePointer) samples_in[i]).position(0).put(b.array(), b.position(), inputSize);
        } else {
          samples_in[i] = new BytePointer(b);
        }
      }
    } else if (samples != null && samples[0] instanceof ShortBuffer) {
      inputFormat = samples.length > 1 ? AV_SAMPLE_FMT_S16P : AV_SAMPLE_FMT_S16;
      inputDepth = 2;
      for (int i = 0; i < samples.length; i++) {
        ShortBuffer b = (ShortBuffer) samples[i];
        if (samples_in[i] instanceof ShortPointer
            && samples_in[i].capacity() >= inputSize
            && b.hasArray()) {
          ((ShortPointer) samples_in[i])
              .position(0)
              .put(b.array(), samples[i].position(), inputSize);
        } else {
          samples_in[i] = new ShortPointer(b);
        }
      }
    } else if (samples != null && samples[0] instanceof IntBuffer) {
      inputFormat = samples.length > 1 ? AV_SAMPLE_FMT_S32P : AV_SAMPLE_FMT_S32;
      inputDepth = 4;
      for (int i = 0; i < samples.length; i++) {
        IntBuffer b = (IntBuffer) samples[i];
        if (samples_in[i] instanceof IntPointer
            && samples_in[i].capacity() >= inputSize
            && b.hasArray()) {
          ((IntPointer) samples_in[i]).position(0).put(b.array(), samples[i].position(), inputSize);
        } else {
          samples_in[i] = new IntPointer(b);
        }
      }
    } else if (samples != null && samples[0] instanceof FloatBuffer) {
      inputFormat = samples.length > 1 ? AV_SAMPLE_FMT_FLTP : AV_SAMPLE_FMT_FLT;
      inputDepth = 4;
      for (int i = 0; i < samples.length; i++) {
        FloatBuffer b = (FloatBuffer) samples[i];
        if (samples_in[i] instanceof FloatPointer
            && samples_in[i].capacity() >= inputSize
            && b.hasArray()) {
          ((FloatPointer) samples_in[i]).position(0).put(b.array(), b.position(), inputSize);
        } else {
          samples_in[i] = new FloatPointer(b);
        }
      }
    } else if (samples != null && samples[0] instanceof DoubleBuffer) {
      inputFormat = samples.length > 1 ? AV_SAMPLE_FMT_DBLP : AV_SAMPLE_FMT_DBL;
      inputDepth = 8;
      for (int i = 0; i < samples.length; i++) {
        DoubleBuffer b = (DoubleBuffer) samples[i];
        if (samples_in[i] instanceof DoublePointer
            && samples_in[i].capacity() >= inputSize
            && b.hasArray()) {
          ((DoublePointer) samples_in[i]).position(0).put(b.array(), b.position(), inputSize);
        } else {
          samples_in[i] = new DoublePointer(b);
        }
      }
    } else if (samples != null) {
      throw new Exception("Audio samples Buffer has unsupported type: " + samples);
    }

    if (samples_convert_ctx == null
        || samples_channels != audioChannels
        || samples_format != inputFormat
        || samples_rate != sampleRate) {
      samples_convert_ctx =
          swr_alloc_set_opts(
              samples_convert_ctx,
              audio_c.channel_layout(),
              outputFormat,
              audio_c.sample_rate(),
              av_get_default_channel_layout(audioChannels),
              inputFormat,
              sampleRate,
              0,
              null);
      if (samples_convert_ctx == null) {
        throw new Exception("swr_alloc_set_opts() error: Cannot allocate the conversion context.");
      } else if ((ret = swr_init(samples_convert_ctx)) < 0) {
        throw new Exception(
            "swr_init() error " + ret + ": Cannot initialize the conversion context.");
      }
      samples_channels = audioChannels;
      samples_format = inputFormat;
      samples_rate = sampleRate;
    }

    for (int i = 0; samples != null && i < samples.length; i++) {
      samples_in[i]
          .position(samples_in[i].position() * inputDepth)
          .limit((samples_in[i].position() + inputSize) * inputDepth);
    }
    while (true) {
      int inputCount =
          samples != null
              ? (samples_in[0].limit() - samples_in[0].position()) / (inputChannels * inputDepth)
              : 0;
      int outputCount =
          (samples_out[0].limit() - samples_out[0].position()) / (outputChannels * outputDepth);
      inputCount =
          Math.min(
              inputCount,
              (outputCount * sampleRate + audio_c.sample_rate() - 1) / audio_c.sample_rate());
      for (int i = 0; samples != null && i < samples.length; i++) {
        samples_in_ptr.put(i, samples_in[i]);
      }
      for (int i = 0; i < samples_out.length; i++) {
        samples_out_ptr.put(i, samples_out[i]);
      }
      if ((ret =
              swr_convert(
                  samples_convert_ctx, samples_out_ptr, outputCount, samples_in_ptr, inputCount))
          < 0) {
        throw new Exception("swr_convert() error " + ret + ": Cannot convert audio samples.");
      } else if (ret == 0) {
        break;
      }
      for (int i = 0; samples != null && i < samples.length; i++) {
        samples_in[i].position(samples_in[i].position() + inputCount * inputChannels * inputDepth);
      }
      for (int i = 0; i < samples_out.length; i++) {
        samples_out[i].position(samples_out[i].position() + ret * outputChannels * outputDepth);
      }

      if (samples == null || samples_out[0].position() >= samples_out[0].limit()) {
        frame.nb_samples(audio_input_frame_size);
        avcodec_fill_audio_frame(
            frame, audio_c.channels(), outputFormat, samples_out[0], samples_out[0].limit(), 0);
        for (int i = 0; i < samples_out.length; i++) {
          frame.data(i, samples_out[i].position(0));
          frame.linesize(i, samples_out[i].limit());
        }
        frame.quality(audio_c.global_quality());
        record(frame);
      }
    }
    return samples != null ? frame.key_frame() != 0 : record((AVFrame) null);
  }
Exemplo n.º 2
0
  void startUnsafe() throws Exception {
    int ret;
    picture = null;
    tmp_picture = null;
    picture_buf = null;
    frame = null;
    video_outbuf = null;
    audio_outbuf = null;
    oc = null;
    video_c = null;
    audio_c = null;
    video_st = null;
    audio_st = null;
    got_video_packet = new int[1];
    got_audio_packet = new int[1];

    /* auto detect the output format from the name. */
    String format_name = format == null || format.length() == 0 ? null : format;
    if ((oformat = av_guess_format(format_name, filename, null)) == null) {
      int proto = filename.indexOf("://");
      if (proto > 0) {
        format_name = filename.substring(0, proto);
      }
      if ((oformat = av_guess_format(format_name, filename, null)) == null) {
        throw new Exception(
            "av_guess_format() error: Could not guess output format for \""
                + filename
                + "\" and "
                + format
                + " format.");
      }
    }
    format_name = oformat.name().getString();

    /* allocate the output media context */
    if ((oc = avformat_alloc_context()) == null) {
      throw new Exception("avformat_alloc_context() error: Could not allocate format context");
    }

    oc.oformat(oformat);
    oc.filename().putString(filename);

    /* add the audio and video streams using the format codecs
    and initialize the codecs */

    if (imageWidth > 0 && imageHeight > 0) {
      if (videoCodec != AV_CODEC_ID_NONE) {
        oformat.video_codec(videoCodec);
      } else if ("flv".equals(format_name)) {
        oformat.video_codec(AV_CODEC_ID_FLV1);
      } else if ("mp4".equals(format_name)) {
        oformat.video_codec(AV_CODEC_ID_MPEG4);
      } else if ("3gp".equals(format_name)) {
        oformat.video_codec(AV_CODEC_ID_H263);
      } else if ("avi".equals(format_name)) {
        oformat.video_codec(AV_CODEC_ID_HUFFYUV);
      }

      /* find the video encoder */
      if ((video_codec = avcodec_find_encoder_by_name(videoCodecName)) == null
          && (video_codec = avcodec_find_encoder(oformat.video_codec())) == null) {
        release();
        throw new Exception("avcodec_find_encoder() error: Video codec not found.");
      }
      oformat.video_codec(video_codec.id());

      AVRational frame_rate = av_d2q(frameRate, 1001000);
      AVRational supported_framerates = video_codec.supported_framerates();
      if (supported_framerates != null) {
        int idx = av_find_nearest_q_idx(frame_rate, supported_framerates);
        frame_rate = supported_framerates.position(idx);
      }

      /* add a video output stream */
      if ((video_st = avformat_new_stream(oc, video_codec)) == null) {
        release();
        throw new Exception("avformat_new_stream() error: Could not allocate video stream.");
      }
      video_c = video_st.codec();
      video_c.codec_id(oformat.video_codec());
      video_c.codec_type(AVMEDIA_TYPE_VIDEO);

      /* put sample parameters */
      video_c.bit_rate(videoBitrate);
      /* resolution must be a multiple of two, but round up to 16 as often required */
      video_c.width((imageWidth + 15) / 16 * 16);
      video_c.height(imageHeight);
      if (aspectRatio > 0) {
        AVRational r = av_d2q(aspectRatio, 255);
        video_c.sample_aspect_ratio(r);
        video_st.sample_aspect_ratio(r);
      }
      /* time base: this is the fundamental unit of time (in seconds) in terms
      of which frame timestamps are represented. for fixed-fps content,
      timebase should be 1/framerate and timestamp increments should be
      identically 1. */
      video_c.time_base(av_inv_q(frame_rate));
      video_st.time_base(av_inv_q(frame_rate));
      if (gopSize >= 0) {
        video_c.gop_size(gopSize); /* emit one intra frame every gopSize frames at most */
      }
      if (videoQuality >= 0) {
        video_c.flags(video_c.flags() | CODEC_FLAG_QSCALE);
        video_c.global_quality((int) Math.round(FF_QP2LAMBDA * videoQuality));
      }

      if (pixelFormat != AV_PIX_FMT_NONE) {
        video_c.pix_fmt(pixelFormat);
      } else if (video_c.codec_id() == AV_CODEC_ID_RAWVIDEO
          || video_c.codec_id() == AV_CODEC_ID_PNG
          || video_c.codec_id() == AV_CODEC_ID_HUFFYUV
          || video_c.codec_id() == AV_CODEC_ID_FFV1) {
        video_c.pix_fmt(AV_PIX_FMT_RGB32); // appropriate for common lossless formats
      } else {
        video_c.pix_fmt(AV_PIX_FMT_YUV420P); // lossy, but works with about everything
      }

      if (video_c.codec_id() == AV_CODEC_ID_MPEG2VIDEO) {
        /* just for testing, we also add B frames */
        video_c.max_b_frames(2);
      } else if (video_c.codec_id() == AV_CODEC_ID_MPEG1VIDEO) {
        /* Needed to avoid using macroblocks in which some coeffs overflow.
        This does not happen with normal video, it just happens here as
        the motion of the chroma plane does not match the luma plane. */
        video_c.mb_decision(2);
      } else if (video_c.codec_id() == AV_CODEC_ID_H263) {
        // H.263 does not support any other resolution than the following
        if (imageWidth <= 128 && imageHeight <= 96) {
          video_c.width(128).height(96);
        } else if (imageWidth <= 176 && imageHeight <= 144) {
          video_c.width(176).height(144);
        } else if (imageWidth <= 352 && imageHeight <= 288) {
          video_c.width(352).height(288);
        } else if (imageWidth <= 704 && imageHeight <= 576) {
          video_c.width(704).height(576);
        } else {
          video_c.width(1408).height(1152);
        }
      } else if (video_c.codec_id() == AV_CODEC_ID_H264) {
        // default to constrained baseline to produce content that plays back on anything,
        // without any significant tradeoffs for most use cases
        video_c.profile(AVCodecContext.FF_PROFILE_H264_CONSTRAINED_BASELINE);
      }

      // some formats want stream headers to be separate
      if ((oformat.flags() & AVFMT_GLOBALHEADER) != 0) {
        video_c.flags(video_c.flags() | CODEC_FLAG_GLOBAL_HEADER);
      }

      if ((video_codec.capabilities() & CODEC_CAP_EXPERIMENTAL) != 0) {
        video_c.strict_std_compliance(AVCodecContext.FF_COMPLIANCE_EXPERIMENTAL);
      }
    }

    /*
     * add an audio output stream
     */
    if (audioChannels > 0 && audioBitrate > 0 && sampleRate > 0) {
      if (audioCodec != AV_CODEC_ID_NONE) {
        oformat.audio_codec(audioCodec);
      } else if ("flv".equals(format_name)
          || "mp4".equals(format_name)
          || "3gp".equals(format_name)) {
        oformat.audio_codec(AV_CODEC_ID_AAC);
      } else if ("avi".equals(format_name)) {
        oformat.audio_codec(AV_CODEC_ID_PCM_S16LE);
      }

      /* find the audio encoder */
      if ((audio_codec = avcodec_find_encoder_by_name(audioCodecName)) == null
          && (audio_codec = avcodec_find_encoder(oformat.audio_codec())) == null) {
        release();
        throw new Exception("avcodec_find_encoder() error: Audio codec not found.");
      }

      if ((audio_st = avformat_new_stream(oc, audio_codec)) == null) {
        release();
        throw new Exception("avformat_new_stream() error: Could not allocate audio stream.");
      }
      audio_c = audio_st.codec();
      audio_c.codec_id(oformat.audio_codec());
      audio_c.codec_type(AVMEDIA_TYPE_AUDIO);

      /* put sample parameters */
      audio_c.bit_rate(audioBitrate);
      audio_c.sample_rate(sampleRate);
      audio_c.channels(audioChannels);
      audio_c.channel_layout(av_get_default_channel_layout(audioChannels));
      if (sampleFormat != AV_SAMPLE_FMT_NONE) {
        audio_c.sample_fmt(sampleFormat);
      } else {
        // use AV_SAMPLE_FMT_S16 by default, if available
        audio_c.sample_fmt(AV_SAMPLE_FMT_FLTP);
        IntPointer formats = audio_c.codec().sample_fmts();
        for (int i = 0; formats.get(i) != -1; i++) {
          if (formats.get(i) == AV_SAMPLE_FMT_S16) {
            audio_c.sample_fmt(AV_SAMPLE_FMT_S16);
            break;
          }
        }
      }
      audio_c.time_base().num(1).den(sampleRate);
      audio_st.time_base().num(1).den(sampleRate);
      switch (audio_c.sample_fmt()) {
        case AV_SAMPLE_FMT_U8:
        case AV_SAMPLE_FMT_U8P:
          audio_c.bits_per_raw_sample(8);
          break;
        case AV_SAMPLE_FMT_S16:
        case AV_SAMPLE_FMT_S16P:
          audio_c.bits_per_raw_sample(16);
          break;
        case AV_SAMPLE_FMT_S32:
        case AV_SAMPLE_FMT_S32P:
          audio_c.bits_per_raw_sample(32);
          break;
        case AV_SAMPLE_FMT_FLT:
        case AV_SAMPLE_FMT_FLTP:
          audio_c.bits_per_raw_sample(32);
          break;
        case AV_SAMPLE_FMT_DBL:
        case AV_SAMPLE_FMT_DBLP:
          audio_c.bits_per_raw_sample(64);
          break;
        default:
          assert false;
      }
      if (audioQuality >= 0) {
        audio_c.flags(audio_c.flags() | CODEC_FLAG_QSCALE);
        audio_c.global_quality((int) Math.round(FF_QP2LAMBDA * audioQuality));
      }

      // some formats want stream headers to be separate
      if ((oformat.flags() & AVFMT_GLOBALHEADER) != 0) {
        audio_c.flags(audio_c.flags() | CODEC_FLAG_GLOBAL_HEADER);
      }

      if ((audio_codec.capabilities() & CODEC_CAP_EXPERIMENTAL) != 0) {
        audio_c.strict_std_compliance(AVCodecContext.FF_COMPLIANCE_EXPERIMENTAL);
      }
    }

    av_dump_format(oc, 0, filename, 1);

    /* now that all the parameters are set, we can open the audio and
    video codecs and allocate the necessary encode buffers */
    if (video_st != null) {
      AVDictionary options = new AVDictionary(null);
      if (videoQuality >= 0) {
        av_dict_set(options, "crf", "" + videoQuality, 0);
      }
      for (Entry<String, String> e : videoOptions.entrySet()) {
        av_dict_set(options, e.getKey(), e.getValue(), 0);
      }
      /* open the codec */
      if ((ret = avcodec_open2(video_c, video_codec, options)) < 0) {
        release();
        throw new Exception("avcodec_open2() error " + ret + ": Could not open video codec.");
      }
      av_dict_free(options);

      video_outbuf = null;
      if ((oformat.flags() & AVFMT_RAWPICTURE) == 0) {
        /* allocate output buffer */
        /* XXX: API change will be done */
        /* buffers passed into lav* can be allocated any way you prefer,
        as long as they're aligned enough for the architecture, and
        they're freed appropriately (such as using av_free for buffers
        allocated with av_malloc) */
        video_outbuf_size =
            Math.max(256 * 1024, 8 * video_c.width() * video_c.height()); // a la ffmpeg.c
        video_outbuf = new BytePointer(av_malloc(video_outbuf_size));
      }

      /* allocate the encoded raw picture */
      if ((picture = av_frame_alloc()) == null) {
        release();
        throw new Exception("av_frame_alloc() error: Could not allocate picture.");
      }
      picture.pts(0); // magic required by libx264

      int size = avpicture_get_size(video_c.pix_fmt(), video_c.width(), video_c.height());
      if ((picture_buf = new BytePointer(av_malloc(size))).isNull()) {
        release();
        throw new Exception("av_malloc() error: Could not allocate picture buffer.");
      }

      /* if the output format is not equal to the image format, then a temporary
      picture is needed too. It is then converted to the required output format */
      if ((tmp_picture = av_frame_alloc()) == null) {
        release();
        throw new Exception("av_frame_alloc() error: Could not allocate temporary picture.");
      }

      AVDictionary metadata = new AVDictionary(null);
      for (Entry<String, String> e : videoMetadata.entrySet()) {
        av_dict_set(metadata, e.getKey(), e.getValue(), 0);
      }
      video_st.metadata(metadata);
    }

    if (audio_st != null) {
      AVDictionary options = new AVDictionary(null);
      if (audioQuality >= 0) {
        av_dict_set(options, "crf", "" + audioQuality, 0);
      }
      for (Entry<String, String> e : audioOptions.entrySet()) {
        av_dict_set(options, e.getKey(), e.getValue(), 0);
      }
      /* open the codec */
      if ((ret = avcodec_open2(audio_c, audio_codec, options)) < 0) {
        release();
        throw new Exception("avcodec_open2() error " + ret + ": Could not open audio codec.");
      }
      av_dict_free(options);

      audio_outbuf_size = 256 * 1024;
      audio_outbuf = new BytePointer(av_malloc(audio_outbuf_size));

      /* ugly hack for PCM codecs (will be removed ASAP with new PCM
      support to compute the input frame size in samples */
      if (audio_c.frame_size() <= 1) {
        audio_outbuf_size = FF_MIN_BUFFER_SIZE;
        audio_input_frame_size = audio_outbuf_size / audio_c.channels();
        switch (audio_c.codec_id()) {
          case AV_CODEC_ID_PCM_S16LE:
          case AV_CODEC_ID_PCM_S16BE:
          case AV_CODEC_ID_PCM_U16LE:
          case AV_CODEC_ID_PCM_U16BE:
            audio_input_frame_size >>= 1;
            break;
          default:
            break;
        }
      } else {
        audio_input_frame_size = audio_c.frame_size();
      }
      // int bufferSize = audio_input_frame_size * audio_c.bits_per_raw_sample()/8 *
      // audio_c.channels();
      int planes =
          av_sample_fmt_is_planar(audio_c.sample_fmt()) != 0 ? (int) audio_c.channels() : 1;
      int data_size =
          av_samples_get_buffer_size(
                  (IntPointer) null,
                  audio_c.channels(),
                  audio_input_frame_size,
                  audio_c.sample_fmt(),
                  1)
              / planes;
      samples_out = new BytePointer[planes];
      for (int i = 0; i < samples_out.length; i++) {
        samples_out[i] = new BytePointer(av_malloc(data_size)).capacity(data_size);
      }
      samples_in = new Pointer[AVFrame.AV_NUM_DATA_POINTERS];
      samples_in_ptr = new PointerPointer(AVFrame.AV_NUM_DATA_POINTERS);
      samples_out_ptr = new PointerPointer(AVFrame.AV_NUM_DATA_POINTERS);

      /* allocate the audio frame */
      if ((frame = av_frame_alloc()) == null) {
        release();
        throw new Exception("av_frame_alloc() error: Could not allocate audio frame.");
      }
      frame.pts(0); // magic required by libvorbis and webm

      AVDictionary metadata = new AVDictionary(null);
      for (Entry<String, String> e : audioMetadata.entrySet()) {
        av_dict_set(metadata, e.getKey(), e.getValue(), 0);
      }
      audio_st.metadata(metadata);
    }

    /* open the output file, if needed */
    if ((oformat.flags() & AVFMT_NOFILE) == 0) {
      AVIOContext pb = new AVIOContext(null);
      if ((ret = avio_open(pb, filename, AVIO_FLAG_WRITE)) < 0) {
        release();
        throw new Exception(
            "avio_open error() error " + ret + ": Could not open '" + filename + "'");
      }
      oc.pb(pb);
    }

    AVDictionary options = new AVDictionary(null);
    for (Entry<String, String> e : this.options.entrySet()) {
      av_dict_set(options, e.getKey(), e.getValue(), 0);
    }
    AVDictionary metadata = new AVDictionary(null);
    for (Entry<String, String> e : this.metadata.entrySet()) {
      av_dict_set(metadata, e.getKey(), e.getValue(), 0);
    }
    /* write the stream header, if any */
    avformat_write_header(oc.metadata(metadata), options);
    av_dict_free(options);
  }