#include "util.h"
#include "log.h"
#include "ffmpeg_decoder.h"
+#include "subtitle.h"
using namespace std;
using namespace boost;
, _audio_codec (0)
, _subtitle_codec_context (0)
, _subtitle_codec (0)
- , _have_subtitle (false)
{
setup_general ();
setup_video ();
avcodec_close (_video_codec_context);
}
- if (_have_subtitle) {
- avsubtitle_free (&_subtitle);
- }
-
if (_subtitle_codec_context) {
avcodec_close (_subtitle_codec_context);
}
if (_format_context->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO) {
_video_stream = i;
} else if (_format_context->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
- _audio_stream = i;
+ if (_audio_stream == -1) {
+ _audio_stream = i;
+ }
+ _audio_streams.push_back (Stream (stream_name (_format_context->streams[i]), i));
} else if (_format_context->streams[i]->codec->codec_type == AVMEDIA_TYPE_SUBTITLE) {
- _subtitle_stream = i;
+ if (_subtitle_stream == -1) {
+ _subtitle_stream = i;
+ }
+ _subtitle_streams.push_back (Stream (stream_name (_format_context->streams[i]), i));
}
}
int frame_finished;
- if (_opt->decode_video) {
- while (avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet) >= 0 && frame_finished) {
- process_video (_frame);
- }
+ while (avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet) >= 0 && frame_finished) {
+ process_video (_frame);
}
if (_audio_stream >= 0 && _opt->decode_audio) {
0, _audio_codec_context->channels, _frame->nb_samples, audio_sample_format (), 1
);
- assert (_audio_codec_context->channels == _fs->audio_channels);
+ assert (_audio_codec_context->channels == _fs->audio_channels());
process_audio (_frame->data[0], data_size);
}
}
return true;
}
- if (_packet.stream_index == _video_stream && _opt->decode_video) {
+ if (_packet.stream_index == _video_stream) {
int frame_finished;
if (avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet) >= 0 && frame_finished) {
0, _audio_codec_context->channels, _frame->nb_samples, audio_sample_format (), 1
);
- assert (_audio_codec_context->channels == _fs->audio_channels);
+ assert (_audio_codec_context->channels == _fs->audio_channels());
process_audio (_frame->data[0], data_size);
}
- } else if (_subtitle_stream >= 0 && _packet.stream_index == _subtitle_stream) {
-
- if (_have_subtitle) {
- avsubtitle_free (&_subtitle);
- _have_subtitle = false;
- }
+ } else if (_subtitle_stream >= 0 && _packet.stream_index == _subtitle_stream && _opt->decode_subtitles) {
int got_subtitle;
- if (avcodec_decode_subtitle2 (_subtitle_codec_context, &_subtitle, &got_subtitle, &_packet) && got_subtitle) {
- _have_subtitle = true;
+ AVSubtitle sub;
+ if (avcodec_decode_subtitle2 (_subtitle_codec_context, &sub, &got_subtitle, &_packet) && got_subtitle) {
+ process_subtitle (shared_ptr<TimedSubtitle> (new TimedSubtitle (sub)));
+ avsubtitle_free (&sub);
}
}
float
FFmpegDecoder::frames_per_second () const
{
- return av_q2d (_format_context->streams[_video_stream]->avg_frame_rate);
+ AVStream* s = _format_context->streams[_video_stream];
+
+ if (s->avg_frame_rate.num && s->avg_frame_rate.den) {
+ return av_q2d (s->avg_frame_rate);
+ }
+
+ return av_q2d (s->r_frame_rate);
}
int
return _video_codec_context->sample_aspect_ratio.den;
}
-void
-FFmpegDecoder::overlay (shared_ptr<Image> image) const
+bool
+FFmpegDecoder::has_subtitles () const
{
- if (!_have_subtitle) {
- return;
- }
-
- /* subtitle PTS in seconds */
- float const packet_time = (_subtitle.pts / AV_TIME_BASE) + float (_subtitle.pts % AV_TIME_BASE) / 1e6;
- /* hence start time for this sub */
- float const from = packet_time + (float (_subtitle.start_display_time) / 1e3);
- float const to = packet_time + (float (_subtitle.end_display_time) / 1e3);
-
- float const video_frame_time = float (last_video_frame ()) / rint (_fs->frames_per_second);
+ return (_subtitle_stream != -1);
+}
- if (from > video_frame_time || video_frame_time < to) {
- return;
- }
+vector<Stream>
+FFmpegDecoder::audio_streams () const
+{
+ return _audio_streams;
+}
- for (unsigned int i = 0; i < _subtitle.num_rects; ++i) {
- AVSubtitleRect* rect = _subtitle.rects[i];
- if (rect->type != SUBTITLE_BITMAP) {
- throw DecodeError ("non-bitmap subtitles not yet supported");
- }
+vector<Stream>
+FFmpegDecoder::subtitle_streams () const
+{
+ return _subtitle_streams;
+}
- /* XXX: all this assumes YUV420 in image */
-
- assert (rect->pict.data[0]);
+void
+FFmpegDecoder::set_audio_stream (int s)
+{
+ _audio_stream = s;
+ setup_audio ();
+}
- /* Start of the first line in the target image */
- uint8_t* frame_y_p = image->data()[0] + rect->y * image->line_size()[0];
- uint8_t* frame_u_p = image->data()[1] + (rect->y / 2) * image->line_size()[1];
- uint8_t* frame_v_p = image->data()[2] + (rect->y / 2) * image->line_size()[2];
+void
+FFmpegDecoder::set_subtitle_stream (int s)
+{
+ _subtitle_stream = s;
+ setup_subtitle ();
+}
- int const hlim = min (rect->y + rect->h, image->size().height) - rect->y;
-
- /* Start of the first line in the subtitle */
- uint8_t* sub_p = rect->pict.data[0];
- /* sub_p looks up into a RGB palette which is here */
- uint32_t const * palette = (uint32_t *) rect->pict.data[1];
-
- for (int sub_y = 0; sub_y < hlim; ++sub_y) {
- /* Pointers to the start of this line */
- uint8_t* sub_line_p = sub_p;
- uint8_t* frame_line_y_p = frame_y_p + rect->x;
- uint8_t* frame_line_u_p = frame_u_p + (rect->x / 2);
- uint8_t* frame_line_v_p = frame_v_p + (rect->x / 2);
-
- /* U and V are subsampled */
- uint8_t next_u = 0;
- uint8_t next_v = 0;
- int subsample_step = 0;
-
- for (int sub_x = 0; sub_x < rect->w; ++sub_x) {
-
- /* RGB value for this subtitle pixel */
- uint32_t const val = palette[*sub_line_p++];
-
- int const red = (val & 0xff);
- int const green = (val & 0xff00) >> 8;
- int const blue = (val & 0xff0000) >> 16;
- float const alpha = ((val & 0xff000000) >> 24) / 255.0;
-
- /* Alpha-blend Y */
- int const cy = *frame_line_y_p;
- *frame_line_y_p++ = int (cy * (1 - alpha)) + int (RGB_TO_Y_CCIR (red, green, blue) * alpha);
-
- /* Store up U and V */
- next_u |= ((RGB_TO_U_CCIR (red, green, blue, 0) & 0xf0) >> 4) << (4 * subsample_step);
- next_v |= ((RGB_TO_V_CCIR (red, green, blue, 0) & 0xf0) >> 4) << (4 * subsample_step);
-
- if (subsample_step == 1 && (sub_y % 2) == 0) {
- int const cu = *frame_line_u_p;
- int const cv = *frame_line_v_p;
-
- *frame_line_u_p++ =
- int (((cu & 0x0f) * (1 - alpha) + (next_u & 0x0f) * alpha)) |
- int (((cu & 0xf0) * (1 - alpha) + (next_u & 0xf0) * alpha));
-
- *frame_line_v_p++ =
- int (((cv & 0x0f) * (1 - alpha) + (next_v & 0x0f) * alpha)) |
- int (((cv & 0xf0) * (1 - alpha) + (next_v & 0xf0) * alpha));
-
- next_u = next_v = 0;
- }
-
- subsample_step = (subsample_step + 1) % 2;
- }
-
- sub_p += rect->pict.linesize[0];
- frame_y_p += image->line_size()[0];
- if ((sub_y % 2) == 0) {
- frame_u_p += image->line_size()[1];
- frame_v_p += image->line_size()[2];
- }
+string
+FFmpegDecoder::stream_name (AVStream* s) const
+{
+ stringstream n;
+
+ AVDictionaryEntry const * lang = av_dict_get (s->metadata, "language", 0, 0);
+ if (lang) {
+ n << lang->value;
+ }
+
+ AVDictionaryEntry const * title = av_dict_get (s->metadata, "title", 0, 0);
+ if (title) {
+ if (!n.str().empty()) {
+ n << " ";
}
+ n << title->value;
}
+
+ if (n.str().empty()) {
+ n << "unknown";
+ }
+
+ return n.str ();
}
-