Example #1
0
//struct SwsContext * sws_getCachedContext (struct SwsContext *context, int srcW, int srcH, enum AVPixelFormat srcFormat, int dstW, int dstH, enum AVPixelFormat dstFormat, int flags, SwsFilter *srcFilter, SwsFilter *dstFilter, const double *param)
//Check if context can be reused, otherwise reallocate a new one.
func Sws_getCachedContext(ctxt *SwsContext, sw, sh int, sf AVPixelFormat, dw, dh int, df AVPixelFormat, f int, sfl, dfl *SwsFilter, p *float64) *SwsContext {
	return (*SwsContext)(C.sws_getCachedContext((*C.struct_SwsContext)(ctxt), C.int(sw), C.int(sh), (C.enum_AVPixelFormat)(sf), C.int(dw), C.int(dh), (C.enum_AVPixelFormat)(df), C.int(f), (*C.struct_SwsFilter)(sfl), (*C.struct_SwsFilter)(dfl), (*C.double)(p)))
}
Example #2
0
//struct SwsContext * sws_getCachedContext (struct SwsContext *context, int srcW, int srcH, enum AVPixelFormat srcFormat, int dstW, int dstH, enum AVPixelFormat dstFormat, int flags, SwsFilter *srcFilter, SwsFilter *dstFilter, const double *param)
//Check if context can be reused, otherwise reallocate a new one.
func Sws_getCachedContext(ctxt *C.struct_SwsContext, sw, sh int, sf C.enum_AVPixelFormat, dw, dh int, df C.enum_AVPixelFormat, f int, sfl, dfl *C.struct_SwsFilter, p *C.double) *C.struct_SwsContext {
	return C.sws_getCachedContext(ctxt, C.int(sw), C.int(sh), sf, C.int(dw), C.int(dh), df, C.int(f), sfl, dfl, p)
}
Example #3
0
func (self *Decoder) Start(videoStream, audioStream *C.AVStream,
	scaleWidth, scaleHeight C.int) *Decoder {

	self.running = true
	self.duration = time.Duration(self.FormatContext.duration * C.AV_TIME_BASE / 1000)
	vCodecCtx := videoStream.codec
	aCodecCtx := audioStream.codec

	self.durationPerSample = time.Second / time.Duration(aCodecCtx.sample_rate)

	// frame pool
	poolSize := 16
	pool := make(chan *C.AVFrame, poolSize)
	self.pool = pool
	numBytes := C.size_t(C.avpicture_get_size(C.PIX_FMT_YUV420P, scaleWidth, scaleHeight))
	for i := 0; i < poolSize; i++ {
		frame := C.av_frame_alloc()
		self.frames = append(self.frames, frame)
		buffer := (*C.uint8_t)(unsafe.Pointer(C.av_malloc(numBytes)))
		self.buffers = append(self.buffers, buffer)
		C.avpicture_fill((*C.AVPicture)(unsafe.Pointer(frame)), buffer, C.PIX_FMT_YUV420P,
			scaleWidth, scaleHeight)
		pool <- frame
	}

	// decode
	self.frameChan = make(chan *C.AVFrame, 512)
	go func() {
		runtime.LockOSThread()

		// scale context
		scaleContext := C.sws_getCachedContext(nil, vCodecCtx.width, vCodecCtx.height, vCodecCtx.pix_fmt,
			scaleWidth, scaleHeight, C.PIX_FMT_YUV420P, C.SWS_LANCZOS, nil, nil, nil)
		if scaleContext == nil {
			log.Fatal("get scale context failed")
		}

		// resample context
		resampleContext := C.swr_alloc_set_opts(nil,
			C.AV_CH_LAYOUT_STEREO, C.AV_SAMPLE_FMT_FLT, aCodecCtx.sample_rate,
			C.int64_t(aCodecCtx.channel_layout), aCodecCtx.sample_fmt, aCodecCtx.sample_rate,
			0, nil)
		if resampleContext == nil {
			log.Fatal("get resample context failed")
		}
		C.swr_init(resampleContext)

		var packet C.AVPacket
		var frameFinished C.int
		var pts int64
		var packetTime time.Duration
		vFrame := C.av_frame_alloc()
		aFrame := C.av_frame_alloc()
		videoIndex := videoStream.index
		audioIndex := audioStream.index
		resampleBuffer := (*C.uint8_t)(C.av_malloc(4096 * 8))
		resampleBufferp := &resampleBuffer

		self.Timer = NewTimer()

		// decode
		for self.running {

			// seek
			if self.seekTarget > 0 {
				if C.av_seek_frame(self.FormatContext, -1,
					C.int64_t(float64(self.seekNext)/float64(time.Second)*float64(C.AV_TIME_BASE)),
					C.AVSEEK_FLAG_BACKWARD) < 0 {
					log.Fatal("seek error")
				}
				for _, codecCtx := range self.openedCodecs {
					C.avcodec_flush_buffers(codecCtx)
				}
				p("frame seek done\n")
			}

		read_packet:
			// read packet
			C.av_free_packet(&packet)
			if C.av_read_frame(self.FormatContext, &packet) < 0 { // read packet
				log.Fatal("read frame error") //TODO stop gracefully
			}

			// get packet time
			if packet.dts != C.AV_NOPTS_VALUE {
				pts = int64(packet.dts)
			} else {
				pts = 0
			}
			if packet.stream_index == videoIndex {
				packetTime = time.Duration(float64(pts) * float64(C.av_q2d(videoStream.time_base)) * float64(time.Second))
			} else if packet.stream_index == audioIndex {
				packetTime = time.Duration(float64(pts) * float64(C.av_q2d(audioStream.time_base)) * float64(time.Second))
			} else { // ignore packet
				goto read_packet
			}
			p("packet time %v at timer time %v\n", packetTime, self.Timer.Now())

			// check seek
			if self.seekTarget > 0 && packetTime > 0 { // if packet time cannot determined, skip
				if packetTime < self.seekTarget { // seek again
					self.seekNext += self.seekStep
					p("seek again %v\n", self.seekNext)
				} else { // seek ok
					p("seek ok\n")
					self.seekTarget = 0
					self.Timer.Jump(packetTime - self.Timer.Now())
				}
			}

			// decode
			if packet.stream_index == videoIndex { // decode video
				if C.avcodec_decode_video2(vCodecCtx, vFrame, &frameFinished, &packet) < 0 {
					continue // bad packet
				}
				if frameFinished <= 0 {
					goto read_packet // frame not complete
				}
				bufFrame := <-pool        // get frame buffer
				C.sws_scale(scaleContext, // scale
					&vFrame.data[0], &vFrame.linesize[0], 0, vCodecCtx.height,
					&bufFrame.data[0], &bufFrame.linesize[0])
				bufFrame.pts = C.int64_t(packetTime) // set packet time
				self.frameChan <- bufFrame           // push to queue
				p("video frame %v\n", packetTime)

			} else if packet.stream_index == audioIndex { // decode audio
			decode_audio_packet:
				l := C.avcodec_decode_audio4(aCodecCtx, aFrame, &frameFinished, &packet)
				if l < 0 {
					continue // bad packet
				}
				if frameFinished <= 0 {
					goto read_packet // frame not complete
				}
				if frameFinished > 0 { // frame finished
					n := C.swr_convert(resampleContext, resampleBufferp, 4096,
						aFrame.extended_data, aFrame.nb_samples)
					if n != aFrame.nb_samples {
						log.Fatal("audio resample failed")
					}
					self.audioFrames <- &AudioFrame{
						time: packetTime,
						data: C.GoBytes(unsafe.Pointer(resampleBuffer), n*8),
					}
				}
				if l != packet.size { // multiple frame packet
					packet.size -= l
					packet.data = (*C.uint8_t)(unsafe.Pointer(uintptr(unsafe.Pointer(packet.data)) + uintptr(l)))
					goto decode_audio_packet
				}
				p("audio frame %v\n", packetTime)

			} else { // other stream
				goto read_packet
			}

		}
	}()

	// sync video
	go func() {
		maxDelta := (time.Second / time.Duration(C.av_q2d(videoStream.r_frame_rate)/10))
		for frame := range self.frameChan {
			delta := time.Duration(frame.pts) - self.Timer.Now()
			p("video frame %v, delta %v, max delta %v\n", time.Duration(frame.pts), delta, maxDelta)
			if delta > 0 {
				if delta > maxDelta {
					self.Timer.Jump(delta)
					print("timer jumped\n")
				} else {
					time.Sleep(delta)
				}
			} else if delta < 0 { // drop frame
				self.RecycleFrame(frame)
				continue
			}
			self.timedFrames <- frame
		}
	}()

	return self
}