ffmpeg学习---4.Playing Sound

来源：互联网发布：变频器模拟软件编辑：程序博客网时间：2024/06/14 07:46

一.最原始版的
http://dranger.com/ffmpeg/tutorial03.html
1. 3playsound.c

cong@msi:/work/ffmpeg/test/4sound$ cat sound.c 
#include "utils.h"
#include <libavformat/avformat.h>
#include <libswscale/swscale.h>
#include <SDL/SDL.h>

#define SDL_AUDIO_BUFFER_SIZE 4096
#define AVCODEC_MAX_AUDIO_FRAME_SIZE 192000
SDL_mutex *affmutex;
SDL_Event sdlevent;
int signal_quit = 1;
typedef struct PacketQueue
{
    AVPacketList * first_pkt, *last_pkt;
    int nb_packets;
    int size;
    SDL_mutex *mutex;
    SDL_cond * cond;
}PacketQueue;
PacketQueue audioq;

static int eventThread(void* data)
{
    while(signal_quit)
    {
        SDL_LockMutex(affmutex);
        while(SDL_PollEvent(&sdlevent))
        {
            switch(sdlevent.type)
            {
                case SDL_QUIT:
                    {
                        signal_quit = 0;
                    }
                    break;
                default:
                    break;

            }
        }
        SDL_UnlockMutex(affmutex);
    } 
}
void packet_queue_init(PacketQueue *q)
{
    memset(q, 0, sizeof(PacketQueue));
    q->mutex = SDL_CreateMutex();
    q->cond = SDL_CreateCond();
}

int packet_queue_put(PacketQueue *q, AVPacket *pkt)
{

    AVPacketList *pkt1;
    if(av_dup_packet(pkt) < 0)
        return -1;
    pkt1 = av_malloc(sizeof(AVPacketList));
    if (!pkt1)
        return -1;
    pkt1->pkt = *pkt;
    pkt1->next = NULL;

    SDL_LockMutex(q->mutex);

    if (!q->last_pkt)
        q->first_pkt = pkt1;
    else
        q->last_pkt->next = pkt1;
    q->last_pkt = pkt1;
    q->nb_packets++;
    q->size += pkt1->pkt.size;
    SDL_CondSignal(q->cond);

    SDL_UnlockMutex(q->mutex);
    return 0;
}

static int packet_queue_get(PacketQueue *q, AVPacket *pkt, int block)
{
    AVPacketList *pkt1;
    int ret;
    SDL_LockMutex(q->mutex);
    for(;;)
    {
        if(0 == signal_quit)
        {
            ret = -1;
            break;
        }
        pkt1 = q->first_pkt;
        if (pkt1)
        {
            q->first_pkt = pkt1->next;
            if (!q->first_pkt)
                q->last_pkt = NULL;
            q->nb_packets--;
            q->size -= pkt1->pkt.size;
            *pkt = pkt1->pkt;
            av_free(pkt1);
            ret = 1;
            break;
        } else if (!block) {
            ret = 0;
            break;
        } else {
            SDL_CondWait(q->cond, q->mutex);
        }
    }
    SDL_UnlockMutex(q->mutex);
    return ret;
}

int audio_decode_frame(AVCodecContext *aCodecCtx, AVPacket *pkt, AVPacket *pkt_temp, AVFrame *frame, uint8_t *audio_buf)
{
    int i;
    int len1, data_size, got_frame;
    int new_packet;
    for(;;)
    {
        while(pkt_temp->size>0 || (!pkt_temp->data && new_packet))
        {
            if(!frame)
            {
                if (!(frame = av_frame_alloc())) //函数avcodec_alloc_frame不用了
                    return AVERROR(ENOMEM);
            }
            else
            {
                av_frame_unref(frame); //函数avcodec_get_frame_defaults(frame)不用了
            }
            new_packet = 0;
            len1 = avcodec_decode_audio4(aCodecCtx, frame, &got_frame, pkt_temp); //decode data is store in frame
            if(len1 < 0)
            {
                pkt_temp->size = 0;
                break;
            }

            pkt_temp->data += len1;
            pkt_temp->size -= len1;

            if(got_frame <= 0) /* No data yet, get more frames */
                continue;
            //dbmsg("add filter: sample=%d", frame->nb_samples); //这个地方有问题:若解码后的数据包与sdl需要的数据格式不匹配则播放产生噪声
            data_size = av_samples_get_buffer_size(NULL, aCodecCtx->channels, frame->nb_samples, aCodecCtx->sample_fmt, 1);
            memcpy(audio_buf, frame->data[0], frame->linesize[0]);
            return data_size;
        }
        if(pkt->data)
            av_free_packet(pkt);
        memset(pkt_temp, 0, sizeof(*pkt_temp));
        if(0 == signal_quit)
            return -1;
         //这个地方是本函数的开始位置,一直等侍音频数据包的到来
        if((new_packet = packet_queue_get(&audioq, pkt, 1)) < 0)
            return -1;
        *pkt_temp = *pkt;
    }
}

void audio_callback(void *userdata, Uint8 *stream, int len)  //这个回调函数中len=4096
{
    int i;
    AVCodecContext *aCodecCtx = (AVCodecContext *)userdata;
    int len1, audio_size;

    static uint8_t audio_buf[(AVCODEC_MAX_AUDIO_FRAME_SIZE * 3) / 2];
    static unsigned int audio_buf_size = 0;
    static unsigned int audio_buf_index = 0;

    AVPacket *pkt = av_mallocz(sizeof(AVPacket));
    AVPacket *pkt_temp = av_mallocz(sizeof(AVPacket));
    AVFrame *frame = NULL;
    //dbmsg("len=%d", len);
    while(len > 0)
    {
        if(audio_buf_index >= audio_buf_size)
        {   //如果没有音频数据包则等侍,若有数据包则解码，并将解码后的数据放在audio_buf中
            audio_size = audio_decode_frame(aCodecCtx, pkt, pkt_temp, frame, audio_buf);
            if(audio_size < 0)
            {
                /* If error, output silence */
                audio_buf_size = 1024;
                memset(audio_buf, 0, audio_buf_size);
            } else {
                audio_buf_size = audio_size;
            }
            audio_buf_index = 0;
        }
        len1 = audio_buf_size - audio_buf_index;
        if(len1 > len)
            len1 = len;
        //将解码后的数据(audio_buf中的数据)copy到stream中进行播放
        memcpy(stream, (uint8_t *)audio_buf + audio_buf_index, len1);
        len -= len1;
        stream += len1;
        audio_buf_index += len1;
    }
}

int main(int argc, char **argv)
{
    int i=0;
    int ret;
    int videoindex= -1;
    int sndindex= -1;
    int frameFinished;
    AVFormatContext *pFormatCtx = NULL;
    AVCodecContext * pCodecCtx;
    AVCodecContext * sndCodecCtx;
    AVCodec * pCodec;
    AVCodec * sndCodec;
    AVFrame * pFrame;
    AVFrame * pFrameYUV;
    AVPacket * packet; 
    struct SwsContext *img_convert_ctx;

    SDL_Surface* psscreen;
    SDL_Overlay* overlay;
    SDL_Rect rect;
    SDL_Thread* sdl_thread;
    SDL_AudioSpec wanted_spec, spec;
    //a. ffmpeg的初始化(虽然名字是register就这么说吧)
    avcodec_register_all();
    avfilter_register_all();
    av_register_all();
    //b.打开视频文件
    pFormatCtx = avformat_alloc_context();
    if(avformat_open_input(&pFormatCtx, argv[1], NULL, NULL)!=0)
        return -1; 
    //c.获取视频文件的流信息,(即查看有几个视频流几个音频流)  
    if(avformat_find_stream_info(pFormatCtx, NULL)<0)
        return -1;
    av_dump_format(pFormatCtx,0, 0, 0);
    //d.获取了视频流的index,这样以后读取一帧之后,根据索引号才能判断这一帧是不是视频帧
    //同理获取了音频流的index,这样以后读取一包之后,根据索引号才能判断这一帧是不是音频帧
    for(i=0; i<pFormatCtx->nb_streams; i++)
    {
        if(pFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO)
        {
            videoindex= i;   //获取视频流的索引
        }
        if(pFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO)
        {
            sndindex= i;    //获取视频流的索引
        }
    }
    if(videoindex== -1)
    {
        dbmsg("no video stream found!");
        //return -1;
    }
    if(sndindex== -1)
    {
        dbmsg("no sound stream found!");
        return -1;
    }
    //e.1为视频流寻找解码器:在c中不仅有视频流的index,还有视频流的编码codec_id，并打开解码器
    if(videoindex != -1)
    {
        pCodecCtx = pFormatCtx->streams[videoindex]->codec;
        pCodec = avcodec_find_decoder(pCodecCtx->codec_id);
        if(pCodec == NULL)
        {
            dbmsg("Codec not found");
            return -1;
        }
        if(avcodec_open2(pCodecCtx, pCodec, NULL) < 0)
            return -1;
    }
    //e.2为音频流寻找解码器,并打开解码器
    if(sndindex != -1)
    {
        sndCodecCtx = pFormatCtx->streams[sndindex]->codec;
        sndCodec = avcodec_find_decoder(sndCodecCtx->codec_id);
        if(sndCodec == NULL)
        {
            dbmsg("Codec not found");
            return -1;
        }

        if(avcodec_open2(sndCodecCtx, sndCodec, NULL) < 0)
            return -1;
    }
    //f.1视频显示的准备:SDL初始化,设置显示模式,创建画布
    pFrame = av_frame_alloc();           //以前的avcodec_alloc_frame函数现在不用了
    pFrameYUV = av_frame_alloc();        //以前的avcodec_alloc_frame函数现在不用了
    SDL_Init(SDL_INIT_EVERYTHING);
    psscreen = SDL_SetVideoMode(pCodecCtx->width, pCodecCtx->height, 0, SDL_SWSURFACE);
    SDL_WM_SetCaption( "FFMPEG Window", NULL);
    //注意这儿的参数SDL_YU12_OVERLAY与SDL_YUY2_OVERLAY一定要与下面sws_scale中的参数配套
    overlay = SDL_CreateYUVOverlay(pCodecCtx->width, pCodecCtx->height, SDL_YV12_OVERLAY, psscreen);
    //overlay = SDL_CreateYUVOverlay(pCodecCtx->width, pCodecCtx->height, SDL_YUY2_OVERLAY, psscreen);
    
    //f.2音频的准备:设置输出audio的格式
    wanted_spec.freq = sndCodecCtx->sample_rate;           //采样率
    wanted_spec.format = AUDIO_S16SYS;                     //SDL中的输入格式, s代表signed有符号的，16代表采样是16位的采样精度，SYS代表大小端格式按照系统默认 
    wanted_spec.channels = sndCodecCtx->channels;          //声道数
    wanted_spec.silence = 0;                               //需不需要静音: 0不需要，但是设为之后仍旧是有声音的,看来这个理解有误,有明白的告诉一下?????
    wanted_spec.samples = 2048;                            //audio_buffer的size,通常这个值设为512到8192,ffplay设置的是1024
    wanted_spec.callback = audio_callback;                 //回调函数
    wanted_spec.userdata = sndCodecCtx;                    //给回调函数传的参数
    if(SDL_OpenAudio(&wanted_spec, &spec) < 0)
    {
        dbmsg("SDL_OpenAudio:%s", SDL_GetError());
        return -1;
    }
    packet_queue_init(&audioq);
    SDL_PauseAudio(0);

    //与SDL的退出相关
    affmutex = SDL_CreateMutex();
    sdl_thread = SDL_CreateThread(eventThread, NULL);

    rect.x = 0;
    rect.y = 0;
    rect.w = pCodecCtx->width;
    rect.h = pCodecCtx->height;

    packet = (AVPacket*)av_malloc(sizeof(AVPacket));
    if(videoindex != -1)
    {
        img_convert_ctx = sws_getContext(pCodecCtx->width, pCodecCtx->height, pCodecCtx->pix_fmt, pCodecCtx->width, pCodecCtx->height, PIX_FMT_YUV420P, SWS_BICUBIC, NULL, NULL, NULL);
        // img_convert_ctx = sws_getContext(pCodecCtx->width, pCodecCtx->height, pCodecCtx->pix_fmt, pCodecCtx->width, pCodecCtx->height, PIX_FMT_YUYV422, SWS_BICUBIC, NULL, NULL, NULL);
        if(img_convert_ctx == NULL)
        {
            dbmsg("img_convert error");
            return -1;
        }
    }
    while( (av_read_frame(pFormatCtx, packet)>=0) && (signal_quit))
    {
        if(packet->stream_index == videoindex)
        {
            if((ret=avcodec_decode_video2(pCodecCtx, pFrame, &frameFinished, packet)) < 0)
            {
                dbmsg("decocode video error");
                return -1;
            }
            if(frameFinished)
            {
                SDL_LockYUVOverlay(overlay);
                pFrameYUV->data[0] = overlay->pixels[0];
                pFrameYUV->data[1] = overlay->pixels[2];
                pFrameYUV->data[2] = overlay->pixels[1];
                pFrameYUV->linesize[0] = overlay->pitches[0];
                pFrameYUV->linesize[1] = overlay->pitches[2];
                pFrameYUV->linesize[2] = overlay->pitches[1];
                sws_scale(img_convert_ctx, (const uint8_t* const*)pFrame->data, pFrame->linesize, 0, 
                        pCodecCtx->height, pFrameYUV->data, pFrameYUV->linesize); 
                SDL_UnlockYUVOverlay(overlay);
                SDL_DisplayYUVOverlay(overlay, &rect);
                SDL_Delay(40);
            }
        }
        if(packet->stream_index == sndindex)
        {
            packet_queue_put(&audioq, packet);
        }
    }
    SDL_WaitThread(sdl_thread, &ret);
    SDL_DestroyMutex(affmutex);
    av_free_packet(packet);
    av_free(pFrameYUV);
    av_free(pFrame);
    avcodec_close(pCodecCtx);
    avcodec_close(sndCodecCtx);
    avformat_close_input(&pFormatCtx);
    return 0;
}

1.2 Makefile

cong@msi:/work/ffmpeg/test/3sound$ cat Makefile 
EXE=sound
CC=gcc
FFMPEG=/work/ffmpeg/out
CFLAGS=-g -O0 -I$(FFMPEG)/include
LDFLAGS = -L$(FFMPEG)/lib/ -lswscale -lswresample -lavformat -lavdevice -lavcodec -lavutil -lavfilter -lm -lSDL
SRC=$(wildcard *.c)
OBJ=$(patsubst %.c,%.o,$(SRC))
DEP=$(patsubst %.c,.%.d,$(SRC))
$(EXE):$(OBJ)
    $(CC) $(CFLAGS) $^ -o $@ $(LDFLAGS)
$(DEP):.%.d:%.c
    @set -e; rm -f $@; \
    $(CC) -MM $< > $@.$$$$; \
    sed 's,/($*/)/.o[ :]*,/1.o $@ : ,g' < $@.$$$$ > $@; \
    rm -f $@.$$$$

-include $(DEP)
clean:
    @rm $(EXE) $(OBJ) $(DEP) -f
run:
    export LD_LIBRARY_PATH=$(FFMPEG)/lib/ \
    && ./$(EXE) ../resource/test.mp3
    #&& ./$(EXE) ../resource/test.wav
    #&& ./$(EXE) ../resource/test.wmv
    #&& ./$(EXE) ../resource/test.rmvb

测试时发现只有wav格式的播放没有问题，

1.3 代码打包

3sound.rar (下载后改名为3sound.tar.gz)

二.改进
2.1 要改进的问题
只有wav播放没有问题,其它格式的播放时会有噪音,看了一下ffplay的代码现在都是用的AVFilter
2.2 代码

cong@msi:/work/ffmpeg/test/3sound_1$ cat sound.c 
#include "utils.h"
#include <libavcodec/avcodec.h>
#include <libavformat/avformat.h>
#include <libswscale/swscale.h>
#include <libswresample/swresample.h>
#include <libavutil/avstring.h>
#include <libavutil/pixfmt.h>
#include <libavutil/log.h>
#include <SDL/SDL.h>
#include <SDL/SDL_thread.h>
#include <stdio.h>
#include <math.h>

#define SDL_AUDIO_BUFFER_SIZE 4096
#define AVCODEC_MAX_AUDIO_FRAME_SIZE 192000
//#define SDL_AUDIO_BUFFER_SIZE 8192
//#define AVCODEC_MAX_AUDIO_FRAME_SIZE 384000
DECLARE_ALIGNED(16,uint8_t,audio_buf2) [AVCODEC_MAX_AUDIO_FRAME_SIZE * 4];
struct SwrContext *swr_ctx = NULL;

SDL_AudioSpec wanted_spec, spec;
SDL_mutex *affmutex;
SDL_Event sdlevent;
int signal_quit = 1;
enum AVSampleFormat audio_src_fmt;
int64_t wanted_channel_layout;
typedef struct PacketQueue
{
    AVPacketList * first_pkt, *last_pkt;
    int nb_packets;
    int size;
    SDL_mutex *mutex;
    SDL_cond * cond;
}PacketQueue;
PacketQueue audioq;

static int eventThread(void* data)
{
    while(signal_quit)
    {
        SDL_LockMutex(affmutex);
        while(SDL_PollEvent(&sdlevent))
        {
            switch(sdlevent.type)
            {
                case SDL_QUIT:
                    {
                        signal_quit = 0;
                    }
                    break;
                default:
                    break;

            }
        }
        SDL_UnlockMutex(affmutex);
    } 
}
void packet_queue_init(PacketQueue *q)
{
    memset(q, 0, sizeof(PacketQueue));
    q->mutex = SDL_CreateMutex();
    q->cond = SDL_CreateCond();
}

int packet_queue_put(PacketQueue *q, AVPacket *pkt)
{

    AVPacketList *pkt1;
    if(av_dup_packet(pkt) < 0)
        return -1;
    pkt1 = av_malloc(sizeof(AVPacketList));
    if (!pkt1)
        return -1;
    pkt1->pkt = *pkt;
    pkt1->next = NULL;

    SDL_LockMutex(q->mutex);

    if (!q->last_pkt)
        q->first_pkt = pkt1;
    else
        q->last_pkt->next = pkt1;
    q->last_pkt = pkt1;
    q->nb_packets++;
    q->size += pkt1->pkt.size;
    SDL_CondSignal(q->cond);

    SDL_UnlockMutex(q->mutex);
    return 0;
}

static int packet_queue_get(PacketQueue *q, AVPacket *pkt, int block)
{
    AVPacketList *pkt1;
    int ret;
    SDL_LockMutex(q->mutex);
    for(;;)
    {
        if(0 == signal_quit)
        {
            ret = -1;
            break;
        }
        pkt1 = q->first_pkt;
        if (pkt1)
        {
            q->first_pkt = pkt1->next;
            if (!q->first_pkt)
                q->last_pkt = NULL;
            q->nb_packets--;
            q->size -= pkt1->pkt.size;
            *pkt = pkt1->pkt;
            av_free(pkt1);
            ret = 1;
            break;
        } else if (!block) {
            ret = 0;
            break;
        } else {
            SDL_CondWait(q->cond, q->mutex);
        }
    }
    SDL_UnlockMutex(q->mutex);
    return ret;
}

int audio_decode_frame(AVCodecContext *aCodecCtx, AVPacket *pkt, AVPacket *pkt_temp, AVFrame *frame)
{
    int i;
    int len1,len2, data_size, got_frame;
    int new_packet;
    int64_t dec_channel_layout;
    audio_src_fmt = AV_SAMPLE_FMT_S16; //AUDIO_S16SYS;
    static once = 1;
    uint8_t *out[] = { audio_buf2 };
    for(;;)
    {
        while(pkt_temp->size>0 || (!pkt_temp->data && new_packet))
        {
            if(!frame)
            {
                if (!(frame = av_frame_alloc())) //av_frame_alloc(); //avcodec_alloc_frame
                    return AVERROR(ENOMEM);
            }
            else
            {
                dbmsg("av_get default frame");
                //avcodec_get_frame_defaults(frame);
                av_frame_unref(frame); //reset frame
            }
            new_packet = 0;
            dbmsg();
            len1 = avcodec_decode_audio4(aCodecCtx, frame, &got_frame, pkt_temp); //decode data is store in frame
            if(len1 < 0)
            {
                pkt_temp->size = 0;
                break;
            }
            //dbmsg("len1=%d, linesize=%d",len1, frame->linesize[0]);

            pkt_temp->data += len1;
            pkt_temp->size -= len1;
            //这儿解码出来之后加入swr_convert
            if(got_frame <= 0) /* No data yet, get more frames */
                continue;
            //dbmsg("add filter: sample=%d", frame->nb_samples);
            data_size = av_samples_get_buffer_size(NULL, aCodecCtx->channels, frame->nb_samples, aCodecCtx->sample_fmt, 1);
            wanted_channel_layout = av_get_default_channel_layout(spec.channels);
            dec_channel_layout = (frame->channel_layout && frame->channels==av_get_channel_layout_nb_channels(frame->channel_layout))?
                frame->channel_layout: wanted_channel_layout;

            dbmsg("format=%d:%d,layout=%d:%d,rate=%d:%d,samples=%d:%d",frame->format,audio_src_fmt,
                    dec_channel_layout,av_get_default_channel_layout(spec.channels),
              frame->sample_rate,spec.freq,frame->nb_samples,spec.samples );
            
            dbmsg("wanted_channel_layout=%d", wanted_channel_layout);
            dbmsg("audio_src_fmt=%d", audio_src_fmt);
            dbmsg("spec.freq=%d", spec.freq);
            //check: format,channel_layout,rate,sample
            dbmsg("once =%d", once);
            if((once) &&( (frame->format!=audio_src_fmt) || (dec_channel_layout!=av_get_default_channel_layout(spec.channels)) ||
              (frame->sample_rate!=spec.freq) || (frame->nb_samples!=spec.samples)) )
            {
                dbmsg();
                if(swr_ctx != NULL)
                    swr_free(&swr_ctx);
                dbmsg("wanted_channel_layout=%d, audio_src_fmt=%d, frame->sample_rate=%d, dec_channel_layout=%d, frame->format=%d, frame->sample_rate=%d",wanted_channel_layout, audio_src_fmt, frame->sample_rate,dec_channel_layout, frame->format, frame->sample_rate);
                swr_ctx = swr_alloc_set_opts(NULL, wanted_channel_layout, audio_src_fmt, frame->sample_rate, dec_channel_layout, frame->format, frame->sample_rate, 0, NULL);
                if(swr_ctx == NULL)
                {
                    dbmsg("swr_ctx == NULL");
                }
                swr_init(swr_ctx);
                dbmsg();
                once = 0;
            }
            if(swr_ctx)
            {
                len2 = swr_convert(swr_ctx, out, sizeof(audio_buf2)/spec.channels/av_get_bytes_per_sample(frame->format),(const uint8_t **)frame->extended_data, frame->nb_samples);
                dbmsg("needed swr_onvert, len2=%d", len2);
                data_size = len2 * spec.channels * av_get_bytes_per_sample(audio_src_fmt);
            }else {
                memcpy(audio_buf2, frame->data[0], frame->linesize[0]);
            }

            dbmsg("data_size =%d", data_size);
            return data_size;
        }
        if(pkt->data)
            av_free_packet(pkt);
        memset(pkt_temp, 0, sizeof(*pkt_temp));
        if(0 == signal_quit)
            return -1;

        if((new_packet = packet_queue_get(&audioq, pkt, 1)) < 0)
            return -1;
        *pkt_temp = *pkt;
    }
}

void audio_callback(void *userdata, Uint8 *stream, int len)
{
    int i;
    AVCodecContext *aCodecCtx = (AVCodecContext *)userdata;
    int len1, audio_size;

    static unsigned int audio_buf_size = 0;
    static unsigned int audio_buf_index = 0;

    AVPacket *pkt = av_mallocz(sizeof(AVPacket));
    AVPacket *pkt_temp = av_mallocz(sizeof(AVPacket));
    AVFrame *frame = NULL;
    dbmsg("len=%d", len);
    while(len > 0)
    {
        if(audio_buf_index >= audio_buf_size)
        {
            audio_size = audio_decode_frame(aCodecCtx, pkt, pkt_temp, frame);
            dbmsg("audio_size=%d", audio_size);
            if(audio_size < 0)
            {
                /* If error, output silence */
                audio_buf_size = 1024;
                memset(audio_buf2, 0, audio_buf_size);
            } else {
                audio_buf_size = audio_size;
            }
            audio_buf_index = 0;
        }
        //dbmsg("len=%d, audio_buf_size=%d, audio_buf_index=%d", len, audio_buf_size, audio_buf_index);
        len1 = audio_buf_size - audio_buf_index;
        //dbmsg("len1=%d", len1);
        if(len1 > len)
            len1 = len;
        //dbmsg("len1 = %d", len1);
        memcpy(stream, (uint8_t *)audio_buf2 + audio_buf_index, len1);
        len -= len1;
        stream += len1;
        audio_buf_index += len1;
    }
}

int main(int argc, char **argv)
{
    int i=0;
    int ret;
    int videoindex= -1;
    int sndindex= -1;
    int frameFinished;
    AVFormatContext *pFormatCtx = NULL;
    AVCodecContext * pCodecCtx;
    AVCodecContext * sndCodecCtx;
    AVCodec * pCodec;
    AVCodec * sndCodec;
    AVFrame * pFrame;
    AVFrame * pFrameYUV;
    AVPacket * packet; 
    struct SwsContext *img_convert_ctx;

    SDL_Surface* psscreen;
    SDL_Overlay* overlay;
    SDL_Rect rect;
    SDL_Thread* sdl_thread;

    avcodec_register_all();
    avfilter_register_all();
    av_register_all();

    pFormatCtx = avformat_alloc_context();

    if(avformat_open_input(&pFormatCtx, argv[1], NULL, NULL)!=0)
        return -1; 

    if(avformat_find_stream_info(pFormatCtx, NULL)<0)
        return -1;
    av_dump_format(pFormatCtx,0, 0, 0);

    for(i=0; i<pFormatCtx->nb_streams; i++)
    {
        if(pFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO)
        {
            videoindex= i;
        }
        if(pFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO)
        {
            sndindex= i;
        }
    }
    if(videoindex== -1)
    {
        dbmsg("no video stream found!");
    }
    if(sndindex== -1)
    {
        dbmsg("no sound stream found!");
    }
    dbmsg("videoindex=%d, sndindex=%d", videoindex, sndindex);
    if(videoindex != -1)
    {
        pCodecCtx = pFormatCtx->streams[videoindex]->codec;
        pCodec = avcodec_find_decoder(pCodecCtx->codec_id);
        if(pCodec == NULL)
        {
            dbmsg("Codec not found");
            return -1;
        }
        if(avcodec_open2(pCodecCtx, pCodec, NULL) < 0)
            return -1;
        pFrame = av_frame_alloc(); //avcodec_alloc_frame
        pFrameYUV = av_frame_alloc();

        SDL_Init(SDL_INIT_EVERYTHING);
        psscreen = SDL_SetVideoMode(pCodecCtx->width, pCodecCtx->height, 0, SDL_SWSURFACE);
        SDL_WM_SetCaption( "FFMPEG Window", NULL);
        overlay = SDL_CreateYUVOverlay(pCodecCtx->width, pCodecCtx->height, SDL_YV12_OVERLAY, psscreen);
        //overlay = SDL_CreateYUVOverlay(pCodecCtx->width, pCodecCtx->height, SDL_YUY2_OVERLAY, psscreen);
    }
    if(sndindex != -1)
    {
        sndCodecCtx = pFormatCtx->streams[sndindex]->codec;
        sndCodec = avcodec_find_decoder(sndCodecCtx->codec_id);
        if(sndCodec == NULL)
        {
            dbmsg("Codec not found");
            return -1;
        }
        if(avcodec_open2(sndCodecCtx, sndCodec, NULL) < 0)
            return -1;
    
        wanted_spec.freq = sndCodecCtx->sample_rate;
        wanted_spec.format = AUDIO_S16SYS;
        wanted_spec.channels = sndCodecCtx->channels;
        wanted_spec.silence = 0;
        wanted_spec.samples = 2048;
        wanted_spec.callback = audio_callback;
        wanted_spec.userdata = sndCodecCtx;
        if(SDL_OpenAudio(&wanted_spec, &spec) < 0)
        {
            dbmsg("SDL_OpenAudio:%s", SDL_GetError());
            return -1;
        }
        dbmsg("freq=%d, channels=%d, samples=%d", spec.freq, spec.channels, spec.samples);
        packet_queue_init(&audioq);
        SDL_PauseAudio(0);
    }
    //sdl get signal quit
    affmutex = SDL_CreateMutex();
    sdl_thread = SDL_CreateThread(eventThread, NULL);

    rect.x = 0;
    rect.y = 0;
    rect.w = pCodecCtx->width;
    rect.h = pCodecCtx->height;

    packet = (AVPacket*)av_malloc(sizeof(AVPacket));
    if(videoindex != -1)
    {
        img_convert_ctx = sws_getContext(pCodecCtx->width, pCodecCtx->height, pCodecCtx->pix_fmt, pCodecCtx->width, pCodecCtx->height, PIX_FMT_YUV420P, SWS_BICUBIC, NULL, NULL, NULL);
        // img_convert_ctx = sws_getContext(pCodecCtx->width, pCodecCtx->height, pCodecCtx->pix_fmt, pCodecCtx->width, pCodecCtx->height, PIX_FMT_YUYV422, SWS_BICUBIC, NULL, NULL, NULL);
        if(img_convert_ctx == NULL)
        {
            dbmsg("img_convert error");
            return -1;
        }
    }
    while( (av_read_frame(pFormatCtx, packet)>=0) && (signal_quit))
    {
        if(packet->stream_index == videoindex)
        {
            dbmsg();
            if((ret=avcodec_decode_video2(pCodecCtx, pFrame, &frameFinished, packet)) < 0)
            {
                dbmsg("decocode video error");
                return -1;
            }
            if(frameFinished)
            {
                SDL_LockYUVOverlay(overlay);
                pFrameYUV->data[0] = overlay->pixels[0];
                pFrameYUV->data[1] = overlay->pixels[2];
                pFrameYUV->data[2] = overlay->pixels[1];
                pFrameYUV->linesize[0] = overlay->pitches[0];
                pFrameYUV->linesize[1] = overlay->pitches[2];
                pFrameYUV->linesize[2] = overlay->pitches[1];
                sws_scale(img_convert_ctx, (const uint8_t* const*)pFrame->data, pFrame->linesize, 0, 
                        pCodecCtx->height, pFrameYUV->data, pFrameYUV->linesize); 
                SDL_UnlockYUVOverlay(overlay);
                SDL_DisplayYUVOverlay(overlay, &rect);
                SDL_Delay(40);
            }
        }
        if(packet->stream_index == sndindex)
        {
            packet_queue_put(&audioq, packet);
        }
    }
    SDL_WaitThread(sdl_thread, &ret);
    SDL_DestroyMutex(affmutex);
    av_free_packet(packet);
    av_free(pFrameYUV);
    av_free(pFrame);
    avcodec_close(pCodecCtx);
    avcodec_close(sndCodecCtx);
    avformat_close_input(&pFormatCtx);
    return 0;
}
cong@msi:/work/ffmpeg/test/3sound_1$

2.3 代码打包

3sound_1.rar (下载后改名为3sound_1.tar.gz)

三.代码
3.1 再次改进
上面的代码太乱了,有很多的全局变量,下面整理一下,学习ffplay做成一个结构体

附录:1.测试的文件格式说明

test.rmvb--> 
    Stream 0
        Type: Audio
        Codec: Cook Auidio(cook)
        Channels: Stereo
        Sample rate: 44100HZ
        Bits per sample: 32
    Stream 1
        Type: Video
        Codec: RealVideo 9/10(4.0)(RV40)
        Resolution: 1024*576
        Decoded format: Planar 4:2:0 YUV


test.wmv -->
    Stream 0 
        Type: Video
        Codec: Windows Media Video 9 (WMV3)
        Language: Chinese
        Resolution: 620x466
        Frame rate: 5
        Decoded format: Planar 4:2:0 YUV
    Stream 1
        Type: Audio
        Codec: Windows Media Audio 2 (WMA2)
        Language: Chinese
        Channels: Mono
        Sample rate: 22050 Hz
        Bits per sample: 32
test.mp3 -->
    Type: Audio
    Codec: MPEG Audio layer 1/2/3 (mpga)
    Channels: Stereo
    Sample rate: 44100 Hz
    Bitrate: 192 kb/s
test.wav -->
    Type: Audio
    Codec: PCM S16 LE (s16l)
    Channels: Stereo
    Sample rate: 44100 Hz
    Bits per sample: 16

0 0