ffmpeg探测网络流格式方法

最近遇到好几个人在问ffmpeg如何处理网络流，刚好前段时间也在做这方面，抽空整理了下，把主要代码发出来，希望对大

家有用。为简单处理，我这里只简单介绍ＵＤＰ接收ＴＳ流，其实只要是socket接收的都可以类似处理。

 

/* main.c
 * Created on: 2011-9-18
 * Author: wudegang
 */

#include "utils.h"
#include <pthread.h>
#include <libavcodec/avcodec.h>
#include <libavformat/avformat.h>

UdpQueue recvqueue;
UdpParam udpParam;

//注册av_read_frame的回调函数，这里只是最简处理，实际应用中应加上出错处理，超时等待...
int read_data(void *opaque, uint8_t *buf, int buf_size) {
    int size = buf_size;
    int ret;
    // printf("read data %d\n", buf_size);
    do {
        ret = get_queue(&recvqueue, buf, buf_size);
    } while (ret);
    // printf("read data Ok %d\n", buf_size);
    return size;
}
#define BUF_SIZE 4096*500
int main(int argc, char** argv)
{
    init_queue(&recvqueue, 1024*500);
    udpParam.argv = argv;
    udpParam.queue = &recvqueue;
    uint8_t *buf = av_mallocz(sizeof(uint8_t)*BUF_SIZE);

    //UDP接收线程
    pthread_t udp_recv_thread;
    pthread_create(&udp_recv_thread, NULL, udp_ts_recv, &udpParam);
    pthread_detach(udp_recv_thread);

    av_register_all();

    AVCodec *pVideoCodec, *pAudioCodec;
    AVCodecContext *pVideoCodecCtx = NULL;
    AVCodecContext *pAudioCodecCtx = NULL;
    AVIOContext * pb = NULL;
    AVInputFormat *piFmt = NULL;
    AVFormatContext *pFmt = NULL;

    //step1:申请一个AVIOContext
    pb = avio_alloc_context(buf, BUF_SIZE, 0, NULL, read_data, NULL, NULL);
    if (!pb) {
        fprintf(stderr, "avio alloc failed!\n");
        return -1;
    }
    //step2:探测流格式
    if (av_probe_input_buffer(pb, &piFmt, "", NULL, 0, 0) < 0) {
        fprintf(stderr, "probe failed!\n");
        return -1;
    } else {
        fprintf(stdout, "probe success!\n");
        fprintf(stdout, "format: %s[%s]\n", piFmt->name, piFmt->long_name);
    }

    pFmt = avformat_alloc_context();
    pFmt->pb = pb; //step3:这一步很关键
    //step4:打开流
    if (avformat_open_input(&pFmt, "", piFmt, NULL) < 0) {
        fprintf(stderr, "avformat open failed.\n");
        return -1;
    } else {
        fprintf(stdout, "open stream success!\n");
    }
    //以下就和文件处理一致了
    if (av_find_stream_info(pFmt) < 0) {
        fprintf(stderr, "could not fine stream.\n");
        return -1;
    }

    av_dump_format(pFmt, 0, "", 0);

    int videoindex = -1;
    int audioindex = -1;
    for (int i = 0; i < pFmt->nb_streams; i++) {
        if ( (pFmt->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO) && (videoindex < 0) ) {
            videoindex = i;
        }
        if ( (pFmt->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) && (audioindex < 0) ) {
            audioindex = i;
        }
    }

    if (videoindex < 0 || audioindex < 0) {
        fprintf(stderr, "videoindex=%d, audioindex=%d\n", videoindex, audioindex);
        return -1;
    }

    AVStream *pVst,*pAst;
    pVst = pFmt->streams[videoindex];
    pAst = pFmt->streams[audioindex];

    pVideoCodecCtx = pVst->codec;
    pAudioCodecCtx = pAst->codec;

    pVideoCodec = avcodec_find_decoder(pVideoCodecCtx->codec_id);
    if (!pVideoCodec) {
        fprintf(stderr, "could not find video decoder!\n");
        return -1;
    }
    if (avcodec_open(pVideoCodecCtx, pVideoCodec) < 0) {
        fprintf(stderr, "could not open video codec!\n");
        return -1;
    }

    pAudioCodec = avcodec_find_decoder(pAudioCodecCtx->codec_id);
    if (!pAudioCodec) {
        fprintf(stderr, "could not find audio decoder!\n");
        return -1;
    }
    if (avcodec_open(pAudioCodecCtx, pAudioCodec) < 0) {
        fprintf(stderr, "could not open audio codec!\n");
        return -1;
    }

    int got_picture;
    uint8_t samples[AVCODEC_MAX_AUDIO_FRAME_SIZE*3/2];
    AVFrame *pframe = avcodec_alloc_frame();
    AVPacket pkt;
    av_init_packet(&pkt);

    while(1) {
        if (av_read_frame(pFmt, &pkt) >= 0) {
            if (pkt.stream_index == videoindex) {
                fprintf(stdout, "pkt.size=%d,pkt.pts=%lld, pkt.data=0x%x.", pkt.size, pkt.pts,(unsigned 

int)pkt.data);
                avcodec_decode_video2(pVideoCodecCtx, pframe, &got_picture, &pkt);
                if (got_picture) {
                    fprintf(stdout, "decode one video frame!\n");
                }
            }else if (pkt.stream_index == audioindex) {
                int frame_size = AVCODEC_MAX_AUDIO_FRAME_SIZE*3/2;
                if (avcodec_decode_audio3(pAudioCodecCtx, (int16_t *)samples, &frame_size, &pkt) >= 0) {
                    fprintf(stdout, "decode one audio frame!\n");
                }
            }
            av_free_packet(&pkt);
        }
    }

    av_free(buf);
    av_free(pframe);
    free_queue(&recvqueue);
    return 0;
}

转自http://topic.csdn.net/u/20111014/11/25c37c13-65e3-4f90-be98-feb964934f23.html?

seed=785392903&r=76622417#r_76622417