【FFMPEG源码分析】从ffplay源码摸清ffmpeg框架(一)

ffplay入口

ffmpeg\fftools\ffplay.c
int main(int argc, char **argv)
{/*******************start 动态库加载/网络初始化等**************/int flags;VideoState *is;init_dynload();av_log_set_flags(AV_LOG_SKIP_REPEATED);parse_loglevel(argc, argv, options);/* register all codecs, demux and protocols */
#if CONFIG_AVDEVICEavdevice_register_all();
#endifavformat_network_init();signal(SIGINT , sigterm_handler); /* Interrupt (ANSI).    */signal(SIGTERM, sigterm_handler); /* Termination (ANSI).  */show_banner(argc, argv, options);parse_options(NULL, argc, argv, options, opt_input_file);if (!input_filename) {show_usage();av_log(NULL, AV_LOG_FATAL, "An input file must be specified\n");av_log(NULL, AV_LOG_FATAL,"Use -h to get full help or, even better, run 'man %s'\n", program_name);exit(1);}/*******************end动态库加载/网络初始化等**************//*******************start 视频帧显示窗口初始化*******************/if (display_disable) {video_disable = 1;}flags = SDL_INIT_VIDEO | SDL_INIT_AUDIO | SDL_INIT_TIMER;if (audio_disable)flags &= ~SDL_INIT_AUDIO;else {/* Try to work around an occasional ALSA buffer underflow issue when the* period size is NPOT due to ALSA resampling by forcing the buffer size. */if (!SDL_getenv("SDL_AUDIO_ALSA_SET_BUFFER_SIZE"))SDL_setenv("SDL_AUDIO_ALSA_SET_BUFFER_SIZE","1", 1);}if (display_disable)flags &= ~SDL_INIT_VIDEO;if (SDL_Init (flags)) {av_log(NULL, AV_LOG_FATAL, "Could not initialize SDL - %s\n", SDL_GetError());av_log(NULL, AV_LOG_FATAL, "(Did you set the DISPLAY variable?)\n");exit(1);}SDL_EventState(SDL_SYSWMEVENT, SDL_IGNORE);SDL_EventState(SDL_USEREVENT, SDL_IGNORE);if (!display_disable) {int flags = SDL_WINDOW_HIDDEN;if (alwaysontop)
#if SDL_VERSION_ATLEAST(2,0,5)flags |= SDL_WINDOW_ALWAYS_ON_TOP;
#elseav_log(NULL, AV_LOG_WARNING, "Your SDL version doesn't support SDL_WINDOW_ALWAYS_ON_TOP. Feature will be inactive.\n");
#endifif (borderless)flags |= SDL_WINDOW_BORDERLESS;elseflags |= SDL_WINDOW_RESIZABLE;#ifdef SDL_HINT_VIDEO_X11_NET_WM_BYPASS_COMPOSITORSDL_SetHint(SDL_HINT_VIDEO_X11_NET_WM_BYPASS_COMPOSITOR, "0");
#endifwindow = SDL_CreateWindow(program_name, SDL_WINDOWPOS_UNDEFINED, SDL_WINDOWPOS_UNDEFINED, default_width, default_height, flags);SDL_SetHint(SDL_HINT_RENDER_SCALE_QUALITY, "linear");if (window) {renderer = SDL_CreateRenderer(window, -1, SDL_RENDERER_ACCELERATED | SDL_RENDERER_PRESENTVSYNC);if (!renderer) {av_log(NULL, AV_LOG_WARNING, "Failed to initialize a hardware accelerated renderer: %s\n", SDL_GetError());renderer = SDL_CreateRenderer(window, -1, 0);}if (renderer) {if (!SDL_GetRendererInfo(renderer, &renderer_info))av_log(NULL, AV_LOG_VERBOSE, "Initialized %s renderer.\n", renderer_info.name);}}if (!window || !renderer || !renderer_info.num_texture_formats) {av_log(NULL, AV_LOG_FATAL, "Failed to create window or renderer: %s", SDL_GetError());do_exit(NULL);}}/*******************end 视频帧显示窗口初始化*******************//******************* 下载数据/demux/decoder*******************/is = stream_open(input_filename, file_iformat);if (!is) {av_log(NULL, AV_LOG_FATAL, "Failed to initialize VideoState!\n");do_exit(NULL);}event_loop(is);return 0;
}

从上面main函数入口来看，main中大部分是一些init工作，而ffmpeg的核心工作demux/decoder的工作都封装在stream_open中了。
想要摸清ffmpeg的框架，学习思路是追踪音视频数据的流动过程。

1. 原始数据通过什么模块获取？
2. 如何通过原始数据匹配到对应的demuxer ，如: ts, mp4, mov, avi等？
3. 如何通过demuxer后的audio/video数据匹配到对应的decoder， 如：h264， vp9， aac等？
4. ffmpeg中数据流转的驱动模型是什么样的？也就是数据是从什么地方驱动开始下载/demux/decode ?

static VideoState *stream_open(const char *filename,const AVInputFormat *iformat)
{VideoState *is = av_mallocz(sizeof(VideoState));is->last_video_stream = is->video_stream = -1;is->last_audio_stream = is->audio_stream = -1;is->last_subtitle_stream = is->subtitle_stream = -1;is->filename = av_strdup(filename);is->iformat = iformat;is->ytop    = 0;is->xleft   = 0;is->continue_read_thread = SDL_CreateCond();/* start video display */frame_queue_init(&is->pictq, &is->videoq, VIDEO_PICTURE_QUEUE_SIZE, 1);frame_queue_init(&is->subpq, &is->subtitleq, SUBPICTURE_QUEUE_SIZE, 0);frame_queue_init(&is->sampq, &is->audioq, SAMPLE_QUEUE_SIZE, 1);packet_queue_init(&is->videoq);packet_queue_init(&is->audioq);packet_queue_init(&is->subtitleq);init_clock(&is->vidclk, &is->videoq.serial);init_clock(&is->audclk, &is->audioq.serial);init_clock(&is->extclk, &is->extclk.serial);is->audio_clock_serial = -1;startup_volume = av_clip(startup_volume, 0, 100);startup_volume = av_clip(SDL_MIX_MAXVOLUME * startup_volume / 100, 0, SDL_MIX_MAXVOLUME);is->audio_volume = startup_volume;is->muted = 0;is->av_sync_type = av_sync_type;is->read_tid     = SDL_CreateThread(read_thread, "read_thread", is);if (!is->read_tid) {av_log(NULL, AV_LOG_FATAL, "SDL_CreateThread(): %s\n", SDL_GetError());
fail:stream_close(is);return NULL;return is;
}

从上面的代码看，stream_open主要是用于初始化VideoState这个结构体，初始化audio/video/subtitile解码后数据的buffer queque。结构体的具体细节先暂不深入。
另外在stream_open中创建了一个thread: read_thread, 看来干活的是这个了。。。

static int read_thread(void *arg)
{VideoState *is = arg;AVPacket *pkt = av_packet_alloc();AVFormatContext *ic = avformat_alloc_context();//启动数据下载模块，demux模块avformat_open_input(&ic, is->filename, is->iformat, &format_opts);is->ic = ic;av_format_inject_global_side_data(ic);if (find_stream_info) {AVDictionary **opts = setup_find_stream_info_opts(ic, codec_opts);avformat_find_stream_info(ic, opts);}if (!video_disable)st_index[AVMEDIA_TYPE_VIDEO] =av_find_best_stream(ic, AVMEDIA_TYPE_VIDEO,st_index[AVMEDIA_TYPE_VIDEO], -1, NULL, 0);if (!audio_disable)st_index[AVMEDIA_TYPE_AUDIO] =av_find_best_stream(ic, AVMEDIA_TYPE_AUDIO,st_index[AVMEDIA_TYPE_AUDIO],st_index[AVMEDIA_TYPE_VIDEO],NULL, 0);if (!video_disable && !subtitle_disable)st_index[AVMEDIA_TYPE_SUBTITLE] =av_find_best_stream(ic, AVMEDIA_TYPE_SUBTITLE,st_index[AVMEDIA_TYPE_SUBTITLE],(st_index[AVMEDIA_TYPE_AUDIO] >= 0 ?st_index[AVMEDIA_TYPE_AUDIO] :st_index[AVMEDIA_TYPE_VIDEO]),NULL, 0);// 启动decoder threadstream_component_open(is, st_index[AVMEDIA_TYPE_AUDIO]);stream_component_open(is, st_index[AVMEDIA_TYPE_VIDEO]);stream_component_open(is, st_index[AVMEDIA_TYPE_SUBTITLE]);for (;;) {if (is->abort_request)break;if (is->paused != is->last_paused) {is->last_paused = is->paused;if (is->paused)is->read_pause_return = av_read_pause(ic);elseav_read_play(ic);}//读取demux后的数据，并存入audio/video/subtitle的buffer queue中av_read_frame(ic, pkt);/* check if packet is in play range specified by user, then queue, otherwise discard */stream_start_time = ic->streams[pkt->stream_index]->start_time;pkt_ts = pkt->pts == AV_NOPTS_VALUE ? pkt->dts : pkt->pts;pkt_in_play_range = duration == AV_NOPTS_VALUE ||(pkt_ts - (stream_start_time != AV_NOPTS_VALUE ? stream_start_time : 0)) *av_q2d(ic->streams[pkt->stream_index]->time_base) -(double)(start_time != AV_NOPTS_VALUE ? start_time : 0) / 1000000<= ((double)duration / 1000000);if (pkt->stream_index == is->audio_stream && pkt_in_play_range) {packet_queue_put(&is->audioq, pkt);} else if (pkt->stream_index == is->video_stream && pkt_in_play_range&& !(is->video_st->disposition & AV_DISPOSITION_ATTACHED_PIC)) {packet_queue_put(&is->videoq, pkt);} else if (pkt->stream_index == is->subtitle_stream && pkt_in_play_range) {packet_queue_put(&is->subtitleq, pkt);} else {av_packet_unref(pkt);}}..............;return 0;
}

从上面read_thread代码看，主要是通过avformat_open_input(…)启动数据下载和demux，再通过stream_component_open(…) 启动audio/video/subtitle的decoder。最后启动死循环通过av_read_frame(…) 读取demux出来的audio/video/subtitle存入对应的buffer queue供decoder使用。
下面再看看decoder是如何消耗数据的？以及decoder完成的数据缓存在什么地方？

static int stream_component_open(VideoState *is, int stream_index)
{AVFormatContext *ic = is->ic;const AVCodec *codec;const char *forced_codec_name = NULL;AVDictionary *opts = NULL;const AVDictionaryEntry *t = NULL;int sample_rate;AVChannelLayout ch_layout = { 0 };int ret = 0;int stream_lowres = lowres;AVCodecContext *avctx = avcodec_alloc_context3(NULL);avcodec_parameters_to_context(avctx, ic->streams[stream_index]->codecpar);const AVCodec * codec = avcodec_find_decoder(avctx->codec_id);avctx->codec_id = codec->id;AVDictionary *opts = filter_codec_opts(codec_opts, avctx->codec_id, ic, ic->streams[stream_index], codec);avcodec_open2(avctx, codec, &opts))；is->eof = 0;ic->streams[stream_index]->discard = AVDISCARD_DEFAULT;switch (avctx->codec_type) {case AVMEDIA_TYPE_AUDIO:/* prepare audio output */audio_open(is, &ch_layout, sample_rate, &is->audio_tgt))‘decoder_init(&is->auddec, avctx, &is->audioq, is->continue_read_thread))；decoder_start(&is->auddec, audio_thread, "audio_decoder", is))；static SDL_AudioDeviceID audio_dev;SDL_PauseAudioDevice(audio_dev, 0);break;case AVMEDIA_TYPE_VIDEO:is->video_stream = stream_index;is->video_st = ic->streams[stream_index];decoder_init(&is->viddec, avctx, &is->videoq, is->continue_read_thread))；decoder_start(&is->viddec, video_thread, "video_decoder", is))；is->queue_attachments_req = 1;break;case AVMEDIA_TYPE_SUBTITLE:is->subtitle_stream = stream_index;is->subtitle_st = ic->streams[stream_index];decoder_init(&is->subdec, avctx, &is->subtitleq, is->continue_read_thread))；decoder_start(&is->subdec, subtitle_thread, "subtitle_decoder", is))break;default:break;}....................;return ret;
}

ffplay内部结构图

通过stream_component_open函数的代码可以看出，先初始化decoder，再创建thread。
至此， ffplay的大致脉络已经摸清楚了，具体看下图:

结构关系图：

后面的文章继续深入分析demuxer和decoder。