ffmpeg主体架构分析

[时间：2016-07] [状态：Open]

[关键词：ffmpeg，libavcodec，libavformat]

FFmpeg接触几年了，用的比较多的是libavcodec和libavformat两个库，偶尔也会用用libswresample（主要处理音频PCM的转换，比如不同的声道数、频率、采样位数、量化位数转换）和libswscale（视频原始数据处理，比如缩放、色度格式、量化位数的转换）。

libavformat的主要机制

libavformat主要完成针对多媒体文件或流媒体（FFmpeg内部成为URL）的数据解析，包括数据读取、格式分析以及包读取；多媒体文件生成。其中主要包含几种主要的结构体：

• AVFormatContext
  
  最核心的结构体，对于每个URL，这里面都会demuxer/muxer，一个AVStream数组以及一个AVIOContext。
• AVStream
  
  记录媒体文件中包含的流信息，比如音频、视频或者数据流及其类型。
• AVInputFormat（demuxer）
  
  作为解析器，是libavformat中很多的结构，其对外接口如下：

int(* read_probe )(AVProbeData *)
	Tell if a given file has a chance of being parsed as this format.

int(* read_header )(struct AVFormatContext *)
	Read the format header and initialize the AVFormatContext structure.

int(* read_packet )(struct AVFormatContext *, AVPacket *pkt)
 	Read one packet and put it in 'pkt'.

int(* read_close )(struct AVFormatContext *)
 	Close the stream.

int(* read_seek )(struct AVFormatContext *, int stream_index, int64_t timestamp, int flags)
 	Seek to a given timestamp relative to the frames in stream component stream_index.

int64_t(* read_timestamp )(struct AVFormatContext *s, int stream_index, int64_t *pos, int64_t pos_limit)
 	Get the next timestamp in stream[stream_index].time_base units.

• AVOutputFormat（muxer）
  
  作为复用器，其对外接口如下：

int(* 	write_header )(struct AVFormatContext *)

int(* 	write_packet )(struct AVFormatContext *, AVPacket *pkt)

int(* 	write_trailer )(struct AVFormatContext *)

int(* 	interleave_packet )(struct AVFormatContext *, AVPacket *out, AVPacket *in, int flush)
 	Currently only used to set pixel format if not YUV420P.

int(* 	query_codec )(enum AVCodecID id, int std_compliance)
 	Test if the given codec can be stored in this container.

void(* 	get_output_timestamp )(struct AVFormatContext *s, int stream, int64_t *dts, int64_t *wall)

int(* 	control_message )(struct AVFormatContext *s, int type, void *data, size_t data_size)
 	Allows sending messages from application to device.

int(* 	write_uncoded_frame )(struct AVFormatContext *, int stream_index, AVFrame **frame, unsigned flags)
 	Write an uncoded AVFrame.

int(* 	get_device_list )(struct AVFormatContext *s, struct AVDeviceInfoList *device_list)
 	Returns device list with it properties.

int(* 	create_device_capabilities )(struct AVFormatContext *s, struct AVDeviceCapabilitiesQuery *caps)
 	Initialize device capabilities submodule.

int(* 	free_device_capabilities )(struct AVFormatContext *s, struct AVDeviceCapabilitiesQuery *caps)
 	Free device capabilities submodule.

int(* 	init )(struct AVFormatContext *)
 	Initialize format.

void(* 	deinit )(struct AVFormatContext *)
 	Deinitialize format.

int(* 	check_bitstream )(struct AVFormatContext *, const AVPacket *pkt)
 	Set up any necessary bitstream filtering and extract any extra data needed for the global header.

• AVIOContext、URLProtocol（协议解析）
  
  这里面涉及输入输出的机制，与具体协议有关，比如http、tcp、udp、rtp、rtsp等。
  
  URLProtocol的接口如下：

typedef struct URLProtocol {
    const char *name;
    int     (*url_open)( URLContext *h, const char *url, int flags);
    /**
     * This callback is to be used by protocols which open further nested
     * protocols. options are then to be passed to ffurl_open()/ffurl_connect()
     * for those nested protocols.
     */
    int     (*url_open2)(URLContext *h, const char *url, int flags, AVDictionary **options);
    int     (*url_accept)(URLContext *s, URLContext **c);
    int     (*url_handshake)(URLContext *c);

    /**
     * Read data from the protocol.
     * If data is immediately available (even less than size), EOF is
     * reached or an error occurs (including EINTR), return immediately.
     * Otherwise:
     * In non-blocking mode, return AVERROR(EAGAIN) immediately.
     * In blocking mode, wait for data/EOF/error with a short timeout (0.1s),
     * and return AVERROR(EAGAIN) on timeout.
     * Checking interrupt_callback, looping on EINTR and EAGAIN and until
     * enough data has been read is left to the calling function; see
     * retry_transfer_wrapper in avio.c.
     */
    int     (*url_read)( URLContext *h, unsigned char *buf, int size);
    int     (*url_write)(URLContext *h, const unsigned char *buf, int size);
    int64_t (*url_seek)( URLContext *h, int64_t pos, int whence);
    int     (*url_close)(URLContext *h);
    int (*url_read_pause)(URLContext *h, int pause);
    int64_t (*url_read_seek)(URLContext *h, int stream_index,
                             int64_t timestamp, int flags);
    int (*url_get_file_handle)(URLContext *h);
    int (*url_get_multi_file_handle)(URLContext *h, int **handles,
                                     int *numhandles);
    int (*url_shutdown)(URLContext *h, int flags);
    int priv_data_size;
    const AVClass *priv_data_class;
    int flags;
    int (*url_check)(URLContext *h, int mask);
    int (*url_open_dir)(URLContext *h);
    int (*url_read_dir)(URLContext *h, AVIODirEntry **next);
    int (*url_close_dir)(URLContext *h);
    int (*url_delete)(URLContext *h);
    int (*url_move)(URLContext *h_src, URLContext *h_dst);
    const char *default_whitelist;
} URLProtocol;

• AVPacket
  
  解复用或者复用之后的数据包，通常包含一帧视频或者一段音频数据。

就我而言，我用的比较多的是demuxer。通常FFmpeg的处理流程是，先通过demuxer的read_probe函数确定URL包含的容器类型，然后调用read_header读取多媒体的信息头，完成基本的初始化；之后正常读取packet通过read_packet和read_timestamp；最后在读取结束的时候调用read_close，完成反初始化操作。另外可以通过read_seek实现媒体文件的seek操作（快进/快退）。

libavcodec主要机制

libavcodec主要结合libavformat实现解码、编码及音视频解析（单独格式分包或打包）。其中主要包含以下结构：

• AVCodecContext
  
  AVCodecContext是libavcodec最核心的结构体，包含编码器或解码器类型，一个AVCodec、一个AVHWAccel以及一些其他编解码参数。
• AVCodec
  
  这个结构包是编码器/解码器的对外的封装格式，其对外接口包括：

void(* 	init_static_data )(struct AVCodec *codec)
 	Initialize codec static data, called from avcodec_register().

int(* 	init )(AVCodecContext *)

int(* 	encode_sub )(AVCodecContext *, uint8_t *buf, int buf_size, const struct AVSubtitle *sub)

int(* 	encode2 )(AVCodecContext *avctx, AVPacket *avpkt, const AVFrame *frame, int *got_packet_ptr)
 	Encode data to an AVPacket.

int(* 	decode )(AVCodecContext *, void *outdata, int *outdata_size, AVPacket *avpkt)

int(* 	close )(AVCodecContext *)

int(* 	send_frame )(AVCodecContext *avctx, const AVFrame *frame)
 	Decode/encode API with decoupled packet/frame dataflow.

int(* 	send_packet )(AVCodecContext *avctx, const AVPacket *avpkt)

int(* 	receive_frame )(AVCodecContext *avctx, AVFrame *frame)

int(* 	receive_packet )(AVCodecContext *avctx, AVPacket *avpkt)

void(* 	flush )(AVCodecContext *)
 	Flush buffers.

Frame-level threading support functions
int(* 	init_thread_copy )(AVCodecContext *)
 	If defined, called on thread contexts when they are created.

int(* 	update_thread_context )(AVCodecContext *dst, const AVCodecContext *src)
 	Copy necessary context variables from a previous thread context to the current one.

• AVHWAccel
  
  这里包含硬件解码的结构体，需要依赖特定硬件才可以正常运行。其统一接口如下：

int(* 	alloc_frame )(AVCodecContext *avctx, AVFrame *frame)
 	Allocate a custom buffer.

int(* 	start_frame )(AVCodecContext *avctx, const uint8_t *buf, uint32_t buf_size)
 	Called at the beginning of each frame or field picture.

int(* 	decode_slice )(AVCodecContext *avctx, const uint8_t *buf, uint32_t buf_size)
 	Callback for each slice.

int(* 	end_frame )(AVCodecContext *avctx)
 	Called at the end of each frame or field picture.

void(* 	decode_mb )(struct MpegEncContext *s)
 	Called for every Macroblock in a slice.

int(* 	init )(AVCodecContext *avctx)
 	Initialize the hwaccel private data.

int(* 	uninit )(AVCodecContext *avctx)
 	Uninitialize the hwaccel private data.

• AVCodecParserContext和AVCodecParser
  
  这是用于特定音频或视频的parser，比如h264、aac等，其统一对外接口如下：

int(* 	parser_init )(AVCodecParserContext *s)

int(* 	parser_parse )(AVCodecParserContext *s, AVCodecContext *avctx, const uint8_t **poutbuf, int *poutbuf_size, const uint8_t *buf, int buf_size)

void(* 	parser_close )(AVCodecParserContext *s)

int(* 	split )(AVCodecContext *avctx, const uint8_t *buf, int buf_size)

• AVFrame
  
  这里面存储了音频、视频解码之后的原始数据或者编码器的输入数据。其中存储的音视频数据具体格式需要参考AVFrame::format（音频格式是AVSampleFormat，视频格式是AVPixelFormat）。

我使用FFmpeg的libavcodec是decoder和parser，当然为了获取音视频原始数据，还需要了解必要的AVFrame结构。

通常decoder的调用逻辑是通过AVCodec的init初始化解码器，调用decode函数解码数据，调用close反初始化解码器，在需要的时候（比如解码结束，切换流）调用flush清空解码器内部缓存数据。

当然很多音视频比特流需要通过parser才送入解码器，这样就可以调用parser_parse实现。

关于硬解码的实现可能有很多细节，有兴趣的可以参考下Hardware acceleration introduction with FFmpeg。