/*****************************************************************************
* Copyright 2008. Pittsburgh Pattern Recognition, Inc.
*
* This file is part of the Frame Accurate Seeking extension library to
* ffmpeg (ffmpeg-fas).
*
* ffmpeg-fas is free software; you can redistribute it and/or modify it
* under the terms of the GNU Lesser General Public License as published by
* the Free Software Foundation; either version 3 of the License, or (at your
* option) any later version.
*
* The ffmpeg-fas library is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
* or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
* License for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with the ffmpeg-fas library. If not, see .
*
******************************************************************************/
#include "ffmpeg_fas.h"
#include "libavformat/avformat.h"
#include "libavcodec/avcodec.h"
#include "seek_indices.h"
#include "private_errors.h"
#include
#include
#include
#include
#define FIRST_FRAME_INDEX 0
#define NUM_POSSIBLE_ERRORS 9
/**** Private Types ***********************************************************/
typedef struct fas_context_struct {
fas_boolean_type is_video_active;
fas_boolean_type is_frame_available;
int current_frame_index;
seek_table_type seek_table;
/* ffmpeg */
AVFormatContext *format_context;
AVCodecContext *codec_context;
int stream_idx;
AVFrame *frame_buffer; // internal buffer
AVFrame *rgb_frame_buffer; // extra AVFrames (for color conversion)
uint8_t *rgb_buffer; // actual data buffer for rgb_frame_buffer (needs to be freed seperately)
fas_boolean_type rgb_already_converted; // color_convert(frame_buffer) == rgb_frame_buffer (so we don't need to repeat conversions)
AVFrame *gray8_frame_buffer;
uint8_t *gray8_buffer;
fas_boolean_type gray8_already_converted;
int64_t current_dts; // decoding timestamp of the most recently parsed packet
int64_t previous_dts; // for previous packet (always use previous packet for seek_table (workaround))
int64_t keyframe_packet_dts; // dts of most recent keyframe packet
int64_t first_dts; // for very first packet (needed in seek, for first keyframe)
} fas_context_type;
static char* invalid_error_code = "not a valid error code";
static char *gbl_error_strings[NUM_POSSIBLE_ERRORS] =
{
"fas_success",
"fas_failure",
"fas_invalid_argument",
"fas_out_of_memory",
"fas_unsupported_format",
"fas_unsupported_codec",
"fas_no_more_frames",
"fas_decoding_error",
"fas_seek_error"
};
char* fas_error_message(fas_error_type error)
{
if ((error < 0) || (error >= NUM_POSSIBLE_ERRORS))
return invalid_error_code;
return gbl_error_strings[error];
}
static void private_show_warning (const char *message);
static fas_error_type private_show_error (const char *message, fas_error_type error);
static fas_error_type private_convert_to_rgb (fas_context_ref_type ctx);
static fas_error_type private_seek_to_nearest_key (fas_context_ref_type context, int target_index, int offset);
fas_error_type private_complete_seek_table (fas_context_ref_type context);
void fas_set_logging (fas_boolean_type logging)
{
if (logging == FAS_TRUE)
{
SHOW_ERROR_MESSAGES = 1;
SHOW_WARNING_MESSAGES = 1;
av_log_level = AV_LOG_INFO;
}
else
{
SHOW_ERROR_MESSAGES = 0;
SHOW_WARNING_MESSAGES = 0;
av_log_level = AV_LOG_QUIET;
}
}
void fas_initialize (fas_boolean_type logging)
{
fas_set_logging(logging);
av_register_all();
return;
}
/* fas_open_video */
fas_error_type fas_open_video (fas_context_ref_type *context_ptr, char *file_path)
{
if (NULL == context_ptr)
return private_show_error ("NULL context pointer provided", FAS_INVALID_ARGUMENT);
// seek_error_type seek_error;
fas_context_ref_type fas_context;
*context_ptr = NULL; // set returned context to NULL in case of error
fas_context = (fas_context_ref_type)malloc (sizeof (fas_context_type));
memset(fas_context, 0, sizeof(fas_context_type));
if (NULL == fas_context)
return private_show_error ("unable to allocate buffer", FAS_OUT_OF_MEMORY);
fas_context->is_video_active = FAS_TRUE;
fas_context->is_frame_available = FAS_TRUE;
fas_context->current_frame_index = FIRST_FRAME_INDEX - 1;
fas_context->current_dts = AV_NOPTS_VALUE;
fas_context->previous_dts = AV_NOPTS_VALUE;
fas_context->keyframe_packet_dts = AV_NOPTS_VALUE;
fas_context->first_dts = AV_NOPTS_VALUE;
fas_context->seek_table = seek_init_table (-1); /* default starting size */
if (av_open_input_file ( &(fas_context->format_context), file_path, NULL, 0, NULL ) != 0)
{
fas_close_video(fas_context);
return private_show_error ("failure to open file", FAS_UNSUPPORTED_FORMAT);
}
if (av_find_stream_info (fas_context->format_context) < 0)
{
fas_close_video(fas_context);
return private_show_error ("could not extract stream information", FAS_UNSUPPORTED_FORMAT);
}
if (SHOW_WARNING_MESSAGES)
dump_format(fas_context->format_context, 0, file_path, 0);
int stream_idx;
for (stream_idx = 0; stream_idx < fas_context->format_context->nb_streams; stream_idx++)
{
if (fas_context->format_context->streams[stream_idx]->codec->codec_type == CODEC_TYPE_VIDEO)
{
fas_context->stream_idx = stream_idx;
fas_context->codec_context = fas_context->format_context->streams[stream_idx]->codec;
break;
}
}
if (fas_context->codec_context == 0)
{
fas_close_video(fas_context);
return private_show_error ("failure to find a video stream", FAS_UNSUPPORTED_FORMAT);
}
AVCodec *codec = avcodec_find_decoder (fas_context->codec_context->codec_id);
if (!codec)
{
fas_context->codec_context = 0;
fas_close_video(fas_context);
return private_show_error("failed to find correct video codec", FAS_UNSUPPORTED_CODEC);
}
if (avcodec_open (fas_context->codec_context, codec) < 0)
{
fas_context->codec_context = 0;
fas_close_video(fas_context);
return private_show_error ("failed to open codec", FAS_UNSUPPORTED_CODEC);
}
fas_context->frame_buffer = avcodec_alloc_frame ();
if (fas_context->frame_buffer == NULL)
{
fas_close_video(fas_context);
return private_show_error ("failed to allocate frame buffer", FAS_OUT_OF_MEMORY);
}
fas_context->rgb_frame_buffer = avcodec_alloc_frame ();
if (fas_context->rgb_frame_buffer == NULL)
{
fas_close_video(fas_context);
return private_show_error ("failed to allocate rgb frame buffer", FAS_OUT_OF_MEMORY);
}
fas_context->gray8_frame_buffer = avcodec_alloc_frame ();
if (fas_context->gray8_frame_buffer == NULL)
{
fas_close_video(fas_context);
return private_show_error ("failed to allocate gray8 frame buffer", FAS_OUT_OF_MEMORY);
}
fas_context->rgb_buffer = 0;
fas_context->gray8_buffer = 0;
fas_context->rgb_already_converted = FAS_FALSE;
fas_context->gray8_already_converted = FAS_FALSE;
*context_ptr = fas_context;
if (FAS_SUCCESS != fas_step_forward(*context_ptr))
return private_show_error ("failure decoding first frame", FAS_NO_MORE_FRAMES);
if (!fas_frame_available(*context_ptr))
return private_show_error ("couldn't find a first frame (no valid frames in video stream)", FAS_NO_MORE_FRAMES);
return FAS_SUCCESS;
}
/* fas_close_video */
fas_error_type fas_close_video (fas_context_ref_type context)
{
if (NULL == context)
return private_show_error ("NULL context provided for fas_close_video()", FAS_INVALID_ARGUMENT);
if (!(context->is_video_active))
{
private_show_warning ("Redundant attempt to close an inactive video");
return FAS_SUCCESS;
}
if (context->codec_context)
if (avcodec_find_decoder (context->codec_context->codec_id))
avcodec_close(context->codec_context);
if (context->format_context)
av_close_input_file (context->format_context);
if (context->rgb_frame_buffer)
av_free (context->rgb_frame_buffer);
if (context->gray8_frame_buffer)
av_free (context->gray8_frame_buffer);
if (context->rgb_buffer)
av_free(context->rgb_buffer);
if (context->gray8_buffer)
av_free(context->gray8_buffer);
if (context->frame_buffer)
av_free (context->frame_buffer);
seek_release_table (&(context->seek_table));
context->is_video_active = FAS_FALSE;
free (context);
return FAS_SUCCESS;
}
/* fas_step_forward */
fas_error_type fas_step_forward (fas_context_ref_type context)
{
if ((NULL == context) || (FAS_TRUE != context->is_video_active)) {
return private_show_error ("invalid or unopened context", FAS_INVALID_ARGUMENT);
}
if (!context->is_frame_available)
{
private_show_warning ("tried to advance after end of frames");
return FAS_SUCCESS;
}
context->current_frame_index++;
AVPacket packet;
while (FAS_TRUE)
{
if (av_read_frame(context->format_context, &packet) < 0)
{
/* finished */
context->is_frame_available = FAS_FALSE;
context->seek_table.completed = seek_true;
return FAS_SUCCESS;
}
int frameFinished;
if (packet.stream_index == context->stream_idx)
{
context->previous_dts = context->current_dts;
context->current_dts = packet.dts;
/* seek support: set first_dts */
if (context->first_dts == AV_NOPTS_VALUE)
context->first_dts = packet.dts;
/* seek support: set key-packet info to previous packet's dts, when possible */
/* note this -1 approach to setting the packet is a workaround for a common failure. setting
to 0 would work just incur a huge penalty in videos that needed -1. Might be worth testing.
*/
if (packet.flags & PKT_FLAG_KEY)
{
//fprintf(stderr, "Packet: (F:%d %lld %lld)\n", context->current_frame_index, packet.pts, packet.dts);
if (context->previous_dts == AV_NOPTS_VALUE)
context->keyframe_packet_dts = packet.dts;
else
context->keyframe_packet_dts = context->previous_dts;
}
avcodec_decode_video(context->codec_context, context->frame_buffer, &frameFinished,
packet.data, packet.size);
if (frameFinished)
{
/* seek support: (try to) add entry to seek_table */
if (context->frame_buffer->key_frame)
{
// fprintf(stderr, "Frame : (PXX F%d: %lld %lld)\n", context->current_frame_index, packet.pts, packet.dts);
seek_entry_type entry;
entry.display_index = context->current_frame_index;
entry.first_packet_dts = context->keyframe_packet_dts;
entry.last_packet_dts = packet.dts;
if (fas_get_frame_index(context) == FIRST_FRAME_INDEX)
entry.first_packet_dts = context->first_dts;
seek_append_table_entry(&context->seek_table, entry);
}
if (context->current_frame_index - FIRST_FRAME_INDEX + 1 > context->seek_table.num_frames)
context->seek_table.num_frames = context->current_frame_index - FIRST_FRAME_INDEX + 1;
break;
}
}
av_free_packet(&packet);
}
context->rgb_already_converted = FAS_FALSE;
context->gray8_already_converted = FAS_FALSE;
av_free_packet(&packet);
return FAS_SUCCESS;
}
/* fas_get_frame_index */
int fas_get_frame_index (fas_context_ref_type context)
{
if (NULL == context)
return private_show_error ("NULL context provided for fas_get_frame_index()", FAS_INVALID_ARGUMENT);
if (FAS_TRUE != context->is_video_active)
return private_show_error ("No video is open for fas_get_frame_index()", FAS_INVALID_ARGUMENT);
return context->current_frame_index;
}
/* fas_get_frame */
fas_error_type fas_get_frame (fas_context_ref_type context, fas_raw_image_type *image_ptr)
{
int buffer_size;
fas_error_type fas_error;
if (NULL == context || FAS_FALSE == context->is_video_active)
return private_show_error ("null context or inactive video", FAS_INVALID_ARGUMENT);
if (NULL == image_ptr)
return private_show_error ("null image_ptr on get_frame", FAS_INVALID_ARGUMENT);
if (!fas_frame_available(context))
return private_show_error ("no frame available for extraction", FAS_NO_MORE_FRAMES);
memset (image_ptr, 0, sizeof (fas_raw_image_type));
buffer_size = context->codec_context->width * context->codec_context->height * 3;
image_ptr->data = (unsigned char *)malloc (buffer_size);
if (NULL == image_ptr->data)
return private_show_error ("unable to allocate space for RGB image", FAS_OUT_OF_MEMORY);
image_ptr->color_space = FAS_RGB24;
image_ptr->width = context->codec_context->width;
image_ptr->height = context->codec_context->height;
image_ptr->bytes_per_line = context->codec_context->width * 3;
fas_error = private_convert_to_rgb(context);
int j;
unsigned char *from;
unsigned char *to;
for (j=0;jcodec_context->height; j++)
{
from = context->rgb_frame_buffer->data[0] + j*context->rgb_frame_buffer->linesize[0];
to = image_ptr->data + j*image_ptr->bytes_per_line;
memcpy(to, from, context->codec_context->width *3);
}
if (FAS_SUCCESS != fas_error)
return private_show_error ("unable to convert image to RGB", FAS_FAILURE);
return FAS_SUCCESS;
}
/* fas_free_frame */
void fas_free_frame (fas_raw_image_type image)
{
if (NULL == image.data)
return;
free (image.data);
return;
}
/* fas_get_seek_table */
seek_table_type fas_get_seek_table (fas_context_ref_type context)
{
seek_table_type null_table;
null_table.array = NULL;
null_table.completed = seek_false;
null_table.num_frames = -1;
null_table.num_entries = 0;
null_table.allocated_size = 0;
if (NULL == context || FAS_FALSE == context->is_video_active)
return null_table;
return context->seek_table;
}
/* fas_put_seek_table */
fas_error_type fas_put_seek_table (fas_context_ref_type context, seek_table_type table)
{
if (NULL == context || FAS_FALSE == context->is_video_active)
return private_show_error ("null context or inactive video", FAS_INVALID_ARGUMENT);
seek_release_table (&context->seek_table);
context->seek_table = seek_copy_table(table);
return FAS_SUCCESS;
}
/* private_complete_seek_table */
fas_error_type private_complete_seek_table (fas_context_ref_type context)
{
if ((NULL == context) || (FAS_FALSE == context->is_video_active))
return private_show_error ("invalid or unopened context", FAS_INVALID_ARGUMENT);
if (context->seek_table.completed)
return FAS_SUCCESS;
fas_error_type fas_error = fas_seek_to_nearest_key (context, context->seek_table.num_frames + FIRST_FRAME_INDEX - 1);
if (FAS_SUCCESS != fas_error)
return private_show_error("failed when trying to complete seek table (1) (first frame not labeled keyframe?)", fas_error);
while (fas_frame_available(context))
{
// printf("%d\n", context->seek_table.num_frames);
fas_step_forward(context);
}
if (!context->seek_table.completed)
return private_show_error("failed when trying to complete seek table (2)", FAS_SEEK_ERROR);
return FAS_SUCCESS;
}
/* fas_seek_to_frame */
fas_error_type fas_seek_to_frame (fas_context_ref_type context, int target_index)
{
fas_error_type fas_error;
if ((NULL == context) || (FAS_FALSE == context->is_video_active))
return private_show_error ("invalid or unopened context", FAS_INVALID_ARGUMENT);
// printf("seeking to %d (from %d)!\n", target_index, context->current_frame_index);
if (target_index == context->current_frame_index)
return FAS_SUCCESS;
fas_error = fas_seek_to_nearest_key (context, target_index);
if (fas_error != FAS_SUCCESS)
return private_show_error ("error advancing to key frame before seek", fas_error);
if (fas_get_frame_index(context) > target_index)
return private_show_error ("error advancing to key frame before seek (index isn't right)", fas_error);
while (fas_get_frame_index(context) < target_index)
{
if (fas_frame_available(context))
fas_step_forward(context);
else
return private_show_error ("error advancing to request frame (probably out of range)", FAS_SEEK_ERROR);
}
return FAS_SUCCESS;
}
/* fas_seek_to_nearest_key */
fas_error_type fas_seek_to_nearest_key (fas_context_ref_type context, int target_index)
{
return private_seek_to_nearest_key(context, target_index,0);
}
/* private_seek_to_nearest_key */
fas_error_type private_seek_to_nearest_key (fas_context_ref_type context, int target_index, int offset)
{
if ((NULL == context) || (FAS_TRUE != context->is_video_active))
return private_show_error ("invalid or unopened context", FAS_INVALID_ARGUMENT);
// printf("HERE: from: %d to: %d offset: %d\n", context->current_frame_index, target_index, offset);
fas_error_type fas_error;
seek_entry_type seek_entry;
seek_error_type seek_error = seek_get_nearest_entry (&(context->seek_table), &seek_entry, target_index, offset);
if (seek_error != seek_no_error)
return private_show_error ("error while searching seek table", FAS_SEEK_ERROR);
if (seek_entry.display_index == context->current_frame_index)
return FAS_SUCCESS;
// printf("HERE: from: %d to: %d (%d) offset: %d\n", context->current_frame_index, target_index, seek_entry.display_index, offset);
// printf("trying to seek to %d (%lld->%lld)\n", seek_entry.display_index, seek_entry.first_packet_dts, seek_entry.last_packet_dts);
// if something goes terribly wrong, return bad current_frame_index
context->current_frame_index = -2;
context->is_frame_available = FAS_TRUE;
int flags = 0;
if (seek_entry.first_packet_dts <= context->current_dts)
flags = AVSEEK_FLAG_BACKWARD;
// printf("av_seek_frame: %lld\n", seek_entry.first_packet_dts);
if (av_seek_frame(context->format_context, context->stream_idx, seek_entry.first_packet_dts, flags) < 0)
return private_show_error("seek to keyframe failed", FAS_SEEK_ERROR);
avcodec_flush_buffers (context->codec_context);
fas_error = fas_step_forward (context);
if (fas_error != FAS_SUCCESS || !context->is_frame_available)
{
// something bad has happened, try previous keyframe
private_show_warning("processing of seeked keyframe failed, trying previous keyframe");
return private_seek_to_nearest_key(context, target_index, offset + 1);
}
while (context->current_dts < seek_entry.last_packet_dts)
{
//printf("frame-times: current: %lld target: %lld is_key: %d\n", context->current_dts, seek_entry.last_packet_dts, context->frame_buffer->key_frame);
fas_error = fas_step_forward(context);
if (fas_error != FAS_SUCCESS)
return private_show_error ("unable to process up to target frame (fas_seek_to_frame)", fas_error);
}
// printf("keyframe vitals: %d looking_for: %lld at: %lld\n", seek_entry.display_index, seek_entry.last_packet_dts, context->current_dts);
if (context->current_dts != seek_entry.last_packet_dts)
{
/* seek to last key-frame, but look for this one */
private_show_warning("missed keyframe, trying previous keyframe");
return private_seek_to_nearest_key(context, target_index, offset + 1);
}
/* Ideally, we could just check if the frame decoded is of the correct time stamp... but... we need several ugly workarounds:
1) Some videos have bad keyframes that don't get decoded properly. In this cases, we need to go back a keyframe.
2) Other times, none of the frames are labeled keyframes. In these cases, we need to allow seeking to frame 0
even when it's not labeled as a keyframe. Messy set of conditions.
*/
if ((!context->frame_buffer->key_frame) && (seek_entry.display_index != 0))
{
private_show_warning("found keyframe, but not labeled as keyframe, so trying previous keyframe.");
/* seek & look for previous keyframe */
/* REMOVE FROM TABLE? */
return private_seek_to_nearest_key(context, seek_entry.display_index - 1, 0);
}
context->current_frame_index = seek_entry.display_index;
return FAS_SUCCESS;
}
/* fas_get_frame_count */
int fas_get_frame_count_fast (fas_context_ref_type context)
{
if (NULL == context || FAS_FALSE == context->is_video_active)
{
private_show_error ("NULL or invalid context", FAS_INVALID_ARGUMENT);
return -1;
}
if (context->seek_table.completed == seek_true)
return context->seek_table.num_frames;
return -1;
}
int fas_get_frame_count (fas_context_ref_type context)
{
int fast = fas_get_frame_count_fast(context);
if (fast >= 0)
return fast;
int current_frame = fas_get_frame_index(context);
fas_error_type fas_error;
fas_error = private_complete_seek_table(context);
if (FAS_SUCCESS != fas_error)
{
private_show_error("failed in get_frame_count trying to complete the seek table", fas_error);
return -1;
}
// seek_show_raw_table(stderr, context->seek_table);
fas_error = fas_seek_to_frame(context, current_frame);
if (FAS_SUCCESS != fas_error)
{
private_show_error("failed in get_frame_count when trying to seek back to original location", fas_error);
return -1;
}
fast = fas_get_frame_count_fast(context);
if (fast < 0)
private_show_warning("get_frame_count failed");
return fast;
}
/* fas_frame_available */
fas_boolean_type fas_frame_available (fas_context_ref_type context)
{
if (NULL == context)
{
private_show_error ("NULL context provided for fas_get_frame_index()", FAS_INVALID_ARGUMENT);
return FAS_FALSE;
}
if (!context->is_video_active)
return FAS_FALSE;
return context->is_frame_available;
}
/* private_show_error */
static fas_error_type private_show_error (const char *message, fas_error_type error)
{
if (SHOW_ERROR_MESSAGES)
fprintf (stderr, " ===> ffmpeg_fas: %s\n", message);
return error;
}
static void private_show_warning (const char *message)
{
if (SHOW_WARNING_MESSAGES)
fprintf (stderr, " ---- ffmpeg_fas: %s\n", message);
return;
}
/* private_convert_to_rgb */
fas_error_type private_convert_to_rgb (fas_context_ref_type ctx)
{
if (ctx->rgb_already_converted)
return FAS_SUCCESS;
if (ctx->rgb_buffer == 0)
{
int numBytes = avpicture_get_size(PIX_FMT_RGB24, ctx->codec_context->width,
ctx->codec_context->height);
ctx->rgb_buffer = (uint8_t *) av_malloc(numBytes*sizeof(uint8_t));
avpicture_fill((AVPicture *) ctx->rgb_frame_buffer, ctx->rgb_buffer, PIX_FMT_RGB24,
ctx->codec_context->width, ctx->codec_context->height);
}
if (img_convert((AVPicture *) ctx->rgb_frame_buffer, PIX_FMT_RGB24, (AVPicture *) ctx->frame_buffer,
ctx->codec_context->pix_fmt,
ctx->codec_context->width, ctx->codec_context->height) < 0)
private_show_error("error converting to rgb", FAS_DECODING_ERROR);
ctx->rgb_already_converted = FAS_TRUE;
return FAS_SUCCESS;
}
/* private_convert_to_gray8 */
fas_error_type private_convert_to_gray8 (fas_context_ref_type ctx)
{
if (ctx->gray8_already_converted)
return FAS_SUCCESS;
if (ctx->gray8_buffer == 0)
{
int numBytes = avpicture_get_size(PIX_FMT_GRAY8, ctx->codec_context->width,
ctx->codec_context->height);
ctx->gray8_buffer = (uint8_t *) av_malloc(numBytes*sizeof(uint8_t));
avpicture_fill((AVPicture *) ctx->gray8_frame_buffer, ctx->gray8_buffer, PIX_FMT_GRAY8,
ctx->codec_context->width, ctx->codec_context->height);
}
if (img_convert((AVPicture *) ctx->gray8_frame_buffer, PIX_FMT_GRAY8, (AVPicture *) ctx->frame_buffer,
ctx->codec_context->pix_fmt,
ctx->codec_context->width, ctx->codec_context->height) < 0)
private_show_error("error converting to gray8", FAS_DECODING_ERROR);
ctx->gray8_already_converted = FAS_TRUE;
return FAS_SUCCESS;
}
int fas_get_current_width(fas_context_ref_type context)
{
return context->codec_context->width;
}
int fas_get_current_height(fas_context_ref_type context)
{
return context->codec_context->height;
}
fas_error_type fas_fill_gray8_ptr(fas_context_ref_type context, unsigned char *y)
{
/* this conversion also seems to screw up sometimes -- pal8 -> gray8? legodragon.avi */
if (private_convert_to_gray8(context) != FAS_SUCCESS)
return FAS_FAILURE;
int width = context->codec_context->width;
int height = context->codec_context->height;
int i;
for (i=0;i < height; i++)
memcpy(y + width * i, context->gray8_frame_buffer->data[0] + context->gray8_frame_buffer->linesize[0] * i, width);
return FAS_SUCCESS;
}
fas_error_type fas_fill_420p_ptrs (fas_context_ref_type context, unsigned char *y, unsigned char *u, unsigned char *v)
{
AVFrame *p = context->frame_buffer;
/* 411p to 420p conversion fails!? ... so i left this -ldb */
if (context->codec_context->pix_fmt != PIX_FMT_YUV420P)
return FAS_FAILURE;
int width = context->codec_context->width;
int height = context->codec_context->height;
int i;
for (i=0;i < height / 2; i++)
{
memcpy(y + width * (2*i) , p->data[0] + p->linesize[0] * (2*i) , width);
memcpy(y + width * (2*i + 1), p->data[0] + p->linesize[0] * (2*i + 1), width);
memcpy(u + width / 2 * i, p->data[1] + p->linesize[1] * i, width / 2);
memcpy(v + width / 2 * i, p->data[2] + p->linesize[2] * i, width / 2);
}
return FAS_SUCCESS;
}