FFmpeg: libavcodec/libtheoraenc.c Source File

00001 /*
00002  * Copyright (c) 2006 Paul Richards <paul.richards@gmail.com>
00003  *
00004  * This file is part of FFmpeg.
00005  *
00006  * FFmpeg is free software; you can redistribute it and/or
00007  * modify it under the terms of the GNU Lesser General Public
00008  * License as published by the Free Software Foundation; either
00009  * version 2.1 of the License, or (at your option) any later version.
00010  *
00011  * FFmpeg is distributed in the hope that it will be useful,
00012  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00013  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00014  * Lesser General Public License for more details.
00015  *
00016  * You should have received a copy of the GNU Lesser General Public
00017  * License along with FFmpeg; if not, write to the Free Software
00018  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00019  */
00020 
00033 /* FFmpeg includes */
00034 #include "libavutil/intreadwrite.h"
00035 #include "libavutil/log.h"
00036 #include "avcodec.h"
00037 
00038 /* libtheora includes */
00039 #include <theora/theora.h>
00040 
00041 typedef struct TheoraContext{
00042     theora_state t_state;
00043 } TheoraContext;
00044 
00048 static int concatenate_packet(unsigned int* offset, AVCodecContext* avc_context, const ogg_packet* packet)
00049 {
00050     char* message = NULL;
00051     uint8_t* newdata = NULL;
00052     int newsize = avc_context->extradata_size + 2 + packet->bytes;
00053 
00054     if (packet->bytes < 0) {
00055         message = "ogg_packet has negative size";
00056     } else if (packet->bytes > 0xffff) {
00057         message = "ogg_packet is larger than 65535 bytes";
00058     } else if (newsize < avc_context->extradata_size) {
00059         message = "extradata_size would overflow";
00060     } else {
00061         newdata = av_realloc(avc_context->extradata, newsize);
00062         if (newdata == NULL) {
00063             message = "av_realloc failed";
00064         }
00065     }
00066     if (message != NULL) {
00067         av_log(avc_context, AV_LOG_ERROR, "concatenate_packet failed: %s\n", message);
00068         return -1;
00069     }
00070 
00071     avc_context->extradata = newdata;
00072     avc_context->extradata_size = newsize;
00073     AV_WB16(avc_context->extradata + (*offset), packet->bytes);
00074     *offset += 2;
00075     memcpy( avc_context->extradata + (*offset), packet->packet, packet->bytes );
00076     (*offset) += packet->bytes;
00077     return 0;
00078 }
00079 
00080 static av_cold int encode_init(AVCodecContext* avc_context)
00081 {
00082     theora_info t_info;
00083     theora_comment t_comment;
00084     ogg_packet o_packet;
00085     unsigned int offset;
00086     TheoraContext *h = avc_context->priv_data;
00087 
00088     /* Set up the theora_info struct */
00089     theora_info_init( &t_info );
00090     t_info.width = avc_context->width;
00091     t_info.height = avc_context->height;
00092     t_info.frame_width = avc_context->width;
00093     t_info.frame_height = avc_context->height;
00094     t_info.offset_x = 0;
00095     t_info.offset_y = 0;
00096     /* Swap numerator and denominator as time_base in AVCodecContext gives the
00097      * time period between frames, but theora_info needs the framerate.  */
00098     t_info.fps_numerator = avc_context->time_base.den;
00099     t_info.fps_denominator = avc_context->time_base.num;
00100     if (avc_context->sample_aspect_ratio.num != 0) {
00101         t_info.aspect_numerator = avc_context->sample_aspect_ratio.num;
00102         t_info.aspect_denominator = avc_context->sample_aspect_ratio.den;
00103     } else {
00104         t_info.aspect_numerator = 1;
00105         t_info.aspect_denominator = 1;
00106     }
00107     t_info.colorspace = OC_CS_UNSPECIFIED;
00108     t_info.pixelformat = OC_PF_420;
00109     t_info.target_bitrate = avc_context->bit_rate;
00110     t_info.keyframe_frequency = avc_context->gop_size;
00111     t_info.keyframe_frequency_force = avc_context->gop_size;
00112     t_info.keyframe_mindistance = avc_context->keyint_min;
00113     t_info.quality = 0;
00114 
00115     t_info.quick_p = 1;
00116     t_info.dropframes_p = 0;
00117     t_info.keyframe_auto_p = 1;
00118     t_info.keyframe_data_target_bitrate = t_info.target_bitrate * 1.5;
00119     t_info.keyframe_auto_threshold = 80;
00120     t_info.noise_sensitivity = 1;
00121     t_info.sharpness = 0;
00122 
00123     /* Now initialise libtheora */
00124     if (theora_encode_init( &(h->t_state), &t_info ) != 0) {
00125         av_log(avc_context, AV_LOG_ERROR, "theora_encode_init failed\n");
00126         return -1;
00127     }
00128 
00129     /* Clear up theora_info struct */
00130     theora_info_clear( &t_info );
00131 
00132     /*
00133         Output first header packet consisting of theora
00134         header, comment, and tables.
00135 
00136         Each one is prefixed with a 16bit size, then they
00137         are concatenated together into ffmpeg's extradata.
00138     */
00139     offset = 0;
00140 
00141     /* Header */
00142     theora_encode_header( &(h->t_state), &o_packet );
00143     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
00144         return -1;
00145     }
00146 
00147     /* Comment */
00148     theora_comment_init( &t_comment );
00149     theora_encode_comment( &t_comment, &o_packet );
00150     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
00151         return -1;
00152     }
00153 
00154     /* Tables */
00155     theora_encode_tables( &(h->t_state), &o_packet );
00156     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
00157         return -1;
00158     }
00159 
00160     /* Clear up theora_comment struct */
00161     theora_comment_clear( &t_comment );
00162 
00163     /* Set up the output AVFrame */
00164     avc_context->coded_frame= avcodec_alloc_frame();
00165 
00166     return 0;
00167 }
00168 
00169 static int encode_frame(
00170     AVCodecContext* avc_context,
00171     uint8_t *outbuf,
00172     int buf_size,
00173     void *data)
00174 {
00175     yuv_buffer t_yuv_buffer;
00176     TheoraContext *h = avc_context->priv_data;
00177     AVFrame *frame = data;
00178     ogg_packet o_packet;
00179     int result;
00180 
00181     assert(avc_context->pix_fmt == PIX_FMT_YUV420P);
00182 
00183     /* Copy planes to the theora yuv_buffer */
00184     if (frame->linesize[1] != frame->linesize[2]) {
00185         av_log(avc_context, AV_LOG_ERROR, "U and V stride differ\n");
00186         return -1;
00187     }
00188 
00189     t_yuv_buffer.y_width = avc_context->width;
00190     t_yuv_buffer.y_height = avc_context->height;
00191     t_yuv_buffer.y_stride = frame->linesize[0];
00192     t_yuv_buffer.uv_width = t_yuv_buffer.y_width / 2;
00193     t_yuv_buffer.uv_height = t_yuv_buffer.y_height / 2;
00194     t_yuv_buffer.uv_stride = frame->linesize[1];
00195 
00196     t_yuv_buffer.y = frame->data[0];
00197     t_yuv_buffer.u = frame->data[1];
00198     t_yuv_buffer.v = frame->data[2];
00199 
00200     /* Now call into theora_encode_YUVin */
00201     result = theora_encode_YUVin( &(h->t_state), &t_yuv_buffer );
00202     if (result != 0) {
00203         const char* message;
00204         switch (result) {
00205             case -1:
00206                 message = "differing frame sizes";
00207                 break;
00208             case OC_EINVAL:
00209                 message = "encoder is not ready or is finished";
00210                 break;
00211             default:
00212                 message = "unknown reason";
00213                 break;
00214         }
00215         av_log(avc_context, AV_LOG_ERROR, "theora_encode_YUVin failed (%s) [%d]\n", message, result);
00216         return -1;
00217     }
00218 
00219     /* Pick up returned ogg_packet */
00220     result = theora_encode_packetout( &(h->t_state), 0, &o_packet );
00221     switch (result) {
00222         case 0:
00223             /* No packet is ready */
00224             return 0;
00225         case 1:
00226             /* Success, we have a packet */
00227             break;
00228         default:
00229             av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed [%d]\n", result);
00230             return -1;
00231     }
00232 
00233     /* Copy ogg_packet content out to buffer */
00234     if (buf_size < o_packet.bytes) {
00235         av_log(avc_context, AV_LOG_ERROR, "encoded frame too large\n");
00236         return -1;
00237     }
00238     memcpy(outbuf, o_packet.packet, o_packet.bytes);
00239 
00240     return o_packet.bytes;
00241 }
00242 
00243 static av_cold int encode_close(AVCodecContext* avc_context)
00244 {
00245     ogg_packet o_packet;
00246     TheoraContext *h = avc_context->priv_data;
00247     int result;
00248     const char* message;
00249 
00250     result = theora_encode_packetout( &(h->t_state), 1, &o_packet );
00251     theora_clear( &(h->t_state) );
00252     switch (result) {
00253         case 0:/* No packet is ready */
00254         case -1:/* Encoding finished */
00255             return 0;
00256         case 1:
00257             /* We have a packet */
00258             message = "gave us a packet";
00259             break;
00260         default:
00261             message = "unknown reason";
00262             break;
00263     }
00264     av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed (%s) [%d]\n", message, result);
00265     return -1;
00266 }
00267 
00268 static const enum PixelFormat supported_pixel_formats[] = { PIX_FMT_YUV420P, PIX_FMT_NONE };
00269 
00271 AVCodec libtheora_encoder =
00272 {
00273     .name = "libtheora",
00274     .type = CODEC_TYPE_VIDEO,
00275     .id = CODEC_ID_THEORA,
00276     .priv_data_size = sizeof(TheoraContext),
00277     .init = encode_init,
00278     .close = encode_close,
00279     .encode = encode_frame,
00280     .pix_fmts = supported_pixel_formats,
00281     .long_name = NULL_IF_CONFIG_SMALL("libtheora Theora"),
00282 };