2 * generic encoding-related code
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #include "libavutil/attributes.h"
22 #include "libavutil/avassert.h"
23 #include "libavutil/frame.h"
24 #include "libavutil/imgutils.h"
25 #include "libavutil/internal.h"
26 #include "libavutil/samplefmt.h"
30 #include "frame_thread_encoder.h"
33 int ff_alloc_packet2(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int64_t min_size)
35 if (size < 0 || size > INT_MAX - AV_INPUT_BUFFER_PADDING_SIZE) {
36 av_log(avctx, AV_LOG_ERROR, "Invalid minimum required packet size %"PRId64" (max allowed is %d)\n",
37 size, INT_MAX - AV_INPUT_BUFFER_PADDING_SIZE);
38 return AVERROR(EINVAL);
41 av_assert0(!avpkt->data);
43 if (avctx && 2*min_size < size) { // FIXME The factor needs to be finetuned
44 av_fast_padded_malloc(&avctx->internal->byte_buffer, &avctx->internal->byte_buffer_size, size);
45 avpkt->data = avctx->internal->byte_buffer;
50 int ret = av_new_packet(avpkt, size);
52 av_log(avctx, AV_LOG_ERROR, "Failed to allocate packet of size %"PRId64"\n", size);
59 int avcodec_default_get_encode_buffer(AVCodecContext *avctx, AVPacket *avpkt, int flags)
63 if (avpkt->size < 0 || avpkt->size > INT_MAX - AV_INPUT_BUFFER_PADDING_SIZE)
64 return AVERROR(EINVAL);
66 if (avpkt->data || avpkt->buf) {
67 av_log(avctx, AV_LOG_ERROR, "avpkt->{data,buf} != NULL in avcodec_default_get_encode_buffer()\n");
68 return AVERROR(EINVAL);
71 ret = av_buffer_realloc(&avpkt->buf, avpkt->size + AV_INPUT_BUFFER_PADDING_SIZE);
73 av_log(avctx, AV_LOG_ERROR, "Failed to allocate packet of size %d\n", avpkt->size);
76 avpkt->data = avpkt->buf->data;
77 memset(avpkt->data + avpkt->size, 0, AV_INPUT_BUFFER_PADDING_SIZE);
82 int ff_get_encode_buffer(AVCodecContext *avctx, AVPacket *avpkt, int64_t size, int flags)
86 if (size < 0 || size > INT_MAX - AV_INPUT_BUFFER_PADDING_SIZE)
87 return AVERROR(EINVAL);
89 av_assert0(!avpkt->data && !avpkt->buf);
92 ret = avctx->get_encode_buffer(avctx, avpkt, flags);
96 if (!avpkt->data || !avpkt->buf) {
97 av_log(avctx, AV_LOG_ERROR, "No buffer returned by get_encode_buffer()\n");
98 ret = AVERROR(EINVAL);
105 av_log(avctx, AV_LOG_ERROR, "get_encode_buffer() failed\n");
106 av_packet_unref(avpkt);
113 * Pad last frame with silence.
115 static int pad_last_frame(AVCodecContext *s, AVFrame *frame, const AVFrame *src)
119 frame->format = src->format;
120 frame->channel_layout = src->channel_layout;
121 frame->channels = src->channels;
122 frame->nb_samples = s->frame_size;
123 ret = av_frame_get_buffer(frame, 0);
127 ret = av_frame_copy_props(frame, src);
131 if ((ret = av_samples_copy(frame->extended_data, src->extended_data, 0, 0,
132 src->nb_samples, s->channels, s->sample_fmt)) < 0)
134 if ((ret = av_samples_set_silence(frame->extended_data, src->nb_samples,
135 frame->nb_samples - src->nb_samples,
136 s->channels, s->sample_fmt)) < 0)
142 av_frame_unref(frame);
146 int avcodec_encode_subtitle(AVCodecContext *avctx, uint8_t *buf, int buf_size,
147 const AVSubtitle *sub)
150 if (sub->start_display_time) {
151 av_log(avctx, AV_LOG_ERROR, "start_display_time must be 0.\n");
155 ret = avctx->codec->encode_sub(avctx, buf, buf_size, sub);
156 avctx->frame_number++;
160 int ff_encode_get_frame(AVCodecContext *avctx, AVFrame *frame)
162 AVCodecInternal *avci = avctx->internal;
167 if (!avci->buffer_frame->buf[0])
168 return AVERROR(EAGAIN);
170 av_frame_move_ref(frame, avci->buffer_frame);
175 static int encode_simple_internal(AVCodecContext *avctx, AVPacket *avpkt)
177 AVCodecInternal *avci = avctx->internal;
178 EncodeSimpleContext *es = &avci->es;
179 AVFrame *frame = es->in_frame;
183 if (avci->draining_done)
186 if (!frame->buf[0] && !avci->draining) {
187 av_frame_unref(frame);
188 ret = ff_encode_get_frame(avctx, frame);
189 if (ret < 0 && ret != AVERROR_EOF)
193 if (!frame->buf[0]) {
194 if (!(avctx->codec->capabilities & AV_CODEC_CAP_DELAY ||
195 (avci->frame_thread_encoder && avctx->active_thread_type & FF_THREAD_FRAME)))
198 // Flushing is signaled with a NULL frame
204 av_assert0(avctx->codec->encode2);
206 if (CONFIG_FRAME_THREAD_ENCODER &&
207 avci->frame_thread_encoder && (avctx->active_thread_type & FF_THREAD_FRAME))
208 /* This might modify frame, but it doesn't matter, because
209 * the frame properties used below are not used for video
210 * (due to the delay inherent in frame threaded encoding, it makes
211 * no sense to use the properties of the current frame anyway). */
212 ret = ff_thread_video_encode_frame(avctx, avpkt, frame, &got_packet);
214 ret = avctx->codec->encode2(avctx, avpkt, frame, &got_packet);
215 if (avctx->codec->type == AVMEDIA_TYPE_VIDEO && !ret && got_packet &&
216 !(avctx->codec->capabilities & AV_CODEC_CAP_DELAY))
217 avpkt->pts = avpkt->dts = frame->pts;
220 av_assert0(ret <= 0);
224 if (!ret && got_packet) {
226 ret = av_packet_make_refcounted(avpkt);
231 if (frame && !(avctx->codec->capabilities & AV_CODEC_CAP_DELAY)) {
232 if (avctx->codec->type == AVMEDIA_TYPE_AUDIO) {
233 if (avpkt->pts == AV_NOPTS_VALUE)
234 avpkt->pts = frame->pts;
235 if (!avpkt->duration)
236 avpkt->duration = ff_samples_to_time_base(avctx,
240 if (avctx->codec->type == AVMEDIA_TYPE_AUDIO) {
241 /* NOTE: if we add any audio encoders which output non-keyframe packets,
242 * this needs to be moved to the encoders, but for now we can do it
243 * here to simplify things */
244 avpkt->flags |= AV_PKT_FLAG_KEY;
245 avpkt->dts = avpkt->pts;
249 if (avci->draining && !got_packet)
250 avci->draining_done = 1;
253 if (ret < 0 || !got_packet)
254 av_packet_unref(avpkt);
258 avctx->frame_number++;
259 av_frame_unref(frame);
263 // Encoders must always return ref-counted buffers.
264 // Side-data only packets have no data and can be not ref-counted.
265 av_assert0(!avpkt->data || avpkt->buf);
270 static int encode_simple_receive_packet(AVCodecContext *avctx, AVPacket *avpkt)
274 while (!avpkt->data && !avpkt->side_data) {
275 ret = encode_simple_internal(avctx, avpkt);
283 static int encode_receive_packet_internal(AVCodecContext *avctx, AVPacket *avpkt)
285 AVCodecInternal *avci = avctx->internal;
288 if (avci->draining_done)
291 av_assert0(!avpkt->data && !avpkt->side_data);
293 if (avctx->codec->type == AVMEDIA_TYPE_VIDEO) {
294 if ((avctx->flags & AV_CODEC_FLAG_PASS1) && avctx->stats_out)
295 avctx->stats_out[0] = '\0';
296 if (av_image_check_size2(avctx->width, avctx->height, avctx->max_pixels, AV_PIX_FMT_NONE, 0, avctx))
297 return AVERROR(EINVAL);
300 if (avctx->codec->receive_packet) {
301 ret = avctx->codec->receive_packet(avctx, avpkt);
303 av_packet_unref(avpkt);
305 // Encoders must always return ref-counted buffers.
306 // Side-data only packets have no data and can be not ref-counted.
307 av_assert0(!avpkt->data || avpkt->buf);
309 ret = encode_simple_receive_packet(avctx, avpkt);
311 if (ret == AVERROR_EOF)
312 avci->draining_done = 1;
317 static int encode_send_frame_internal(AVCodecContext *avctx, const AVFrame *src)
319 AVCodecInternal *avci = avctx->internal;
320 AVFrame *dst = avci->buffer_frame;
323 if (avctx->codec->type == AVMEDIA_TYPE_AUDIO) {
324 /* extract audio service type metadata */
325 AVFrameSideData *sd = av_frame_get_side_data(src, AV_FRAME_DATA_AUDIO_SERVICE_TYPE);
326 if (sd && sd->size >= sizeof(enum AVAudioServiceType))
327 avctx->audio_service_type = *(enum AVAudioServiceType*)sd->data;
329 /* check for valid frame size */
330 if (avctx->codec->capabilities & AV_CODEC_CAP_SMALL_LAST_FRAME) {
331 if (src->nb_samples > avctx->frame_size) {
332 av_log(avctx, AV_LOG_ERROR, "more samples than frame size\n");
333 return AVERROR(EINVAL);
335 } else if (!(avctx->codec->capabilities & AV_CODEC_CAP_VARIABLE_FRAME_SIZE)) {
336 /* if we already got an undersized frame, that must have been the last */
337 if (avctx->internal->last_audio_frame) {
338 av_log(avctx, AV_LOG_ERROR, "frame_size (%d) was not respected for a non-last frame\n", avctx->frame_size);
339 return AVERROR(EINVAL);
342 if (src->nb_samples < avctx->frame_size) {
343 ret = pad_last_frame(avctx, dst, src);
347 avctx->internal->last_audio_frame = 1;
348 } else if (src->nb_samples > avctx->frame_size) {
349 av_log(avctx, AV_LOG_ERROR, "nb_samples (%d) != frame_size (%d)\n", src->nb_samples, avctx->frame_size);
350 return AVERROR(EINVAL);
356 ret = av_frame_ref(dst, src);
364 int attribute_align_arg avcodec_send_frame(AVCodecContext *avctx, const AVFrame *frame)
366 AVCodecInternal *avci = avctx->internal;
369 if (!avcodec_is_open(avctx) || !av_codec_is_encoder(avctx->codec))
370 return AVERROR(EINVAL);
375 if (avci->buffer_frame->data[0])
376 return AVERROR(EAGAIN);
381 ret = encode_send_frame_internal(avctx, frame);
386 if (!avci->buffer_pkt->data && !avci->buffer_pkt->side_data) {
387 ret = encode_receive_packet_internal(avctx, avci->buffer_pkt);
388 if (ret < 0 && ret != AVERROR(EAGAIN) && ret != AVERROR_EOF)
395 int attribute_align_arg avcodec_receive_packet(AVCodecContext *avctx, AVPacket *avpkt)
397 AVCodecInternal *avci = avctx->internal;
400 av_packet_unref(avpkt);
402 if (!avcodec_is_open(avctx) || !av_codec_is_encoder(avctx->codec))
403 return AVERROR(EINVAL);
405 if (avci->buffer_pkt->data || avci->buffer_pkt->side_data) {
406 av_packet_move_ref(avpkt, avci->buffer_pkt);
408 ret = encode_receive_packet_internal(avctx, avpkt);
416 #if FF_API_OLD_ENCDEC
417 static int compat_encode(AVCodecContext *avctx, AVPacket *avpkt,
418 int *got_packet, const AVFrame *frame)
420 AVCodecInternal *avci = avctx->internal;
426 if (frame && avctx->codec->type == AVMEDIA_TYPE_VIDEO) {
427 if (frame->format == AV_PIX_FMT_NONE)
428 av_log(avctx, AV_LOG_WARNING, "AVFrame.format is not set\n");
429 if (frame->width == 0 || frame->height == 0)
430 av_log(avctx, AV_LOG_WARNING, "AVFrame.width or height is not set\n");
433 if (avctx->codec->capabilities & AV_CODEC_CAP_DR1) {
434 av_log(avctx, AV_LOG_WARNING, "The deprecated avcodec_encode_* API does not support "
435 "AV_CODEC_CAP_DR1 encoders\n");
436 return AVERROR(ENOSYS);
439 ret = avcodec_send_frame(avctx, frame);
440 if (ret == AVERROR_EOF)
442 else if (ret == AVERROR(EAGAIN)) {
443 /* we fully drain all the output in each encode call, so this should not
449 av_packet_move_ref(&user_pkt, avpkt);
451 ret = avcodec_receive_packet(avctx, avpkt);
453 if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
458 if (avpkt != avci->compat_encode_packet) {
459 if (avpkt->data && user_pkt.data) {
460 if (user_pkt.size >= avpkt->size) {
461 memcpy(user_pkt.data, avpkt->data, avpkt->size);
462 av_buffer_unref(&avpkt->buf);
463 avpkt->buf = user_pkt.buf;
464 avpkt->data = user_pkt.data;
465 av_init_packet(&user_pkt);
467 av_log(avctx, AV_LOG_ERROR, "Provided packet is too small, needs to be %d\n", avpkt->size);
468 av_packet_unref(avpkt);
469 ret = AVERROR(EINVAL);
475 avpkt = avci->compat_encode_packet;
477 if (!avci->compat_decode_warned) {
478 av_log(avctx, AV_LOG_WARNING, "The deprecated avcodec_encode_* "
479 "API cannot return all the packets for this encoder. "
480 "Some packets will be dropped. Update your code to the "
481 "new encoding API to fix this.\n");
482 avci->compat_decode_warned = 1;
483 av_packet_unref(avpkt);
493 av_packet_unref(&user_pkt);
498 int attribute_align_arg avcodec_encode_audio2(AVCodecContext *avctx,
500 const AVFrame *frame,
503 int ret = compat_encode(avctx, avpkt, got_packet_ptr, frame);
506 av_packet_unref(avpkt);
511 int attribute_align_arg avcodec_encode_video2(AVCodecContext *avctx,
513 const AVFrame *frame,
516 int ret = compat_encode(avctx, avpkt, got_packet_ptr, frame);
519 av_packet_unref(avpkt);