2 * RTP parser for VP9 payload format (draft version 0) - experimental
3 * Copyright (c) 2015 Thomas Volkert <thomas@homer-conferencing.com>
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "libavcodec/bytestream.h"
25 #include "rtpdec_formats.h"
27 #define RTP_VP9_DESC_REQUIRED_SIZE 1
29 struct PayloadContext {
34 static av_cold PayloadContext *vp9_new_context(void)
36 return av_mallocz(sizeof(PayloadContext));
39 static void vp9_free_dyn_buffer(AVIOContext **dyn_buf)
41 uint8_t *ptr_dyn_buffer;
42 avio_close_dyn_buf(*dyn_buf, &ptr_dyn_buffer);
43 av_free(ptr_dyn_buffer);
47 static av_cold void vp9_free_context(PayloadContext *data)
52 static av_cold int vp9_init(AVFormatContext *ctx, int st_index,
55 av_dlog(ctx, "vp9_init() for stream %d\n", st_index);
56 av_log(ctx, AV_LOG_WARNING,
57 "RTP/VP9 support is still experimental\n");
62 ctx->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL;
67 static int vp9_handle_packet(AVFormatContext *ctx, PayloadContext *rtp_vp9_ctx,
68 AVStream *st, AVPacket *pkt, uint32_t *timestamp,
69 const uint8_t *buf, int len, uint16_t seq,
72 int has_pic_id, has_layer_idc, has_ref_idc, has_ss_data, has_su_data;
73 av_unused int pic_id = 0, non_key_frame = 0;
74 av_unused int layer_temporal = -1, layer_spatial = -1, layer_quality = -1;
75 int ref_fields = 0, has_ref_field_ext_pic_id = 0;
76 int first_fragment, last_fragment;
79 /* drop data of previous packets in case of non-continuous (lossy) packet stream */
80 if (rtp_vp9_ctx->buf && rtp_vp9_ctx->timestamp != *timestamp) {
81 vp9_free_dyn_buffer(&rtp_vp9_ctx->buf);
84 /* sanity check for size of input packet: 1 byte payload at least */
85 if (len < RTP_VP9_DESC_REQUIRED_SIZE + 1) {
86 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet, got %d bytes\n", len);
87 return AVERROR_INVALIDDATA;
91 decode the required VP9 payload descriptor according to section 4.2 of the spec.:
95 |I|L|F|B|E|V|U|-| (REQUIRED)
99 L: Layer indices present
100 F: Reference indices present
101 B: Start of VP9 frame
103 V: Scalability Structure (SS) present
104 U: Scalability Structure Update (SU) present
106 has_pic_id = buf[0] & 0x80;
107 has_layer_idc = buf[0] & 0x40;
108 has_ref_idc = buf[0] & 0x20;
109 first_fragment = buf[0] & 0x10;
110 last_fragment = buf[0] & 0x08;
111 has_ss_data = buf[0] & 0x04;
112 has_su_data = buf[0] & 0x02;
114 /* sanity check for markers: B should always be equal to the RTP M marker */
115 if (last_fragment >> 2 != flags & RTP_FLAG_MARKER) {
116 av_log(ctx, AV_LOG_ERROR, "Invalid combination of B and M marker\n");
117 return AVERROR_INVALIDDATA;
120 /* pass the extensions field */
121 buf += RTP_VP9_DESC_REQUIRED_SIZE;
122 len -= RTP_VP9_DESC_REQUIRED_SIZE;
125 decode the 1-byte/2-byte picture ID:
129 I: |M|PICTURE ID | (RECOMMENDED)
131 M: | EXTENDED PID | (RECOMMENDED)
134 M: The most significant bit of the first octet is an extension flag.
135 PictureID: 8 or 16 bits including the M bit.
139 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n");
140 return AVERROR_INVALIDDATA;
143 /* check for 1-byte or 2-byte picture index */
146 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n");
147 return AVERROR_INVALIDDATA;
149 pic_id = AV_RB16(buf) & 0x7fff;
153 pic_id = buf[0] & 0x7f;
164 L: | T | S | Q | R | (CONDITIONALLY RECOMMENDED)
167 T, S and Q are 2-bit indices for temporal, spatial, and quality layers.
168 If "F" is set in the initial octet, R is 2 bits representing the number
169 of reference fields this frame refers to.
173 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet");
174 return AVERROR_INVALIDDATA;
176 layer_temporal = buf[0] & 0xC0;
177 layer_spatial = buf[0] & 0x30;
178 layer_quality = buf[0] & 0x0C;
180 ref_fields = buf[0] & 0x03;
189 decode the reference fields
193 F: | PID |X| RS| RQ| (OPTIONAL) .
194 +-+-+-+-+-+-+-+-+ . - R times
195 X: | EXTENDED PID | (OPTIONAL) .
198 PID: The relative Picture ID referred to by this frame.
199 RS and RQ: The spatial and quality layer IDs.
200 X: 1 if this layer index has an extended relative Picture ID.
205 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n");
206 return AVERROR_INVALIDDATA;
209 has_ref_field_ext_pic_id = buf[0] & 0x10;
211 /* pass ref. field */
212 if (has_ref_field_ext_pic_id) {
214 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n");
215 return AVERROR_INVALIDDATA;
218 /* ignore ref. data */
224 /* ignore ref. data */
234 decode the scalability structure (SS)
240 | T | S | Q | R | (OPTIONAL) .
241 +-+-+-+-+-+-+-+-+ -\ .
242 | PID |X| RS| RQ| (OPTIONAL) . . - PAT. LEN. times
243 +-+-+-+-+-+-+-+-+ . - R times .
244 X: | EXTENDED PID | (OPTIONAL) . .
245 +-+-+-+-+-+-+-+-+ -/ -/
247 PID: The relative Picture ID referred to by this frame.
248 RS and RQ: The spatial and quality layer IDs.
249 X: 1 if this layer index has an extended relative Picture ID.
252 avpriv_report_missing_feature(ctx, "VP9 scalability structure data\n");
253 return AVERROR_PATCHWELCOME;
257 decode the scalability update structure (SU)
262 avpriv_report_missing_feature(ctx, "VP9 scalability update structure data\n");
263 return AVERROR_PATCHWELCOME;
267 decode the VP9 payload header
271 //XXX: implement when specified
273 /* sanity check: 1 byte payload as minimum */
275 av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n");
276 return AVERROR_INVALIDDATA;
279 /* start frame buffering with new dynamic buffer */
280 if (!rtp_vp9_ctx->buf) {
281 /* sanity check: a new frame should have started */
282 if (first_fragment) {
283 res = avio_open_dyn_buf(&rtp_vp9_ctx->buf);
286 /* update the timestamp in the frame packet with the one from the RTP packet */
287 rtp_vp9_ctx->timestamp = *timestamp;
289 /* frame not started yet, need more packets */
290 return AVERROR(EAGAIN);
294 /* write the fragment to the dyn. buffer */
295 avio_write(rtp_vp9_ctx->buf, buf, len);
297 /* do we need more fragments? */
299 return AVERROR(EAGAIN);
301 /* close frame buffering and create resulting A/V packet */
302 res = ff_rtp_finalize_packet(pkt, &rtp_vp9_ctx->buf, st->index);
309 RTPDynamicProtocolHandler ff_vp9_dynamic_handler = {
311 .codec_type = AVMEDIA_TYPE_VIDEO,
312 .codec_id = AV_CODEC_ID_VP9,
314 .alloc = vp9_new_context,
315 .free = vp9_free_context,
316 .parse_packet = vp9_handle_packet